《Linux基金會：2024健康數據互操作性的開放架構研究報告（英文版）（42頁）.pdf》由會員分享，可在線閱讀，更多相關《Linux基金會：2024健康數據互操作性的開放架構研究報告（英文版）（42頁）.pdf（42頁珍藏版）》請在三個皮匠報告上搜索。

1、An Open Architecture for Health Data InteroperabilityHow Open Source Can Help the Healthcare Sector Overcome the“Information Dark Ages”Anna Hermansen,The Linux FoundationForeword by Leo Anthony Celi,MIT Laboratory for Computational PhysiologyOctober 2024Healthcare is a unique and complex sector that

2、 is highly regulated,is risk averse,and must consider a diverse set of stakeholders.Health data is siloed andits exchange hamstrungby entrenched incumbent record systems that lack interoperability.The European HealthData Space and the Trusted Exchange Framework&Common Agreement aretwo examples of in

3、itiatives working to enhance healthdata exchange.Common misperceptions of open source,such as a lack of technical&legal support and commercial inviability,also exist in the digital health sector.Open source digital health solutions can increase health equity,de-risk innovation,and remove vendor lock

4、-in.Open source solutions are gaining traction in Europeand developing countries,embracing efficiency and agility in regions that prioritize collaboration and cost savings.Two examples of opensource solutions are DHIS2,used for data managementin over 100 countries,and SORMAS,used foroutbreak monitor

5、ing inover 15 countries.Artificial intelligence holds significant promise in healthcare,and its data needs may catalyze the development of more effective data exchange infrastructure.A precompetitive digital health architecture would standardize the components of the system and allow for the develop

6、ment of applications that are portable,sustainable,and interoperable.To build lifelong records,data must be semantically standardized and patient centric,separating the data from the applications and using a common data store.Innovative and agile solutions can work around incumbent platforms,allowin

7、g for bottom-up development and an opening up of the digital health market.A neutral foundation is necessary to create a center of gravity for open source health solutionsto collaborate,learn,and standardize around a precompetitive layer.Copyright 2024 The Linux Foundation|October 2024.This report i

8、s licensed under the Creative Commons Attribution-NoDerivatives 4.0 International Public LicenseAn Open Architecture for Health Data InteroperabilityContentsForeword.04Executive summary.05The state of the health tech sector.06Healthcare as a unique and complex sector.06The people and processes behin

9、d digital health transformation.08Innovation in digital health.10Nonstandardized data exchange infrastructure.12Health data standards and interoperability.13EHR incumbent market power.16Open source for health data management.18(Mis)perceptions of open source.19Benefits of open source.22Open source h

10、ealthcare solutions around the world.27AI the catalyst.30Moving the sector forward:Recommendations from experts.32Building a digital health architecture.32Standardizing data,semantically.34Trying new business models:Innovation around incumbents.36Next steps:The role of the foundation.36Conclusion.38

11、Methodology.38References.39Acknowledgments.41About the author.41ForewordThe proliferation of digital health data has led to siloed systems,making it difficult for stakeholders to access and share information,which ultimately limits research and patient care.The healthcare sector faces significant hu

12、rdles in data interoperability due to the nonstandardized nature of electronic health records(EHRs),which complicates access to and sharing of health data.The unique characteristics of healthcare,such as unpredictable demand,ethical obligations,and strict regulations,contribute to the complexity of

13、managing health systems and hinder effective digitalization.Effective governance is critical but often overlooked in health technology projects.Poor governance can impede the success of digital health initiatives and lead to ineffective implementations.As this report from the Linux Foundation illust

14、rates,open source solutions offer a promising avenue for enhancing healthcare data management by promoting collaboration,transparency,and cost-effectiveness,although adoption faces resistance due to misperceptions.Proprietary EHR systems create vendor lock-in,limiting flexibility and the ability to

15、customize solutions to meet specific healthcare needs,which can negatively impact patient care.Open source projects benefit from community involvement,which can provide ongoing support,reduce dependency on single vendors,and enhance the sustainability of healthcare solutions.The ethical dimension of

16、 healthcare aligns well with open source principles,emphasizing the need for transparency and shared knowledge to improve patient outcomes and public health.The practice of using encryption for protecting intellectual property(IP)dates back to Galileo in 1610 and has roots in ancient civilizations,h

17、ighlighting the long-standing need for safeguarding valuable information.The modern open science movement aims to democratize access to scientific knowledge,addressing historical inequities and promoting transparency,accessibility,and inclusivity in research and innovation.Key milestones,such as the

18、 Budapest Open Access Initiative and UNESCOs 2021 Recommendation on Open Science,underscore efforts to make science more equitable and accessible.The COVID-19 pandemic demonstrated the potential of open science for rapid data sharing and collaboration,highlighting the importance of equitable access

19、in global health responses.But despite its ideals,open science faces challenges,including structural inequalities and entrenched systemic practices that hinder equitable access to knowledge and material resources.Open science does not unfold in a vacuum but in an ecosystem besieged with power dynami

20、cs and knowledge hierarchies,marginalizing certain groups and countries and affecting their credibility and access to scientific discourse and engagement.The dominance of the Global North in scientific endeavors creates disparities,with actors including innovators from low-and middle-income countrie

21、s often facing barriers to collaboration.A radical reimagination of open science is necessary,advocating for the inclusion of marginalized voices and addressing existing power imbalances in the technological community.Ultimately,the impact of open science,just like any other disruptive innovation,hi

22、nges on who sets the agenda and who controls the movement.At this moment of reimagination,important change can start within the open source community.We hope that this report inspires readers to initiate the next wave of development,collaboration,and innovation for the health technology sector.LEO A

23、NTHONY CELI Clinical Research Director and Senior Research Scientist MIT Laboratory for Computational Physiology4AN OPEN ARCHITECTURE FOR HEALTH DATA INTEROPERABILITYExecutive summaryThis Linux Foundation research report reviews the adoption,perceptions,and potential of open source in health data in

24、frastructure.It starts with an overview of the unique characteristics of the healthcare sector and how this impacts innovation and digitization.It then focuses on the primary problem statement of this research,namely the lack of interoperability between data systems and how this hinders data sharing

25、.It addresses a number of reasons why there is a lack of interoperability,including the market power of incumbent electronic health record(EHR)systems.Turning to open source,the research discusses the perceptions of and challenges to open source adoption in the health tech sector and how the sectors

26、 interoperability challenges make it a relevant application for open source.After addressing the way that artificial intelligence(AI)could potentially catalyze a transformation of health data collection practices,the report concludes with recommendations on how to move forward,including the developm

27、ent of an architecture of standards and technologies to increase innovation and data sharing.5AN OPEN ARCHITECTURE FOR HEALTH DATA INTEROPERABILITYThe state of the health tech sectorThe digital transformation of healthcare has led to the exponential growth of data stores useful for providers,researc

28、hers,institutions,and companies to improve care for patients.A significant aspect of this digital transformation was the introduction of EHRs around the world,which are digital systems used to capture and manage health data,primarily in a hospital setting1.Despite widespread adoption of EHRs,accessi

29、ng this data remains difficult,as these systems do not interoperate well with each other.For any individual engaging in healthcare systems around the worldas a patient,a provider,a researcher,an administrator,or an IT managerit is clear that the sector suffers from poor portability of data.2 This pr

30、oblem stems from technological,regulatory,cultural,and operational particularities that,in combination,make this sector uniquely challenging to effectively digitalize.Open source communities have driven major new technological shifts,from the digitization of vehicles to 5G telecommunications to AI,b

31、y building the shared infrastructure for those transformations.Health data systems are in need of a similar transformation that prioritizes collaboration,transparency,and interoperability.These facets are fundamental to the open source value proposition,and yet major stakeholders in healthcare are a

32、bsent from open source communities.LF Research undertook this research study to understand open source adoption in this space,what challenges exist,and how the LF could play a role in accelerating interoperable,digital healthcare systems.From the perspective of open source technologists,this report

33、captures the reasons for the lack of data interoperability in the healthcare sector and the impact that this has on care and research,highlighting the obstacles to open source adoption as well as the ways in which a health-specific open source architecture of standards and technologies could address

34、 this issue.Healthcare as a unique and complex sectorThe healthcare sector has a number of particular characteristics that,combined,make it unique.From a market perspective,healthcare is not a typical industry sector due to a number of factors:3 The irregular,non-fungible,and lifetime demand for hea

35、lthcare:When a patient demands healthcare,this typically comes at unpredictable times and often becomes urgent.Arun Kumbhat,Director of GoToMarket,HR,and PR Services for Libra Social Research Foundation,called health“a non-fungible subject,”which,when combined with its nonlinearity and urgency,makes

36、 it“completely different from consumer goods,direct-to-consumer businesses,banking,finance,and insurance.”He also explained how a patient may require follow-up or preemptive treatments throughout the progression of a disease,which necessitates a“continuous personal health record.”The ethical and mor

37、al expectation placed on physicians:Unlike most other markets,the relationship between the patient and the provider rests on trust that the provider has the best interests of the patient in mind during treatment.This trust layer is significant due to the risk on the line for the provider,where their

38、 decisions could mean the difference between life and death.4 This places a moral and social obligation at the center of this market,which according to a professor at a European university of technology,makes it“a very specific sector Its public or social purpose driven.”The strict and complex regul

39、atory landscape:Healthcare is a highly regulated sector.4 As Alex Scammon,Head of Open Source Development at G-Research,explained,the regulatory characteristics of the market introduce significant financial considerations for those developing healthcare products and services.He noted that the high-r

40、isk nature of a healthcare product increases the time and cost it takes to get approval,and,as such,“it takes a huge amount of resources to get things past the regulatory hurdles That does seem more unique in healthcare than in big tech,for example.”Although these regulations are important from a sa

41、fety perspective,some were written without the internet in mind or before smartphones,such as HIPAA,making them less relevant and potentially obstructive in the current context.6AN OPEN ARCHITECTURE FOR HEALTH DATA INTEROPERABILITYThese unique aspects of the healthcare market contribute to the compl

42、exity of health systems management,where many different stakeholders in highly dynamic settings with strict governance mechanisms accomplish numerous and diverse tasks and decision processes.5 6 As Dr.Tony Shannon,Head of Digital Services at the Office of the Government Chief Information Officer wit

43、hin the government of Ireland,argued,“hospitals probably are the most complex organizational unit on the planetthere is so much going on in a typical hospital.”Any transformations,technological or otherwise,in a healthcare system must consider a significant number of stakeholders and their interests

44、,the regulations and risks in place,and the unique trust relationship between patient and provider.4“Hospitals probably are the most complex organizational unit on the planet.”Dr.Tony ShannonHead of Digital Services Government Chief Information Officer Government of Ireland7AN OPEN ARCHITECTURE FOR

45、HEALTH DATA INTEROPERABILITYThe people and processes behind digital health transformation Given the complexity of the health sector,its digital transformation has been a slow and fragmented process.The transition from paper to digital charts is a clear example of this.A senior consultant at a global

46、 consulting firm,explained that the implementation of digital records for the most part“didnt bring the tool,the technology,into the process of what the doctor is doing.And so its an extra step,instead of being integrated into what theyre doing.”She went on to say,“if the technology isnt following t

47、heir pathway,theyre not going to use it.”This usability issue is well summed up in the fact that there is an entire industry of medical scribes to put health data into the digital system.“What other industry would tolerate an industry of scribes working alongside professionals to feed the informatio

48、n beast?”Shannon lamented.Shannons own interpretation of the challenge with digital transformation is due to the fact that it comes down to a complex convergence of people,process,and technology,with human resources all siloed within three domains:“The medical professionals are trained in the clinic

49、al domain but dont understand technology;the technologists are trained in the technical domain but dont understand the clinical domain;and the management administrators are trained in management science but dont understand either the clinical process or the technical stack.”Despite this complexity,h

50、e argued that all clinical domains contain generic patterns in their processes,which are key to more effectively managing health data.Identifying these generic patterns helps institutions align and interoperate IT infrastructure across healthcare environments,but without this alignment,the problem r

51、emains complex and siloed.7 8 This convergence of technology,people,and process echoed across the interviews.Dr.Pankaj Gupta,Non-Executive Director of Libra Social Research Foundation,made a similar comment,arguing that“in healthcare IT,you need the business processes that are semantically standardi

52、zed,you need physical and IT infrastructure that is well developed,you need the human resources that have crossover skills between healthcare and technology.”A professor at a European university of technology explained the difficulty in ensuring that all these different facets work well together.“My

53、 medical colleagues,its not their business to deal with data or technology,you know,they dont have time.So,you need new positions or interested parties who can do that.But its very difficult to connect and trust other partners who might be proficient in technology or data analytics.”From a health re

54、search perspective,David Buckeridge,Professor at McGill University,concurred:“The challenge is,you need a lot of perspectives to make sense of this,right?You need the research ethics board perspective,you need a substantive analytical research perspective,and you need the IT security perspective.The

55、y all have to come together and look at this issue and make a sort of determination about it.And its challenging to get those people together,period,but also to get them together and have a discussion around issues like this.”Similarly,Jared Keller,an independent data sharing researcher and consulta

56、nt,pointed out that the proper management of a health technology requires technical people,lawyers,and businesspeople:“Theyre never all the same person,and its hard to get them to talk to each other.”Understanding these different perspectives is important to have well-functioning,effectively governe

57、d,and sustainable projects.Limiting the focus of the project to defining the technology infrastructure ignores the potential institutional or social obstacles,which are just as important to consider.9 Keller made the point that,in his experience,governance is always the last thing to undergo conside

58、ration.“There is a temptation to start with the tech,but that can lead people astray.”Buckeridge 8AN OPEN ARCHITECTURE FOR HEALTH DATA INTEROPERABILITYagreed that governance is one of the most significant obstacles to implementing digital health projects,but he pointed out that it is also one of the

59、 hardest elements to get right.“With everything Ive built in terms of data analytics infrastructure,the governance is the hardest part.And if you dont figure it out ahead of time,its a real headache,but at the same time,if you wait to figure it out,you may not get anything done.”Putting governance c

60、onsiderations on the same level as technical ones is particularly challenging in a sector where resources are not always well financed or well allocated.As a senior consultant at a global consulting firm explained,“the fact that the Canadian healthcare system isI wont say bankrupt,but the belts are

61、being tightenedthe risk with that is,when youre doing a big project,the budgets get cut.And often the budgets get cut in the nontechnical streams,which is change management,clinical adoption,training,communication,and so on.And thats where this industry needs it the most.And so,I think the tightenin

62、g of budgets is hampering the success of implementations.”Without these governance and change management mechanisms in place,digital innovations in the health sector may not have the ability to really take off.“Often the budgets get cut in the nontechnical streams,which is change management,clinical

63、 adoption,training,communication,and so on.And thats where this industry needs it the most.And so,I think the tightening of budgets is hampering the success of implementations.”9AN OPEN ARCHITECTURE FOR HEALTH DATA INTEROPERABILITYInnovation in digital healthThe unique aspects of the healthcare mark

64、et make it a challenging area for innovation.First,being social purpose driven means that it is not as much of a profit-driven sector.A professor at a European university of technology sees digital health investments falling around the world,tying it to the way that healthcare represents a public go

65、od:“There is a limit to the profits you can yield from digital healthcare solutions.”Peter Lee,Head of Microsoft Research,agreed with this sentiment,suggesting that there is a fundamental ethical difference between big tech and healthcare.As he explained,investors in big tech companies such as Micro

66、soft demand significant revenue growth,which no healthcare business is capable of doing,reducing the motivation to invest in this sector.This leaves big tech companies to go through a cycle of investing in healthcare and then giving up,according to Lee.Ian McNicoll,CEO of freshEHR,and a professor at

67、 a European university of technology both put the blame on the competencies of tech companies.As The professor stated,“healthcare is too far from their core competence.They jumped into the healthcare value chain without the medical competence.”McNicoll pointed out that this cycle happens“because big

68、 tech is all driven by data analysis and not by direct care They have a lot of experience in data analysis,but its downstream from the frontline stuff.And frankly,because the frontline stuff is just too damn hard,they dont want to get involved in that.”This leads to restricted investments in healthc

69、are innovations,with concerns about what it will take for true digital transformation in this sector.10 On the other side of the spectrum,startups also have difficulties entering the market,as they provide a product or service that is very specific to a particular problem in the health market.A prof

70、essor at a European university of technology pointed out that when the market is as big as it is,and the product only fits a specific need,“if you do not get into the bigger offerings,you might simply not be useful.”As a result,the sector suffers from“pilotitis,”where innovative interventions have s

71、een success only in niche and restricted contexts,lacking the ability to scale up beyond their pilot stage.9,11 As McNicoll argued,without an ability to plug their offering into other areas of the market,“its a real drag on innovationthe new little startups who come in,theyre having to rebuild the w

72、hole infrastructure themselves.”The startups particular product or service is also subject to compliance requirements,which can be very costly and time consuming,such as the HiTrust certification in the U.S.12 Second,the slow pace of innovation is in part due to the higher risks in healthcare.Gawand

73、e(2012)notes how slowly new processes and medicines trickle down in medicine compared with other consumer areas,where the gap in discovery and implementation is“appallingly”large.13 The“fail fast and iterate”concept that moves most consumer innovations along is not possible in a sector where the ris

74、ks are too high to tolerate flaws.4 From the procurement side,investment in innovative technologies is limited,and instead,hospitals and clinics tend to procure products from vendors.As Buckeridge posited,this procurement strategy“is seen as a way to mitigate risk and decrease requirements for skill

75、ed personnel at some level.”Niraj Dalmia,Partner in Omnia AI at Deloitte Canada,explained the risk-aversion from a public sector perspective,stating,“In the private sector,there is some more appetite to take risk,especially in non-healthcare sectors due to the fail fast and iterate benefits The publ

76、ic sector,I would say,is looking for precedent and looking for a little bit more surety that its been successful,its been done.And I think,fair enough10AN OPEN ARCHITECTURE FOR HEALTH DATA INTEROPERABILITYthe public sector is spending tax dollars on these softwares,and they want to absolutely make s

77、ure its going to be successful Its a fine balance thats tricky to achieve,but it has to be obtained to counter the productivity paradox in the healthcare sector that comes with digital and information system implementations.”There are also important risks to consider from a data privacy perspective.

78、As a senior consultant at a global consulting firm explained,her work in risk advisory for her clients is“around the privacy and cyber within a technology platform,which is obviously very important for technology implementations where theres health information included in it.”The stringent regulatio

79、ns around protecting data make for risk-averse technology implementations.The unique characteristics of the healthcare market,in combination with the complexity of the various stakeholders and their relationships to one another,has made digital transformation of the sector challenging and restricted

80、.“In the private sector,there is some more appetite to take risk,especially in non-healthcare sectors due to the fail fast and iterate benefits The public sector,I would say,is looking for precedent and looking for a little bit more surety that its been successful,its been done.”Niraj DalmiaPartner

81、in Omnia AI at Deloitte Canada11AN OPEN ARCHITECTURE FOR HEALTH DATA INTEROPERABILITYNonstandardized data exchange infrastructureIn this complex regulatory,financial,technical,and governance environment,an important aspect of the health system suffershealth data.As the sector has digitized,and the d

82、ata available in digital formats has grown exponentially,these different factors have locked data into silos that become challenging or even impossible to access and share.As a senior consultant at a global consulting firm plainly stated,“its really hard to bring data together to provide value for e

83、ither the patient or the clinician.”Researchers also struggle with accessing data and,as a result,face limitations in their ability to produce outcomes from real-world evidence.14“We dont have enough data to do what we want to do.We would like 100 times more,1,000 times more than what we have,”expla

84、ined Scammon in reference to his work on an open source diagnostic model for hip dysplasia.This siloed context makes it very challenging to track data across a patients journey.McNicoll referenced a cancer patients journey as an example:As they move from symptoms to diagnosis to surgery to treatment

85、,“theres a lot of handover of information at every step and its a real challenge to follow the patient through their data.”Whats more,the health data itself is“uniquely complex,wide,messy,and fractal,”McNicoll pointed out.He explained,“its never actually been mapped.Its all whats in clinicians heads

86、,its all in bits of paper.Theres lots ofconfusion and differences of meaning;context is very,very important.”Noah Harlan,Senior Advisor at Findhelp,explained how this complexity gets worse when incorporating social determinants of health(SDOH)systems:“There are too many interlocking ecosystems as th

87、e notion of care expands outward.And there are too many places where different pieces and combinations of the information need to move.”Different data,even data that some may consider adjacent to care,need to be able to come together to build an adequate picture of the patientand this relies on stan

88、dards.“We dont have enough data to do what we want to do.We would like 100 times more,1,000 times more than what we have.”Alex ScammonHead of Open Source Development at G-Research12AN OPEN ARCHITECTURE FOR HEALTH DATA INTEROPERABILITYHealth data standards and interoperabilityHealth data is defined u

89、sing various standards that are meant to make it“computable,understandable,replicable,reusable,and interoperable”across health systems.15 There are a number of standards developed for different stages of the health data lifecycle:content standards for data creation,code systems for data formatting,i

90、nformation standards for data analysis,exchange standards for data flow,and privacy standards for data protection.15 Some more popular standards are Fast Healthcare Interoperability Resources(FHIR),SNOMED Clinical Terms(CT),Logical Observation Identifiers Names and Codes(LOINC),and the Observational

91、 Medical Outcomes Partnership(OMOP)Common Data Model(CDM).Many of these are global standards,but there are also regional efforts to standardize health data.For example,a professor at a European university of technology describes how each European country has its own national healthcare record platfo

92、rm,and the European Union is working to build standards to join these platforms together and connect them as a single market.Despite progress in adoption and implementation,these standards have not solved the interoperability issue.The interviewees pointed out that these standards are not adequate t

93、o provide real,sustainable interoperability between different health systems.A senior consultant at a global consulting firm explained the issue from a governance and implementation perspective.In Canada,where healthcare is a provincial jurisdiction,the federal government“can put standards up there,

94、but the provinces dont have to follow it.They can do their own thing.”This impacts intra-as well as cross-provincial data sharing,where each system can use its own standard.Harlan also commented on this from the American perspective:“Interoperability is very hard.The social care programs have differ

95、ent rules,the rules vary from state to state,plus youve got federal rules.”McNicoll gave a similar reasoning from the perspective of the clinicians role and dynamic within the U.K.health system:“Its hard to standardize clinicians to get them to collect data in the same way.They wield more power than

96、 somebody in a different profession,you know,a lawyer,for instance,or somebody whos within IBM where there will be company-wide standards.Thats hard to do in health because of the power of clinicians.”A senior consultant at a global consulting firm agreed.“When you look at Canadian clinicians,theyre

97、 sole proprietors,for the most part They can do their own thing a little bit more so a lot of our time is spent in getting them to agree to a protocol or a pathway or something,across the hospitals,or even within a hospital.”FOUR COMMON HEALTH DATA STANDARDSFHIR:Used to standardize medical data in e

98、xchange by defining the structure for the core information set that is common to most applications,regardless of how the data is stored.16 OMOP CDM:Used to standardize data in rest and for analysis by defining the structure and content of data using a core vocabulary of medical terms.17 SNOMED CT:Cl

99、inical terminology providing a standardized way to represent clinical phrases captured by the clinician,defining the representation of and the relationship between components of the clinical process.18 LOINC:Universal medical terminology of code names and identifiers for laboratory tests and observa

100、tions.19 13AN OPEN ARCHITECTURE FOR HEALTH DATA INTEROPERABILITYTo properly implement these standards requires authority.Janice Spence,Senior Standards Specialist,and Linda Parisien,Certified Terminology Standards Specialist,both at Canada Health Infoway,explained how the Canadian government is curr

101、ently driving the adoption of SNOMED CT.However,they pointed to the fact that different EHR providers are not wholly compliant and often have limited capacity to fully implement terminology standards.“There is some functionality that is available,but its not maximized,”Parisien commented.Spence expl

102、ained the problem from a governance perspective:“There needs to be drivers in place to say to the EHR vendors or digital health solution providers that they are required to adopt and support approved pan-Canadian and international terminology and data exchange standards,to comply with the standards,

103、and to maintain those standards.Right now,across the country,we have that gap there Were looking to see where we can have some levers in place to ensure that a compliance framework is actually put in place and maintained.”This leads to vendors who say they are using a standard without being wholly c

104、ompliant or interoperable,according to an architect at a U.S.health department:“All vendors claim their product is interoperable.And thats 100%not true The problem with all of the standards is that every vendor can implement them slightly differently,which means there is in fact no interoperability.

105、”Described from his data standardization perspective,McNicoll pointed out that there is standardization at the data exchange level,using a standard such as FHIR,but there is not standardization at the data store level.“When we exchange data,we exchange data in this agreed format.And then you can imp

106、ort the agreed format into your system But the systems still store and manage data in their own internal database formats.”Instead,he argued that the standard should be“inside”the data,creating a common format for data,and then the applications can work directly on top of that data.The lack of inter

107、operability is also the outcome of many different solutions being developed for different purposes in different contexts.Gupta and others pointed to the piecemeal development of health data systems,where“theres no way you can triangulate all these data points,because all these vertical health system

108、s built their IT systems at different points in time,in different technologies.And the majority of them do not talk to each other,because they were never designed to do so,”Gupta explained.However,the answer is not necessarily to have one uniform softwareas a senior consultant at a global consulting

109、 firm commented,“interoperability is the answer,not homogeneous technology.”A professor at a European university of technology also argued that“one giant solution is not possible in innovation The way countries are moving is interoperability of diverse platforms.”Despite this understanding,we see gl

110、obal EHR markets consolidating around a handful of giant providers.14AN OPEN ARCHITECTURE FOR HEALTH DATA INTEROPERABILITYEXAMPLES OF DATA EXCHANGE INITIATIVESSeveral public and private initiatives and frameworks support health information exchange.One such example is the Trusted Exchange Framework

111、and Common Agreement(TEFCA),which the U.S.Department of Health and Human Services Office of the National Coordinator for Health Information Technology introduced in January 2022.TEFCA aims to enhance the digital transfer of health information between participating entities,known as Qualified Health

112、Information Networks(QHINs).Its goal is to establish universal governance,policy,and technical standards to achieve nationwide interoperability,thereby simplifying healthcare data sharing and exchange across organizations.Through its Common Agreement,TEFCA provides a governance framework with baseli

113、ne legal and technical requirements that apply to all health information networks,ensuring consistent and standardized data exchange.20 The European Union recently approved a parallel initiative known as the European Health Data Space(EHDS).The Commissions approach emphasizes data rights,aiming to p

114、rovide citizens with full control over their health data.This will be facilitated by a legal and policy framework designed to ensure the“free movement of health data,”allowing health data to seamlessly follow individuals wherever they go.In practice,citizens will be able to access their health infor

115、mation in a standardized European format,free of charge,and regardless of their location within the E.U.Beyond enhancing personal access,the EHDS aims to foster benefits for research,scientific development,and industry by promoting a“genuine single market”for EHR systems.This standardization aims to

116、 simplify the market entry process for EHR providers across member states,promoting uniformity and interoperability.The EHDS is expected to result in substantial cost savingsover 5.5 billion over the next decadethrough improved access to and exchange of health data.The initiative is scheduled for fo

117、rmal adoption in the autumn of 2024.21 15AN OPEN ARCHITECTURE FOR HEALTH DATA INTEROPERABILITYEHR incumbent market powerToday,the developed world largely relies on a handful of major proprietary EHR platforms to handle hospital data.The two that primarily came up in interviews were Epic and Cernerwh

118、o collectively hold over 50%of market share in the U.S.and are some of the top EHR vendors in the U.K.and Canada.22 23 24 Many interviewees brought up this consolidation and its impacts on data management.Buckeridge explained that historically,“most of the major academic health centers that had stro

119、ng informatics programs in the U.S.would build and run their own electronic medical record,but almost all of those have been replaced in the U.S.now by Epic,for example.”St.Michaels Hospital in Toronto exemplifies this:As Jamie Beverley,Director of Product Development for the Data Science and Advanc

120、ed Analytics team at Unity Health Toronto,pointed out,it is replacing its current patient record system with Epic.A senior consultant at a global consulting firm also mentioned that most of her Canadian clients choose to implement Epic or Cerner.As an architect at a U.S.health department explained,“

121、in the U.S.,theres just one consolidation after another to make an increasing monopoly on healthcare networks.”These incumbent EHR players are not oriented toward standardization.From an economic perspective,these players want to keep their users locked into their systems so that they will continue

122、to pay the licensing fees.Standardizing their infrastructure could mean opening up the opportunity for clients to integrate their solutions with other standardized vendors.In a privately delivered system such as in the U.S.,the hospital or clinics interest in keeping its patient locked into its syst

123、em compounds this.As an architect at a U.S.health department explained,“big healthcare networks want to lock their patients into their health network and keep the cash flow from that patient cohort.Similarly,EHR vendors want to have as many customers as possible paying license fees and so on for the

124、ir platform.”Neither of these parties has incentives to standardize in order to share data.Shannon agreed with this,stating,“there are active efforts to contain data,and not share it,that are built into the economic model of the healthcare systems that are there today.”Even in a publicly funded syst

125、em,the motivation to make data portable is not always there.As Dalmia explains,“the motivation to share data across Canada has been limited,”although he has seen growing interoperability initiatives across provinces in Canada.He argues that it requires“a culture and mindset change,and the appetite t

126、o do that.”This lack of interoperabilityparticularly at a change management and governance levelrequires resources and funding that publicly funded systems may not prioritize.An important consideration in this consolidation around a few EHRs is that these systems do not cover all aspects of healthca

127、re,and there is an“edge effect”that reduces the interoperability of a patients data journey.When administrators purchase one chunk of software,thinking this will achieve interoperability,“this misses the point,because for example here in Quebec with Epic,it doesnt cover everything.It doesnt cover pr

128、imary care.Its not going to cover public health,”said Buckeridge.Harlan explained it from the SDOH perspective,where an individual who misses a food stamp registration may end up in the hospital because theyre hungry,and everything up to the point of the hospital bill happens outside of Epic.“All of

129、 the negative externalities around that,all of the cost drivers around that,all of that processEpic has no role to play in it.”16AN OPEN ARCHITECTURE FOR HEALTH DATA INTEROPERABILITYUltimately,the standardization and sustainability of health data records is crucial,as this sector needs records that

130、are lifelong.Current systems are lacking the interoperability to produce lifelong records because of the absence of incentive,resources,and market competition to make them so.As a result of this lack of interoperability,“were living in the information Dark Ages as far as healthcare is concerned at t

131、he moment,”Shannon posited.“If you look at the U.S.,if you look at many parts of the Western world,you see a healthcare system in crisis.And its multifactorial,its people,its process,but its in no small part down to the poor state of health IT today.Its killing people.”He went on to say that“the hea

132、lth IT market is full of vested interests at the moment.And theyre resistant to that kind of advancement.The industry needs to be challenged;the monopolistic practices need to be challenged.”“There are active efforts to contain data,and not share it,that are built into the economic model of the heal

133、thcare systems that are there today.”17AN OPEN ARCHITECTURE FOR HEALTH DATA INTEROPERABILITYOpen source for health data managementIs open source the way to challenge this industry?Open sourceand,in particular,open scienceis a foundational tenet of healthcare.Shannon commented that“its a medical,cult

134、ural thing that you publish what you learn and get peer review,or else youre hiding a secret,you knowits not the moral thing to do.”The moral obligations of the sector,according to Shannon,align well with the moral obligations of open source to encourage transparency and sharing.He gave an example o

135、f the medical family that hid its original forceps tool to protect its private practice,which,by the current standard of“publish or perish,”is unfathomable to todays doctors.This collaboration is crucial to solve the universality of challenges that exist in this sector,making it a classic open sourc

136、e use case,according to Shannon:“We will only solve the wicked challenges in healthcare with a more open collaborative approach to the digital dimension.”“We will only solve the wicked challenges in healthcare with a more open collaborative approach to the digital dimension.”Dr.Tony Shannon Head of

137、Digital Services,Government Chief Information Officer,Government of Ireland18AN OPEN ARCHITECTURE FOR HEALTH DATA INTEROPERABILITY(Mis)perceptions of open sourceDespite this natural connection,it became clear from interviews that many common(mis)conceptions of open source exist in this sector,and th

138、e adoption of open source experiences resistance because of real and perceived challenges.Despite many of these perceptions being overcome in other sectors,it appears they still exist in healthcare.The following are common beliefs and misperceptions that came up in the interviews.Open source clashes

139、 with private sector interestsAs mentioned above,interviewees felt that open source carries an ethical obligation that may go hand in hand with the ethics of healthcare.As Gupta argued,the Libra Social Research Foundation develops open source software projects where,as a developer,“you can build you

140、r own product and use it,and if you create anything new on top of the core,that is yoursbut then if you modify the core,then you have toyoure sort of morally bound tobring it back into the open source.”Although this may not be the case in practice,open source is bound up with morality,at least for G

141、upta.This morality adds a dimension that may fit well with health technology goals,according to some intervieweesbut not to all.Despite some interviewees arguing that financial opportunity has held back technological innovation,Scammon felt that financial gain is the root of what drives proprietarya

142、nd not open sourceinnovation.He argued that healthcare is adversarial to open source because“proprietary vendors are driven by money.Thats the root of what is aggressive against open source,because theres a lot of money to be made in medical IP.And a lot of these vendors who are making a lot of mone

143、y will see this as a threat The very simple answer is that its money.”Shannon agreed with the idea that the sector is still very driven financially,arguing that“parts of the world where theres a lot of money will be slow to embrace open source and healthcare.”However,he felt that in regions where th

144、ere is less money in this sector,there will“emerge a set of tools that will disrupt the Western healthcare systems,whether they like it or not;its just interestingly coming from a place where innovation is happening,because the current incumbents arent there.”This argument played out in various othe

145、r interviews as a tension between the private sector and open source.From an open data perspective,Keller described how much harder it is to incentivize the private sector to openly publish their data,as it is hard for companies to understand the financial value particularly when they are fearful of

146、 giving away a competitive advantage.Gupta explained briefly how this conflict between financial interests and open source played out in India.When the government of India established the open source mandate for digital health standards,“there were reservations from the private sector because they h

147、ad to balance their other interests.”He went on to describe that although the mandate“sent shockwaves into the private sector some of them realized that the government is doing it for the greater good.”An architect at a U.S.health department argued that the arguments against open source are unfounde

148、d because of the commercial interest behind them.“Clearly they are managers,theyre nontechnical people.They dont care about innovation,they dont care about data.”He went on to describe the lobbying culture in the U.S.and how it hindered the adoption of the Veterans Affairs open source Veterans Healt

149、h Information Systems and Technology Architecture(VistA)platform.“This religion,effectively,that commercial by definition is better Thats nothing but pure misinformation on behalf19AN OPEN ARCHITECTURE FOR HEALTH DATA INTEROPERABILITYof vendors and contractors that want to sell the government their

150、commercial products.”He went on to say that today,“unless you have a national healthcare system,open source in healthcare is not a very good fit.Because there are so many other incentives in playmainly money-driven ones,profit-seeking ones.”Open source lacks support compared with proprietaryThe lack

151、 of proprietary backing for open source software concerned some of the interviewees.Buckeridge linked this hesitation for open source procurement to a lack of IT resources to manage this kind of environment,where,despite a policy in Quebec to favor open source,he sees a trend away from needing in-ho

152、use,local expertise to manage the software.In reference to his hospitals development group,he said:“We use virtualization,and for that,we dont use open source,we use VMwaremainly because our system administrator is not so comfortable with that,and his skillset doesnt really extend to creating an ope

153、n source environment for that.”He explained this as a conflation between open source software and a locally developed application,the latter of which can be detrimental to a hospital if it is poorly managed and the person who builds it moves on,“and nobody really quite knows what to do with it.”Beve

154、rley also pointed to the issue of resourcing,explaining how when their group tried to open source and share their models with other hospitals,they realized that the hospitals“didnt have the teams and the infrastructure to run those models,and then,to retrain them after they started drifting and to m

155、onitor them as they started drifting Hosting maintenance,ML Ops,monitoringthats been a gap in adoption of our open source efforts.”Dalmia also reflected on this,where projects do not“have the skills that are required in understanding,managing,and developing open source tools,so they just go with app

156、lication packages This does create a little bit of that comfort that its already been developed,tested,and tried and that youre not building from scratch.”Some interviewees linked the trend to vendor-backed software as a response to liability concerns.Chinmay Singh,Founder of iWish and TeleVox Healt

157、hcare,argued,“If I take an open source product,then the requirements around liability are not addressed,right.Like,if this software malfunctions,who do I go and see?I need to see somebody.”He goes on to clarify,“Now,thats not an insurmountable problem,because a lot of open source is being used in he

158、alth IT But the liability issue is the primary deterrent.”Jason Clark,DevOps Engineering Manager at CarrumHealth,brought up the same issue,articulating that from his perspective with academic medical institutions,“they want to have someone to call,they want a support agreement.”An architect at a U.S

159、.health department also noted that“anybody,such as a hospital or a business,will want to have a vendor with 24/7 support for their software need.They need support 24/7,they need a real company behind it.”Open source is commercially inviableThe concern around the sustainability of open source softwar

160、e also came up in the context of its commercial viability.McNicoll expressed a concern with sustainability,stating that“philosophically,”he agreed with advocating for open source in healthcare,but he felt that“practically,its very hard to sustain.”In his work in Jamaica,for example,they initially wo

161、rked with open source systems but didnt ultimately choose them because“they were hard to support.”He argued that this comes down to commercial viability of open source,where most of the viable open source projects“tend to have a big IT sponsor 20AN OPEN ARCHITECTURE FOR HEALTH DATA INTEROPERABILITYo

162、r government sponsorship,”without which open source projects have a hard time making something work commercially.Open source is only used for smaller projectsA number of interviewees expressed,either directly or subconsciously,that open source is best suited for smaller projects.A professor at a Eur

163、opean university of technology commented,“at a smaller scale,developers love to use open source solutions to come up with certain modules for a platform.But it has its own limits,right?”A senior consultant at a global consulting firm mentioned,“I dont necessarily work a lot with open source projects

164、,because at Deloitte,we often have very large projects.”She also stated,“I dont know anyone in the public health space who would use open source for the healthcare back office.”It was clear from some comments that there is a perception that open source is not a viable option for larger health inform

165、ation projects.This also points to a common preoccupation of open source at the application layer instead of considering its use at the protocol layer.These various challenges and perceptions of open source hinder its adoption in the healthcare sector,whether these perceptions are valid or are based

166、 on a lack of knowledge of open source.The following section reviews the common benefits of open source and how they apply in this sector to disrupt misconceptions and identify avenues for adoption.21AN OPEN ARCHITECTURE FOR HEALTH DATA INTEROPERABILITYBenefits of open sourceFor those in the open so

167、urce community,the usual benefits seen across industry verticals and technology horizontals also apply to the healthcare sector.Open source encourages health equity by reducing costsAffording proprietary solutions can be challenging,particularly for developing countries.As Scammon commented,his work

168、 on the open source hip dysplasia model“needs to exist for the world,because millions of infants need to get treated every year.And if it only exists in a proprietary form,it will only be used in places that have money.And thats not fair.”He described how their open source model is available for fre

169、e,with the scanning instrument as the only cost.Health systems that require certain tooling can therefore rely on open source alternatives to save on the expense of subscription and licensing fees.Similarly,Gupta explained the financial incentive for adopting open data standards in India:“While writ

170、ing the metadata and data standards for health,the directive given to us from the Ministry of Health was very clear:You need to have open standards.Dont go and start proposing proprietary standards or value sets for which India will have to pay.Imagine,for 1.4 billion people,if India had to pay per

171、transactionits going to be a nightmare,its just unaffordable.So that was a very clear direction that it has to be open source.”Kumbhat argued that“national-level transformations in healthcare,particularly in developing countries,are only possible with open source.Because it needs crowdsourced innova

172、tions,you want to avoid vendor lock-ins,you want to avoid obsolescence.And ultimately,it would reflect in the cost of delivering healthcare to the bottom of the pyramid.And developing countries cant really afford proprietary systems in that sense.”The ethical characteristic of healthcare makes it a

173、good application area for open source as a way to reduce costs.An architect at a U.S.health department made the case that this is a sector that should be putting the public before profits.He said,“for public healthcare systems that are taxpayer funded we should be giving that money back to the publi

174、c for services and not just helping shareholders benefit on the backs of taxpayers.Thats just fundamentally wrong,especially for something as basic as healthcare.”22AN OPEN ARCHITECTURE FOR HEALTH DATA INTEROPERABILITY“National-level transformations in healthcare,particularly in developing countries

175、,are only possible with open source.Because it needs crowdsourced innovations,you want to avoid vendor lock-ins,you want to avoid obsolescence.”Arun KumbhatDirector of GoToMarket,HR,and PR Services for Libra Social Research FoundationOpen source building blocks de-risk innovationAn important outcome

176、 of reducing costs through open source is that,according to Kumbhat,it“de-risks”innovation.He described this through his experience with various startups,where“a lot of time and money is spent in designing a product that is relevant to the market,”and this design cost is high without being able to b

177、uild on the expertise of others who have open sourced their solutions.Indias work developing and open sourcing its healthcare tooling means that innovators can benefit from and build on top of that.“Innovation needs to be crowdsourced,it needs to be inexpensive.It needs to be de-risked if youre goin

178、g to provide effective healthcare to a large population,right?”When a software is open sourced,Kumbhat explained,“the risk of standards and interoperability is removed,because thats already part of the minimum viable product.Thats a very significant risk reduction.”When this de-risking happens,vendo

179、rs can instead focus on the money-making aspects of the industry,according to Scammon:“In reality,theres still a lot of money to be made on more advanced problems if we let OSS take care of the basics.”The open sourcing of software projects creates a roster of“building blocks,”as a few of the interv

180、iewees described them.These building blocks become very useful in contexts where work is consistently being duplicated,such as the example Scammon provided.He spoke about Pharmaverse,an open source clinical trial platform that all pharmaceutical companies can use and contribute to.25 He said,“Right

181、now,every single pharmaceutical company has to build their own clinical trial platform,and every one of those needs to get approved by the Food and Drug Administration(FDA).And thats a huge amount of time and effort for each one of the companies.And its a huge amount of time and effort for the FDA.I

182、f we just had one shared one that everyone contributed to were saving everybody time,saving tons of money,and allowing them to focus on their core business.”23AN OPEN ARCHITECTURE FOR HEALTH DATA INTEROPERABILITYDalmia commented that moving away from vendor entrenchment means a more productive envir

183、onment for building solutions.Instead,using open source software“allows a little bit of flexibility with almost Lego bricks,as opposed to these monolithic applications.”He qualified that this approach requires maintaining the right in-house or contracted skillsets to keep these projects working,but

184、despite this need,open source is an important consideration for his clients procurement decisions.He gave a specific example where one of his clients wanted to convert a chunk of their codebase into open source“because they liked the flexibility;they have the skills;they did think that its a more mo

185、dern way of doing it.”Open source removes vendor lock-inInterviewees discussed concerns over the persistence and preservation of health data when using proprietary systems and how open source removes the risk of losing data if a company decides to remove its solution.As Singh argued,open source mean

186、s procurers are not beholden to the whim of a single provider,who may decide to kill a product or service:“The use of open source takes you away from these proprietary systems,which is a good thingit is less whimsical.”He mentioned the VistA open source EHR solution,arguing that VistA would never be

187、 able to say,“from today,no supportit would not happen.But that does happen in healthcare all the time.And these companies,the way they develop the technology solution,theyre trying to hold on to that data because that creates a vendor lock-in,and with open source,thats not an issue.You can clearly

188、see that with VistA.”Buckeridge commented that avoiding vendor lock-in is the rationale for open source adoption in his lab:“Were right now architecting our analytics stack,as were going to be rebuilding that in the next year or two,and thinking about what we want it to look like when we have Epic c

189、ome,which is a behemoth of the system.Our main strategy is to get data out into an open source stack quickly so that we dont have lock-in to a particular vendor or product.”Beverley also expressed a concern with vendor lock-in in terms of his ability to customize a software to a hospitals specific n

190、eeds.He commented,“Ive grown scared of vendor lock-in,and I think lots of groups are scaredthe idea of committing to commercial products and being in a place where you can no longer customize those things to the needs of an environment,to patients,or to healthcare staff.”His use of open source tooli

191、ng has allowed him to develop solutions for St.Michaels Hospital in Toronto that are tailored to his specific context.This kind of sustainability is crucial for a sector such as healthcare that requires lifelong records.Instead of relying on a proprietary vendor whose financial interests may conflic

192、t with building out sustainable health records,an open source platform is the better fit.An architect at a U.S.health department contrasted open source with proprietary vendors,where if you want to change vendors,“your data is trapped in the previous vendors database,and youll never get your data ba

193、ck.”He felt that because of this,“theres more argument for open source in organizations that care about data governance.”24AN OPEN ARCHITECTURE FOR HEALTH DATA INTEROPERABILITYOpen source is trustworthyNot only do open source projects have the long-term benefit of removing vendor lock-in,but they ar

194、e also more sustainable due to the fact that they are perceived as more trustworthy.An architect at a U.S.health department argued that developers are more likely to trust open source code.“If you want adoption of your software,and you want trustin other words,you want a lot of developers developing

195、 on your platformthe only way you get trust is by open sourcing your code and saying,Look,were open source.And then you get developers who are interested:Okay,I can play with this,I can use this,I can learn about this without having to pay anything.”If developers trust the code,they will want to con

196、tinue to use and contribute to it,making it a more sustainable project.He also extended the importance of trust to users of the product or service,stating,“You wouldnt trust your data in anything but something open sourcesomething that has a governance model that isnt about monetizing your data.”Cla

197、rk also discussed the trust placed in open source.He framed it within his work with AI models,where the democracy of decision-making,the transparency of the code and algorithms,and the ability for developers to improve the model themselves without appealing to a business are all crucial to building

198、successful models.This inherent trust,from the governance structure down to the very code,makes open source an attractive option for healthcare providers who need to build lifelong records.Open source has a community around itAnother important aspect of open source is the community that surrounds it

199、.McNicoll spoke to this as the power of having crowdsourced governance:“The way that we manage these little components of OpenEHRwhich is absolutely open source,crowdsourced,crowd managedevery archetype is its own little governance space.That is the genius thing,we have to let this stuff evolve.Its

200、on a fast,agile turnaround.”This means that the projects can adapt and be flexible to the community managing them and be as effective as possible.This community also provides an important support system for those who implement open source software.Beverley explained that,in their procurement choices

201、,they look for software with strong community support.“In five years,are there still going to be people on the job market who know this tool and are developing it and updating it,giving it security patches and all that.And thats beenweve had really positive experiences with open source tooling In an

202、 emergency situation,we can reach out and get help.”He also pointed to the community to describe how open source software outpaces proprietary offerings:“Theres a lot of really great tooling out there,and I dont think the proprietary stuff really keeps up a lot of the time with some of the open sour

203、ce tools.Id say its pretty infrequent that we hit a point where what is open source and battle tested over 20 years of communities using this softwareits pretty rare that we find that that doesnt fit our needs.”This community reduces the need for local expertise in a hospital setting.In Buckeridges

204、example of virtualization software,there are open source alternatives to VMWare that have reliable organizations behind them,such as Mirantis,Docker,IBM,and Red Hat.It also means that there are various options to use when seeking out a support agreement.25AN OPEN ARCHITECTURE FOR HEALTH DATA INTEROP

205、ERABILITYDespite Singhs and Clarks concerns about this aspect,open source reduces the reliance on one vendor,where a software such as the Linux operating system has the support of multiple vendors,such as Red Hat,Oracle,IBM,and Canonical,who can sign an agreement.developers trust the code,they will

206、want to continue to Overall,these benefits demonstrate the different ways that healthcare is a fit for open source,and some actually run counter to the perceptions of the previous sectionwhere removing vendor lock-in,de-risking innovation,and reducing costs demonstrate commercial viability and susta

207、inability,and the level of perceived support depends on how you choose to define the community and resources around your software stack.26AN OPEN ARCHITECTURE FOR HEALTH DATA INTEROPERABILITYOpen source healthcare solutions around the worldThere are a few regions and projects around the world that h

208、ave realized the benefits of open source for healthcare,adopting solutions and adapting them to their specific contexts.Interviewees gave examples of these projects and the context as to what conditions allow them to take hold.Open source has seen greater traction in healthcare systems in Europe tha

209、n in the U.S.and Canada.As a professor at a European university of technology explained,the coordination between E.U.member states leads to greater standardization and collaboration,two concepts that go hand in hand with open source:“In Europe,we have national health records in member states,and now

210、 we have cross-border solutions,and hopefully at a certain point we can have European health records that become a base or common ground for all other digital solutions in general.”McNicoll referenced an uptake of open sourcebased projects in Europe,pointing to how the regionalization in Europe crea

211、tes hubs for open source innovation where integrated and standardized systems can grow without the drag of prioritizing purely proprietary interests.He described developments in Catalonia,Karolinska,and London,where governments are adopting the open standards solution OpenEHR as a way to address the

212、 growing appreciation to separate the open standard data layer from the proprietary application layer.In India,digital health became the governments key mandate because of the cost of healthcare on the household in a country where“a large percentage of the population falls below the poverty line eve

213、ry year,”explained Gupta,who led the taskforce that developed open standards in India.He described the work of the National Digital Health Mission,which developed the open source“minimum viable product definitions and building blocks”for the country to develop its digital health ecosystem,following“

214、a very clear directive from the government that we do not want anything proprietary.”One significant outcome of this development was the Ayushman Bharat Health Account,a national digital health identifier that is used to build a longitudinal health record through all the health data that has been li

215、nked with that identifier.26 Kumbhat made the point that these open source solutions can be of benefit outside of India,as well:“Considering that India is a country of more than 1.4 billion people,theres a great amount of diversity in terms of maturity,technology levels,populations,and healthcare sy

216、stems,and therefore any kind of challenge that you will see in developing countries or even in the West is probably going to be a subset of what we face here.”More generally,Shannon pointed to open source adoption taking place in the developing world,where he has seen“a huge amount of movement aroun

217、d OpenMRS,an open source medical healthcare system.”He described it as“the leading open source healthcare system internationally Its all over the world,you know,its in 80 countries This tells you that this movement around open source is already ongoing.”These regional developments show a growing int

218、erest in implementing open source for greater standardization,reducing costs,and improving sustainability.The following textboxes are examples of two open source projects that are in use today:VistA and OpenEHR.27AN OPEN ARCHITECTURE FOR HEALTH DATA INTEROPERABILITYVISTAVeterans Affairs doctors and

219、technologists developed an open source EHR called VistA in the 1970s and 1980s.27 This system was built ground-up by doctor-programmer teams who were decentralized throughout the organization but saw the benefit of building on common infrastructure.This system took into account what the clinicians n

220、eeded from an information collection system,creating what Singh argued“is one of the best EHRs you will find.”An architect at a U.S.health department explained that“it was the culture of innovation,of decentralized development,that allowed VistA to evolve.”VistA took hold at the VA despite the top-d

221、own central office that slowly“chipped away at the VAs ability to innovate.”VistA stands alone as one of very few open source implementations at the hospital level in the U.S.OPENEHROpenEHR started in 2003 with the goal of establishing a new way of building lifelong health records based on open sour

222、ce licensed components and specifications.28 It defines a set of open source blueprints for standards-based data stores(clinical data repositories)and a library of healthcare data model components,which facilitates the building of healthcare applications and systems and is not an out-of-the-box open

223、 source medical application/solution.The licensing deliberately allows for the building of both open and closed source applications and technology solutions around this open core.“OpenEHR is about that data layer,”said McNicoll.He went on to explain,“It says nothing about applications or,indeed,abou

224、t the database technology that the OpenEHR stack sits upon.”This means that the software can be used interchangeably with any application that sits on top of it.Shannon explained that the philosophy behind OpenEHR was to avoid the deployment of a monolithic,singular solution in a particular setting

225、and instead incorporate a“building block approach.”The Ripple project adds several open source tiers to the core“data definitional”OpenEHR offering,including an open source UI framework and an open source clinical data repository store.29 McNicoll pointed out that OpenEHR implementations have been g

226、rowing recently.It has been picked up by commercial vendors,particularly in Europe and Australia,who felt that it responded to the challenges they were experiencing in“data management using traditional data stores because their data is so complicated.”As he pointed out,these vendors needed“a low-cod

227、e data solution that would start to make them much more flexible as a way of competing with the Epics and Cerners of this world.”28AN OPEN ARCHITECTURE FOR HEALTH DATA INTEROPERABILITYThe slow but clear adoption of open source in healthcare systems in Europe,India,and other areas of the world signal

228、s a desire to move away from the“archaic”infrastructure that burdens the healthcare sector and instead to embrace an approach that encourages innovation,efficiency,and agility.Shannon felt optimistic about this transition,arguing that“what will happen over time is that as people increasingly see how

229、 dysfunctional the current proprietary nature of the healthcare IT market is,and as these other alternative approaches gain traction,you will see a shift.Its going to take years for that to happen at scale,but that shift is now ongoing because,simply put,the current approach is just not sustainable.

230、”EXAMPLES OF GLOBAL HEALTH INFORMATION EXCHANGESDHIS2DHIS2 is an open source software platform for data collection,management,and analysis,widely used as a health information management system in over 100 countries,with national-scale implementations in more than 70 of them.The system supports multi

231、ple levels of data collection and analysis,including aggregate data(e.g.,routine health facility data,staffing,equipment,infrastructure,and population estimates),event data(e.g.,disease outbreaks,surveys,audits,and patient satisfaction surveys),and individual-level longitudinal data(e.g.,vaccination

232、 records,lab sample collection,patient treatment and follow-up,and student progress).Due to its scalability and adaptability,DHIS2 has applications in sectors beyond health,functioning as an information management,reporting,and monitoring system at district or national levels in areas such as educat

233、ion and civil registration.30 Surveillance Outbreak Response Management and Analysis SystemGerman and Nigerian public health and research institutions developed the Surveillance Outbreak Response Management and Analysis System(SORMAS)in response to the 20142015 Ebola epidemic in West Africa,and it b

234、ecame open source in 2016.The system focuses on disease surveillance,outbreak response,and management during epidemic and pandemic situations.SORMAS seeks to support the surveillance,management,and analysis of infectious diseases,providing public health agencies with tools to monitor infection rates

235、,manage caseloads,and conduct contact tracing.More than 15 countries currently use SORMAS for outbreak monitoring and response efforts.31 29AN OPEN ARCHITECTURE FOR HEALTH DATA INTEROPERABILITYAI the catalystAs in other industries that have embraced open source,a disruptive factor may speed up this

236、transition.For healthcare,AI could play that role,given its significant potential to solve problems and address constraints in healthcare.The opportunities for its application span activities of automation,detection,and predictionoften with more precision,and at much faster speeds,than healthcare pr

237、oviders.Interviewees listed a number of different AI use cases.In terms of automation,Singh focused on AI activities that bring down the costs of operation,such as building nursing schedules,while Muhammad Mamdani,Vice President of Data Science and Advanced Analytics at Unity Health Toronto,similarl

238、y explained the assignment tool that his team implemented at St.Michaels Hospital to assign nurses to different zones in the emergency department.A senior consultant at a global consulting firm discussed AI support in completing forms,including flagging issues as information is added to an EHR syste

239、m.Dalmia spoke about automating nurses check-up calls after patient discharge:“We can have conversational AI have a chat with them,get the notes,and see if it needs to be escalated to a nurse.That will free up so much nurse capacity.”For prediction,Mamdani discussed the CHARTwatch tool implemented i

240、n two hospitals in Toronto,which predicts whether a patient will die or go to the ICU to help prevent deaths by monitoring the patient and paging the medical team when needed.32 Dalmia also discussed the opportunity to predict how many patients may enter a hospital over the course of a few weeks to

241、better manage scheduling,with the overall goal of“helping the health system meet the demand that it is facing right now.”From a diagnostic perspective,Singh also pointed out that AI can help“for instance,with radiology and those things.The problem is that the FDA comes into play,”which makes adoptio

242、n in diagnostic use cases slower and more hesitant.Despite the promise of AI,interviewees also pointed out the current vulnerabilities of the technology and identified where bolstering is necessary to build sustainable AI infrastructure.Primarily,this technology has massive data requirements that ma

243、ny organizations cannot meet.33 Kumbhat stated that“AI is likely to fail qualitatively unless there is a ground-up flow of real-time,authentic data available.”Dalmia also highlighted this problem,saying that data is one of the biggest challenges for AI adoption.“Theres a lot of bad quality data,expi

244、red datasets,even datasets that might not be digitized and in manual fax format or paper format.So,in that environment,theres a lot of foundational work thats required to be done even before we go into any AI work.”Privacy and governance concerns came up as well as a way to build“trustworthy AI,”a t

245、erm Dalmia used.This involves AI models that incorporate bias and ethics mechanisms,with governance structures that maintain transparency and privacy in an environment as sensitive as healthcare.33 In terms of governance,A senior consultant at a global consulting firm brought up another challenge in

246、 that AIs incorporation into a hospital information system would require collaboration and agreement among doctors with varying needs and interests.“They would have to agree on whether this is a value to them as a group,not just individually,because you cant necessarily have AI for each clinician,”s

247、he pointed out.As discussed above,this kind of collaboration is complicated in the healthcare setting.30AN OPEN ARCHITECTURE FOR HEALTH DATA INTEROPERABILITYThe potential of this technology,and what it needs to move forward,creates an interesting liquidity moment,where AI could cause a wider and fas

248、ter shift in the digital health sector.He commented that,as it stands,the current system is not set up for AI to be truly transformative for healthcare.Shannon concurred.“Healthcare is going to get disrupted by digital,whether it wants to or not.AI is coming now.And so,the role,the process,around th

249、e knowledge capture and control by the profession alone is now going to be challenged by AI The disruptive forces on healthcare will continue to hit it until the change happens.Its inevitable.Its just a question of time,you know?”Dalmia made a similar comment:“I dont think there is an if,its a when,

250、But how will it be rolled out?And how do we make sure that its done in the best way,to benefit everyone?”Shannons position was clear:“We want the AI model to be open source so that it can be peer-reviewed,so that I know how the decisions are being made.And theres no black box there.”“AI is likely to

251、 fail qualitatively unless there is a ground-up flow of real-time,authentic data available.”Arun KumbhatDirector of GoToMarket,HR,and PR Services for Libra Social Research Foundation31AN OPEN ARCHITECTURE FOR HEALTH DATA INTEROPERABILITYMoving the sector forward:Recommendations from expertsAlthough

252、at times expressed as an intractable problem,interviewees gave a few recommendations to untangle data collection and management from the systems that maintain entrenched interests and a lack of interoperability.The following section reviews these recommendations and some next steps for the digital h

253、ealth sector.Building a digital health architecture As McNicoll clearly stated,“we need to fundamentally change the way that were architecting health systems.”The digital health sector is lacking an architecture that defines and standardizes the different components,nodes,and technologies within the

254、 system and how they all fit together.No particular company owns this architecture,since it is community developed,and it provides the protocol on which proprietary applications can be built and interoperate.This is similar to the World Wide Web,where emails from different providers can be sent amon

255、g themselves,and browsers from any vendor are able to access any website,because they are all based on a common set of standards and open source protocols.Other sectors,such as telecom and finance,have similar architectures:For example,the 5G Super Blueprint,which LF Networking developed,provides th

256、e instructions for pulling their sectors open source building blocks together and incorporates a landscape that shows where all telecommunications providers and cloud technologies plug in to this ecosystem.34 The Apperta Foundation suggests the adoption of an open architecture:“It is vendor and tech

257、nology neutral,eliminates lock-in,facilitates innovation,and forces vendors to compete on quality,value,and service.”6 A number of interviewees supported the development of an architecture,using different terms to describe this concept,including“digital backbone”(Gupta),“protocol”(Harlan),and“refere

258、nce model”(Shannon).Lee explained the value of the architecture for smaller vendors and startups,who can manage one“box,”or area,of the larger picture,which can be applied in different systems across different jurisdictions that are using this architecture.In this case,the EHR would represent one“bo

259、x”in the healthcare diagram,while other applications can individually support admissions,discharge,transfer management,after-visit summaries,and more instead of having to expand incumbent EHR systems to encompass more and more outside of their original intent.As McNicoll stated,building this open st

260、andardized architecture means that the startups do not have to“rebuild the whole infrastructure themselves”when they develop a solution in a specific corner of the sector.Building on top of an open,precompetitive architecture means that applications can interoperate with a federated set of systems a

261、nywhere.At this point,Shannon explained,the solution provider is“competing on the service rather than on the architecture.”Harlan made the same point,arguing that“where open source is at its best is when youve got protocol layer stuff,particularly around data exchange and information exchange,and tr

262、ying to get all of those pipes working together As you get closer and closer to the application layer,open source has less and less of a role to play.And thats really where organizations can build discrete products and compete in both the financial marketplace and the marketplace of ideas.”This will

263、 give health organizations“the ability to swap in and out different components,”explained Buckeridge.32AN OPEN ARCHITECTURE FOR HEALTH DATA INTEROPERABILITYWithout an architecture,innovators lack the ability to build an application that is sustainable,portable,adaptable,and interoperable for differe

264、nt health information systems.Buckeridge gave the example of the CODA infrastructure developed to share COVID-19 patient data in a federated way.35 He explained that it was hard to scale the project beyond COVID-19 data in the hospitals“because we didnt have the same architecture being governed the

265、same way within our environment.”Despite this project being open source,the underlying variation in infrastructure and governance across hospitals did not support scaling the project.Some floated OpenEHR,VistA,and other solutions as current systems that could be used as an architecture.An architect

266、at a U.S.health department sees VistA“as a foundation for the next generation of EHRs in the United States where the kernel is the interface layer and the database is the data management layer,as the foundation.Because the applications are very specific to your institution.”He explained how differen

267、t health services outside of the VA,including in New York,Jordan,and India,forked the VistA codebase and now all use this identical infrastructure,despite the fact that they have built their own specific applications on top of it.Shannon likewise advocated for the adoption of OpenEHR,which he descri

268、bed as a“future proof”architectural standard.CASE STUDY:PAYLOADSHarlan provided a clear example of where an architecture can bring greater efficiency to the healthcare system and ultimately better care to patients:payloads.Described as a referral that an organization sends to another individual that

269、 contains notes,health records,or other documents,Harlan explained that moving these payloads around means that they must be interoperable for different types of platforms to consume them and that they need to protect the personally identifiable information in them.A protocol for these payloads woul

270、d shape and standardize what the payload looks like,what the API endpoints are,and what permissioning looks like so that individuals can receive and participate securely in the process.It can also provide a map for application developers to provide tools that can plug into the system and participate

271、 in the protocol exchange.This approach would support greater interoperability within the health system but also address the“edge effect”described above,which is an important aspect of addressing SDOH,where health outcomes relate to processes outside of the traditional health system.33AN OPEN ARCHIT

272、ECTURE FOR HEALTH DATA INTEROPERABILITYStandardizing data,semanticallyA digital health architecture requires standardization,meaning that there is still a need for standards currently in use.“We still need the FHIR exchange,we need other standards such as OMOP We need things like SNOMED.The architec

273、ture is only part of the solution,”McNicoll stated.This includes standard data formats and standard APIs so that these applications can talk together.As the Apperta Foundation explains,building a“minimum viable”open platform that creates true portability and interoperability of health data and appli

274、cations needs to be based on core standards,such as FHIR HL7,SNOMED CT,or OpenEHR.6 Harlan pointed out the need for standardized terminology,such as the definition of gross income,in order to harmonize fields across organizations forms and remove ambiguity.In Europe,a professor at a European univers

275、ity of technology argued that standards are the critical piece to move data sharing forward.She said,“if we have standards and we have strategic support,that means resources to adjust the national data platforms Its the standards actually connecting the technology but also data transmission data saf

276、ety.Data sharing rules.”Advocating for the effective and widespread adoption of existing data standards,as Spence and Parisien are doing at Canada Health Infoway,is an important step in this process.Policy is a part of this;Dalmia argued that“as applications are modernized,there needs to be some goo

277、d policy that theres some data standards to adhere to that will allow us to share data.”He pointed to efforts in Canada to build the“lowest common denominator”of data fields for sharing across the country.The problem with this work,he stated,is that there is often a lack of funding and incentives to

278、 do it.When asked about a minimum dataset,McNicoll pointed out that this helps to“not try to boil the ocean,”but it jeopardizes the progress of standardizing when there are groups that fall outside of that minimum.“The trouble is,if you do that,because the data is so complex,you keep just having min

279、imal datasets all over the place.So,you dont pick up the edge cases for that one project,”he explained.His work mapping data with the NHS Christie cancer hospital instead tries to capture as many edge cases as possible.As discussed above,standards need to address semantic interoperability.An archite

280、ct at a U.S.health department spoke to the work being done by the World Wide Web Committee and the Solid Foundation to develop semantically interoperable technology in order to link data.He stated,“where we need to go is to adopt the World Wide Web standard for linked data as the data model for heal

281、thcare,which is a form of JSON-LD.”Open source is key to semantic data standards,according to McNicoll.“The sweet spot for the open bits is around the data and the data definitions.”OpenEHRs approach is to“separate the apps from the data,store the data in common ports using the OpenEHR standards and

282、 ask the applications to work directly on that common data.So,the standard is inside the data store rather than outside.”To build sustainable and lifelong records,standardization should be reconceptualized to consider health data as patient-centric as opposed to organization-centric.McNicoll explain

283、ed this difference by arguing that data should be“in a patient-centric data store,and everybody reads and writes to that as they go along We shouldnt have multiple copies of a patients problem list.We shouldnt have multiple copies of their allergies list;there should be one for the patient.And we al

284、l read and write to that.”He elaborated that there is an important distinction between patient-held and patient-centric,where he has seen confusion in the past:“Theyre two separate things.Patient-centric is about the way you organize the information.And actually,theres no necessity for the patient t

285、o be able to see or interact with that at all Patient-held records,or patient access to records,is a separate thing.Patients can still have access to organization-centric records.”34AN OPEN ARCHITECTURE FOR HEALTH DATA INTEROPERABILITYPatient-centric records mean more effective data capture,where in

286、tegrated data reduces duplication of effortswhich can have impacts ranging from freeing up resources all the way to saving lives.As an architect at a U.S.health department explained,“the only way to run a cost-effective,seamless healthcare system is to have integrated data.Otherwise,youll have dupli

287、cated tests,duplicative reports.And if you dont have a coherent,complete picture of a patients data,you cannot make good decisions.”Shannon also explained how standardizing data at the patient level is the lowest level,meaning you can go on to“feed other levels,you can then support the cohort manage

288、ment,and you can support the enterprise of the population.”From conversations with our interviewees,the key to moving to a more interoperable health data environment is in the standardization of data within an open architecture.As Harlan argued,“if we can standardize and create open source protocols

289、 around how we are moving data around then we essentially are creating something that doesnt get owned by any single entity as the warehouser.And,instead,all the participants in the ecosystem just simply stand up their kind of node,and they can begin interoperating according to whatever rules are wr

290、itten into the network.”This architecture means built-in interoperability on which innovators can build and,ultimately,from which healthcare stakeholders can benefit.“As applications are modernized,there needs to be some good policy that theres some data standards to adhere to that will allow us to

291、share data.”Niraj DalmiaPartner in Omnia AI at Deloitte Canada35AN OPEN ARCHITECTURE FOR HEALTH DATA INTEROPERABILITYTrying new business models:Innovation around incumbentsAs seen throughout this study,incumbent players hold a tight grip over data sharing.Instead of trying to reduce the market power

292、 of these organizations directly,some interviewees suggested working around them.Shannon gave the example of a“dual track approach,”where more innovative and agile solutions are built up in areas such as community care or where work is happening across borders,while the more monolithic structures re

293、main in the hospitals.“You accept that youre going to have some monoliths for some time to come,but you dont just leave those as they are,”he explained.“You also seed some parallel innovation,which can then grow from the bottom up.”Harlan also discussed how non-incumbent players are gaining wider an

294、d wider traction in this space.He gave the example of Amazon buying One Medical,where“now you have Amazon running a pharmacy,running a national healthcare network of primary care.”He posited,“the world today is filled with non-incumbent players who are gaining increasing amounts of resources to figh

295、t against incumbents,”who might be more interested in using products and services from innovative healthcare startups where there are options that allow them to use more open source software on which to run their applications.He argued,“I can pretty much guarantee you that Amazon is not beholden to

296、the wishes of an incumbent EHR.”This shift in ownership of health providers,at least in the U.S.,may open up the market away from the larger incumbents.Next steps:The role of the foundation How can we get to a point where we have a standardized architecture?As a professor at a European university of

297、 technology explained,this is not a job for one digital health company,whose incentives are to drive profit,not standardization and integrationthis requires a platform to govern relationships and standards.She suggested that,instead,“open source should take this role of providing a general platform

298、with governance rules with standards,you know,and really developing a medium for all the players to safely exchange data and services.”Harlan listed the stakeholders needed to collaborate on this kind of effort.“You need a not-very-large but strong set of foundational companies to stand it up.And yo

299、ure going to need probably at least one of the major health platforms,you know,preferably a couple of them,because if you can show that the two of them can participate,then that leads the way.I think you need some of the social care platforms.”Practically,he argued that the Linux Foundation(LF)may b

300、e the right group to bring these stakeholders together.“I think that the LF stands in a unique positionand I mean uniqueto bring together the only companies in the world,probably,that have this scale,”Harlan posited.Shannon described the LFs role in this collaboration from more of a philosophical pe

301、rspective,following a conversation on healthcare as a public good.He said,“I think the LF has done a really good job of tackling some of the big societal issues that are out there with open source,”and healthcare is the next one.The role of the foundation is not just to build the architecture but to

302、 create a meeting point for all the different open source work taking place in the healthcare sector.Scammon argued that there currently lacks an umbrella organization bringing together these various projects in healthcare.As he said,what the LF does best is that it“plants the flag,invites people in

303、to the tent,and then tries to create the community and the hype in the best possible way so that theres momentum and a center of gravity around this.And what will come out of it is a place where a whole bunch of these different disparate groups can come and share notes,gain visibility,and make their

304、 voices heard as a whole on a political and regulatory level.”Once these projects come together,activities naturally fall out of that collaboration,where people working across the sectorincluding people who dont36AN OPEN ARCHITECTURE FOR HEALTH DATA INTEROPERABILITYtraditionally work together,and ev

305、en beyond healthcarecan address common questions such as“How can we share data in a safe way?”“If you can bring all of them together,all of a sudden,you have a lot of voices who can contribute to a bunch of those conversations,”said Scammon.Leadership of this collaboration is crucial.“Whats required

306、 from a foundation is leadership in a complex environment,”Shannon explained.Leadership to align different groups around similar patterns and solutions,and leadership to also educate about what is taking place in this sector to solve these intractable issues.Before actually building the architecture

307、,stakeholders require education on the value of this architecture:“The common ground all jurisdictions share is a need for clinical and management and technical people to come together to support the care process with an architecture,”said Shannon,and leadership must demonstrate how everyone all fit

308、s together and benefits from this kind of collaboration.Scammon also discussed the value of education.He pointed out the misconception that open sourcing software will restrict an organizations ability to make a profit.He gave the example of a medical company he was working with that“was under the m

309、istaken belief that if it open sourced its work,that it wasnt going to be allowed to use it to make money.We had to sit down and repeat over and over,No,this is an Apache license,you can do whatever you want with it,including building a platform and using it to make money.Go for it.But it was becaus

310、e they werent educated on all of the different licenses and whether the one that we were using allowed them to do what they wanted.It takes a bunch of educationthats one of the ways that we can defang open source.”As seen in other industries undergoing technological transformations,the foundation pl

311、ays an important role in a context where multiple stakeholders with competing priorities and expectations must come together to build a precompetitive layer that all parties use and trust.“The common ground all jurisdictions share is a need for clinical and management and technical people to come to

312、gether to support the care process with an architecture.”Dr.Tony Shannon Head of Digital Services,Government Chief Information Officer,Government of Ireland37AN OPEN ARCHITECTURE FOR HEALTH DATA INTEROPERABILITYConclusionCurrent health data management systems around the world lack the sustainability

313、 to meet healthcare demands and technological advances such as AI.The complexity of health systems,the market dynamics,and the unique characteristics of the sector all lead to poor interoperability and innovation.There are strong philosophical,economic,and technological reasons why open source shoul

314、d be considered when looking for solutions,but many challenges and perceptions of open source hinder its adoption.This study calls for the adoption of an open architecture to build interoperability into the foundations of health systems,which will improve innovation,standardize efforts,and,ultimatel

315、y,improve health outcomes.A foundation would play an important role in the adoption,development,and maintenance of this architecture,bringing collaborators together to solve shared problems and benefit from a standard protocol on which they can build their own applications.The LF invites stakeholder

316、s across the sector to consider these recommendations and the role they can play in their adoption.Methodology The researcher interviewed 20 health technologists working across three different geographic regions in various health specializations.Members of the projects working group developed and re

317、viewed a question guide.The interviews were recorded,transcribed,and then coded following a thematic analysis approach.Codes were grouped according to their thematic overlap with each other,which developed into emergent topic areas on the uniqueness of the healthcare market;interoperability and data

318、 sharing issues;the perceptions,benefits,and adoption of open source in this sector;and actionable next steps.Findings were written up and peer-reviewed before publication.38AN OPEN ARCHITECTURE FOR HEALTH DATA INTEROPERABILITYReferences1.Aminpour,Farzaneh,Farahnaz Sadoughi,and Maryam Ahamdi.“Utiliz

319、ation of open source electronic health record around the world:A systematic review.”J Res Med Sci 19,no.1(Jan 2014):57-64.PMID:24672566;PMCID:PMC3963324.2.Theodos,Kim,and Scott Sittig.“Health Information Privacy Laws in the Digital Age:HIPAA Doesnt Apply.”Perspect Health Inf Manag 18(Dec 2020).PMID:

320、33633522;PMCID:PMC7883355.3.Arrow,Kenneth J.“Uncertainty and the Welfare Economics of Medical Care:Reply(The Implications of Transaction Costs and Adjustment Lags).”The American Economic Review 55,no.1/2(1965):154158.http:/www.jstor.org/stable/18161844.Wachter,Robert M,and Erik Brynjolfsson.“Will Ge

321、nerative Artificial Intelligence Deliver on Its Promise in Health Care?”JAMA 331,no.1(2024:6569.doi:10.1001/jama.2023.250545.Kannampalli,Thomas Gl,Guido F.Schauer,Trevor Cohen,and Vimla L.Patel.“Considering complexity in healthcare systems.”Journal of Biomedical Informatics 44,no.6(2011):943-947.htt

322、ps:/doi.org/10.1016/j.jbi.2011.06.006.6.“Defining an Open Platform.”Apperta Foundation.June 2018.https:/apperta.org/assets/Apperta_Defining_an_Open_Platform.pdf7.Shannon,Tony,“Healthcare:Generic Healthcare IT strategy:Options A/B/C.”Frectally Speaking.May 30,2014.https:/ Clinical Programmes:Aligning

323、 Process Improvements with Information Technologies.”Frectally Speaking.November 25,2011.https:/ 9.Huang,Fei,Sean Blaschke,and Henry Lucas.“Beyond pilotitis:taking digital health interventions to the national level in China and Uganda.”Global Health 13,no.49(2017).https:/doi.org/10.1186/s12992-017-0

324、275-z10.Padmanabhan,Paddy.“Is healthcare too hard for Big Tech firms?”Healthcare IT News.August 23,2021.https:/ L,Martin G Seneviratne,Matt L Berg,et al.“A full-STAC remedy for global digital health transformation:open standards,technologies,architectures and content.”Oxford Open Digital Health 1 (2

325、023).https:/doi.org/10.1093/oodh/oqad01812.HiTrust.Accessed October 9,2024.https:/ Big Medicine Can Learn from the Cheesecake Factory,”The New Yorker.August 6,2012.https:/ A Regier,Samantha Pollard.“Developing Data Sharing Models for Health Research with Real-World Data:A Scoping Review of Patient a

326、nd Public Preferences.”Journal of Medical Systems 46,no.2(October 2022).doi:10.1007/s10916-022-01875-315.“What are health data standards?”Canadian Institute for Health Information,Accessed October 9,2024.https:/www.cihi.ca/en/submit-data-and-view-standards/data-standards/what-are-health-data-standar

327、ds16.“FHIR Overview.”HL7,March 26,2023.https:/www.hl7.org/fhir/overview.html17.“Standardized Data:The OMOP Common Data Model.”The Observational Medical Outcomes Partnership.Accessed October 9,2024.https:/www.ohdsi.org/data-standardization/18.“What is SNOMED CT?”SNOMED International.Accessed October

328、9,2024.https:/www.snomed.org/what-is-snomed-ct19.“Learn LOINC.”Regenstrief Institute,Inc.Accessed October 9,2024.https:/loinc.org/learn/20.“Trusted Exchange Framework and Common Agreement(TEFCA).”Office of the National Coordinator for Health Information Technology.Accessed October 9,2024.https:/www.

329、healthit.gov/topic/interoperability/policy/trusted-exchange-framework-and-common-agreement-tefca21.“European Health Data Space:Empowering individuals and facilitating health data sharing.”European Commission.May 22,2024.https:/ec.europa.eu/commission/presscorner/detail/en/fs_24_134722.Drees,Jackie.“

330、KLAS:Epic,Cerner dominate EHR market share.”Beckers Health IT.April 30,2019.https:/ Electronic Health Records Market Analysis.”Insights10.May 10,2024.https:/ QJBvAbXlvl_6ikPY25DHZW2FeVOkv 24.Witowski,Nicole.“Top 10 EHR vendors in Canadian hospitals.”Definitive Healthcare.February 1,2024.https:/ Bhar

331、at Digital Mission.”National Health Authority.Accessed October 9,2024.https:/abdm.gov.in/26.Pharmaverse.Accessed October 9,2024.https:/pharmaverse.org/27.“About VistA.”World VistA.Accessed October 9,2024.https:/worldvista.org/AboutVistA28.openEHR.Accessed October 9,2024.https:/openehr.org/29.“EtherC

332、IS:Enterprise Clinical Data Repository.”Ripple Foundation.Accessed October 9,2024.https:/www.ripple.foundation/ethercis/30.“About DHIS2.”Health Information Systems Programme(HISP).Accessed October 9,2024.https:/dhis2.org/about-2/31.“SORMAS overview.”SORMAS Foundation.Accessed October 9,2024.https:/s

333、ormas.org/sormas/overview/32.Glauser,Wendy.“AI tool fundamentally changes the game for ICU care.”Healthy Debate.June 22,2021.https:/healthydebate.ca/2021/06/topic/ai-tool-changes-the-game-icu-care/33.Youssef,Alaa,Madelena Y Ng,Jin Long,et al.“Organizational Factors in Clinical Data Sharing for Artificial Intelligence in Health Care.”JAMA Netw Open 6,no.12(2023).doi:10.1001/jamanetworkopen.2023.484