Health Informatics 3.0

Health informatics 3.0

Dipak Kalra University College London

It’s time for health informatics to evolve it and in the formats (modality, device, Summary from enabling the information society level of detail) that are best suited to Web 3.0 promises us smart computer services that will interact with to creating the knowledge society: that individual at that time. This is in- each other and leverage knowledge about us and our immediate health informatics 3.0. deed a grand vision! context to deliver prioritised and relevant information to support de- The vision of the semantic web (a Healthcare now needs us to provide such cisions and actions. Healthcare must take advantage of such new critical element of Web 3.0) is of in- knowledge-integrating and collaborat- knowledge-integrating services, in particular to support better co- formation objects that are richly in- ing services: not as prototypes and pi- operation between professionals of different disciplines working in dexed by knowledge models and bound lots, but on a wide scale. different locations, and to enable well-informed co-operation be- to collaborating components within a Increasing sophistication of care, tween clinicians and patients. To grasp the potential of Web 3.0 we distributed computing ecosystem [1], with an ageing population [2] and more will need well-harmonised semantic resources that can richly con- bundled as smart services that interact and more illnesses becoming long-term nect virtual teams and link their strategies to real-time and tailored with each other to deliver personalised conditions (such as HIV [3]), requires evidence. Facts, decision logic, care pathway steps, alerts, education (tailored, relevant and prioritised) in- greater co-operation between profes- need to be embedded within components that can interact with mul- formation that supports the decisions sionals of different disciplines work- tiple EHR systems and services consistently. Using Health and actions of individuals and commu- ing in different locations, knowing Informatics 3.0 a patient’s current situation could be compared with nities. Services will collaborate to de- each others care goals, progress made the outcomes of very similar patients (from across millions) to deliver an integrated and holistic solution and any difficulties encountered. This liver personalised care recommendations. The integration of EHRs to its users. In other words, individuals needs more than sharing the record of with biomedical sciences (‘omics) research results and predictive will no longer utilise computers to what has been done (EHRs), but shar- models such as the Virtual Physiological Human could help speed search for and locate specific items of ing each other’s clinical care strategy, up the translation of new knowledge into clinical practice. The mis- information to digest and weight up, options and logistic constraints, to de- sion, and challenge, for Health Informatics 3.0 is to enable healthy but will pose problems and receive an- termine optimal ways of aligning ef- citizens, patients and professionals to collaborate within a knowl- swers to questions like: forts. We accomplish this best today edge-empowered social network in which patient specific informa- • What are my best care management by discussing patients in team meet- tion and personalised real-time evidence are seamlessly interwoven. options for this particular patient? ings. Physical (or online) meetings • What will happen to renal function could only cover a minority of patients, Keywords if I discontinue this drug? and team collaboration needs a way of Semantic web, semantic interoperability, patient engagement, per- • What is the optimal dose in this pa- scaling up virtually, through knowl- sonalised medicine, trusted services tient to maximise effectiveness and edge services. minimise adverse effects? The rapid pace of bio-science dis- Yearb Med Inform 2011:8-14 • How much should we budget for covery, the advent of personalised children’s cancer services next year? therapies, and the overwhelming vol- ume of new clinical evidence, com- Web 3.0 computers will work out what bined with demands for safe, high qual- information they need, where to get it ity and equitable standards of care, has from and how to access it. They will already encouraged professionals to filter and prioritise and personalise replace historic sources of knowledge what they find, and present the user (textbook plus apprenticeship plus an- with a polished set of reasoned options ecdote) with evidence based care. The from which to make a final choice. access to evidence has itself evolved Smart services will present individu- from individuals searching the litera- als with information when they need ture (with greater or lesser skills at

IMIA Yearbook of Medical Informatics 2011 9 Health informatics 3.0

critical appraisal) to peer-developed Much of health informatics, and guidelines that have filtered and com- also the needs of health care, have Harmonised Semantic bined published evidence into a speci- focussed up to now on managing col- Resources fication of best practice (50+ pages lections of data: for example improv- each!), and now progressively being ing interfaces and devices for data cap- The cornerstone of health informatics mapped into clinical workflows (as care ture, information models and databases 3.0, of collaborating components and pathways) to make them more usable. to store and process the data, inter- smart services, is knowledge integra- But we need to advance from these operability standards to communicate tion. Yet, today, our knowledge repre- forms of evidence (which are ulti- data and allow us to grow longitudinal sentations across healthcare, clinical mately informed by large expensive health records, single condition guide- research and bio-medicine are limited, studies that take years to deliver re- lines and simple decision support, and poorly interoperable. sults) to empirical data integrated analysis of reporting data sets and data Investments in health informatics re- across large populations, collected sys- warehouses, and a little bit of near pa- search, standards, product develop- tematically and therefore comparable: tient tele-care. The interpretation of com- ment and deployment are gradually real time outcomes. plex images and signals has perhaps had enabling the integration of data within Individualised evidence would take the greatest real-world impact so far. Bio- and between organisations. This is this a step further, in which a patient’s informatics is bringing us fresh ways to known colloquially as syntactic inter- current situation could be compared consider diseases and treatments, but we operability: the ability to exchange with the patterns and outcomes of very need to improve its integration with rou- data, but only being able to use it similar patients (from across millions) tine health care (beyond inherited disor- meaningfully if parties have agreed in in order to deliver point of care recom- ders and chemotherapy). advance on how they will represent mendations that are highly relevant. Such Health IT has so far helped us to meaning within the data. In the hier- case based systems already exist, but are cope with the increasing data volumes, archy of semantic interoperability lev- usually focused on a single condition, but as we finally succeed in joining els defined in the SemanticHEALTH intensively curated and only within one up disjointed parts of our health sys- Report [5], Web 3.0 services will re- (innovative) hospital. Linking case based tem how will we avoid information quire Level 3 interoperability: full se- analyses to emergent knowledge could overload? Will our clinicians and our mantic inter-operability with sharable radically shorten the “bench to bedside” managers simply get more data, or will context and seamless co-operability translation from over a decade to months. they have useful information presented (i.e. received data can be combined Such capability should be available to in ways that can be applied efficiently seamlessly with local data and proc- inform the routine care of all patients, and effectively? Will they spend more essed homogeneously). whenever needed. time sifting through health informa- We need guideline and decision sup- Most of our current guidelines are tion and published knowledge, or less? port systems, notification and alerting for single diseases, but patients rarely Will service management decisions components, and analytic tools to be have only one. The most realistic way (e.g. commissioning) be more wisely able to consistently process integrated of building up an understanding of informed or remain simplistic but health data drawn from multiple EHR how diseases and treatments interact, backed by larger volumes of data, as systems. New generation personalised and the optimal ways to manage mul- poor in quality as today? medicine needs the integration of EHRs tiple conditions, is to study (data mine) Next generation health care poses with biomedical sciences (‘omics) re- large-scale EHR repositories. Euro- some quite powerful challenges that research results and predictive models pean level disease specific repositor- quire us to harness and exploit knowl- such as the Virtual Physiological Hu- ies are already proving useful for edge at a much greater scale than we man. Clinical and public health re- knowledge discovery (e.g. [4]), but presently can. To grasp the potential search, and clinical trials, need to lev- these are expensive to create and main- of Web 3.0, we need to develop well- erage the value locked in our existing tain. Large population repositories harmonised semantic resources that can EHRs. The knowledge representations (ideally multi-national) are now richly connect virtual teams and link that underpin each of these sub-disci- needed. This will require health their strategies to real-time evidence plines need to be formalised and informatics to develop scalable and to enable better, safer and more effi- aligned. convincing solutions for privacy protec- cient care. We need to accelerate new Important progress has been made tion, de-identification, information- research and leverage better the value in this direction, and there are some knowledge integration and the real-time of bio-informatics research. We need early successes. We have known for analysis of distributed repositories, and to engage and empower patients and many years that prescribing decision friendly ways of presenting this knowl- healthy citizens in personalised support systems can help prevent seri- edge to busy clinicians. healthcare. ous error, and that they require access

IMIA Yearbook of Medical Informatics 2011 10 Kalra

to comprehensive allergy, diagnosis lections of coherent and quality-assured collaborating through the Joint Initia- and past medication data from the EHR semantic resources: clinical models tive Council) but more importantly in a processable form, cross-mapped such as archetypes and templates, between individual standards. These to data on drugs, their effects and ac- mapped to EHR interoperability stand- are still usually developed, balloted and tive ingredients. Such decision support ards and bound to well specified multi- published in isolation. Yet few has been shown long ago to work con- lingual terminology value sets, in- informatics solutions rely upon only sistently across vendor products [6]. To dexed and correlated with each other one standard: too little effort is invested provide this level of interoperability via ontologies and referenced from in working out and validating a solu- on a wide scale, the meaning of the modular care pathway components. tion-oriented portfolio of standards, clinical information within EHRs needs Similarly, libraries of interoperable and in delivering coherent guidance to to be formally and consistently repre- rules and queries to support decision- adopters (industry, purchasers and us- sented in order to be understood making, to generate alerts and to rep- ers). IHE is making a contribution, computationally and re-used by other resent the steps within care pathways, especially for DICOM standards, but systems whose information models are need to be built up and adopted con- solution orientation must be designed different, and be mapped to a wide sistently, so that services that use them into standards from the outset, not range of knowledge resources (such as behave uniformly and predictably. For worked out as a best fit afterwards. The conditional criteria within guidelines, example, a multi-disease care pathway terminology-record structure binding eligibility criteria in trial protocols, alert- should be able to seamlessly combine issue is a classical example of a cum- ing criteria in surveillance systems). the parts of single-disease pathways bersome solution to a largely prevent- Achieving such interoperability without inconsistency or duplication. able problem [9]. SDOs need to shift across the breadth of health and An antibiotic prescribing system their processes towards ensuring indus- healthcare is the challenge that needs should be able to check a patient’s try and user relevance, appropriate lev- urgently to be addressed. It is hard to documented health status including els of sophistication and complexity believe that the origins of the EHR ar- renal and liver function, pregnancy that are affordable and so can be read- chetype are now almost 15 years old status, allergies, recent antibiotic pre- ily taken up by markets, promoting and [7]. This was conceived as the most scriptions, the relevant clinical guide- supporting adoption, and learning from basic level of knowledge representa- line, the outcome of other patients re- experience. Many SDOs do almost tion for the EHR: standardised clini- cently prescribed antibiotics in that none of this today. Successful align- cal models to provide a common rep- hospital for the same indication, local ment will also depend upon the will- resentation to which data from and regional resistance patterns and the ingness of each SDO to cede certain heterogeneous systems can be mapped, hospital’s preferred formulary in or- areas of scope in favour of standards to support harmonised integration and der to present the prescriber with a developed by other SDOs. consistent analysis. The need to build prioritised short-list of options that are We have limited global experience up a library of clinical content of this all known to be safe and effective [8]. of developing harmonised collections kind has been accepted the world over, Multi-lingual resources are needed (libraries) of semantic resources of suf- but the value of authoring this knowl- to support cross-border care and to en- ficient scale. There is a paucity of ex- edge using a compatible approach (ar- able cross-border aggregation for re- amples of good practice in how seman- chetypes, or something else) and some search and population health manage- tic resources should be defined, degree of cross-specialty coherence ment. Multi-jargon resources are also validated and widely accepted, how us- seems to have escaped attention. Tens needed so that users of differing lev- ers should be trained to improve the of countries have each invested mil- els of expertise can review and under- quality and consistency of EHR docu- lions each in building up patchy, in- stand health information and knowl- mentation, and how to best enable pa- consistent and incompatible collections edge - for example a specialist and tients and healthy citizens to become of clinical models, some bound to ter- patient sharing the same information knowledge-empowered participants in minology, probably of variable qual- but viewing it differently. maintaining good health and manag- ity (since validation is nearly non-ex- Generating semantically coherent ing illness. istent) and with limited uptake within resources requires strengthening the It is high time for a multi- products. Yet, in 1996, this was only alignment of the various standards de- stakeholder international initiative to envisaged as the starting block, not the velopment organisations (SDOs) and build up this high quality clinical con- holy grail! investing in harmonising the artefacts tent, through a formalised, co- The challenge ahead is substantial. they produce. Health informatics ordinated and well-governed process. Useful semantic interoperability hinges standards development has historically Experts at a recent ARGOS project upon widespread and dependable ac- been piecemeal: not just between workshop in Washington [10] agreed cess to published and maintained col- SDOs (many of which are nowadays that semantic interoperability is too

IMIA Yearbook of Medical Informatics 2011 11 Health informatics 3.0

complex to be tackled right across clinical professional bodies and tentially valuable source of outcomes health care at once. Attention should eHealth programmes. data, if only we could harmonise the first be on areas of practice that can The benefits of semantic inter-op- information in each EHR and accu- leverage existing clinical consensus and erability occur some way downstream rately profile each patient to make accepted evidence, for example chronic of the investments needed in tools, valid and precise comparisons between diseases such as heart failure, popula- clinical content and in knowledge- fine grained sub-populations. Data tion health challenges such as child- driven systems. The lack of clear busi- warehouses currently support such hood obesity. Even within such a tar- ness models to justify these medium functions, but are usually limited to a geted strategy, not all clinical data to long term investments has probably small number of local data feeds (e.g. needs to be semantically processed. Pri- been a significant barrier to progress 3-4 hospitals) and require expensive ority should be given to the data that to date, and an alignment of value and time consuming mapping and have known computational value i.e. propositions is needed for this effort data cleaning before being usable. We for which there is a knowledge related to progress and scale up. This in turn have to get beyond this, to the point exploitation, for example care pathway requires inter-governmental strategic where EHRs are populated in the first support or patient safety. Research is co-operation. place with high quality and semanti- needed on the criteria that help to de- cally sound information, and can termine what parts of, and how much then be co-analysed consistently to of, a health record is useful to struc- provide a virtual population denomi- ture and code to achieve Level 3 nator of millions. Research is still interoperability. Accelerate and Leverage needed, in particular on how to de- Starting with priority areas does not Knowledge Discovery identify fine grained clinical data (as imply that a piecemeal solution will opposed to the simple removal of de- work: semantic interoperability is a ho- The present gold standard for new mographic data items) and how to as- listic problem, and needs to be addressed clinical evidence is a clinical trial or a sess and compensate for variable data through a coherent strategy. The start- population health (cohort) study: both quality within EHRs. ing points need to be clearly positioned of these are expensive to undertake and Secondly, we need to speed up the on a roadmap towards richer inter-op- may take years to deliver results. Pub- transfer of newly discovered knowledge erability, and the approaches scalable lished study results that could impact into clinical practice. For readily us- to more complete semantic inter-op- on strategic clinical decisions then take able clinical knowledge, such as a new erability as an evolutionary process. further years to influence routine clini- therapeutic indication, we know that It is probably fair to say that for- cal practice on a wide scale. This integrating that knowledge into clini- mal knowledge representation (ontol- “bench to bedside” delay of up to a cal information systems that are used ogy) is still primarily seen as a research decade prolongs the use of potentially at the point of care will increase adop- capability in health informatics rather inefficient and ineffective practice and tion. The more challenging innovations than ready to scale. Yet ontology will risks preventable harm to patients. The to adopt are radically new ways of be vital as the means to broker across EU eHealth for Safety study [12] an- thinking, such as molecular medicine different knowledge representations, ticipates modelling and simulation and predictive medicine. between models of meaning and mod- tools to have a significant impact on Much of molecular medicine fo- els of use [11], between different com- patient safety through better predic- cuses on cellular and sub-cellular find- munities such as bio-science research, tion, prevention and personalisation. ings, and maps these to phenotypes. population health research and clini- Firstly, though, we need to acceler- Phenotypes are coarse simplifications cal practice, between service provid- ate the discovery of new knowledge of a person’s health conditions and ers and service planers. Each of these from large populations of existing health status: they might be suitable to stakeholders needs their own view of health records. Data mining research support molecular knowledge discov- the knowledge space. This is not solved does not necessarily replace the value ery but are not usually sufficiently rich by one or a few giant ontologies, but of clinical trials, but it can flag up new to selectively target this knowledge in by focussed ontological services that hypotheses that need further investi- clinical practice. There is a need for are each fit for a dedicated stakeholder gation. EHRs can provide population bio-informatics research to integrate group and purpose - at the appropriate prevalence data and fine grained co- more fully with the totality of a pa- level of detail and coverage - yet are morbidity data to optimise a research tient’s clinical picture and clinical able to collaborate with others. protocol, and help identify candidates course, not just selected phenotypic Growing this body of harmonised to recruit (almost half of all pharma data items, and therefore to integrate semantic resources should now be a Phase III trial delays are due to recruit- with the EHR. This is vital to support high priority for health informatics, ment problems [13]). EHRs are a po- fully personalised medicine.

IMIA Yearbook of Medical Informatics 2011 12 Kalra

Predictive medicine, such as that ena- physical and virtual team-working that with aspects of self care, mainly real- bled through the Virtual Physiological should be anticipated and prepared for. ised through Personal Health Record Human [14], offers an exciting new Much of the pure research from bio- (PHR) systems, which inform and en- way of communicating to clinicians and science still needs a translational phase gage patients in health and social care, to patients how subtle changes in a in which this applied knowledge can or enable individuals who are not need- treatment or health condition might be identified. This additional stage of ing or seeking health care to keep track impact on body function and eventu- research is vital if next-generation of their level of fitness, manage pre- ally outcome. Much of the modelling, (3.0) computational services are to vention or monitor health status. This though, has been developed in vitro make best and safest use of new knowl- sector is growing rapidly. Although using test data, or clinical data that have edge to provide sound recommenda- new entrants to the field (such as been specifically extracted and tions to clinicians. Google and Microsoft) offer ways of cleaned. This has been necessary to de- sending information to clinical sys- sign the models, but is not the reality tems, there is as yet no real re- of the data on which they will be run architecting of healthcare services in practice: clinical data are full of around the patient. noise, missing values, and some data Patient and Citizen Centric The distinction between a PHR and entry errors. Research is needed on the Services an EHR is not hard and fast: there will tolerance of the models to these noisy be kinds of shared care records where conditions, and on ways of measuring There are growing pressures to recog- the distinction is neither clear nor de- and presenting the level of confidence nise the central role of the public as sirable: i.e. the patient and his or her of a prediction. informed partners in decisions about care team might share and manage a Just as healthcare has been a cottage their personal health care, the safety common health record transparently. industry of parochial standards, so and efficacy of treatment choices and In parallel there is a need for individu- these scientific communities have each in service priority setting. People are als to engage in and take greater own- standardised within silos, optimising increasingly expecting to exercise per- ership of their health and healthcare, internal sharing but not facilitating in- sonal and informed autonomy over including wellness (which existing tegration or collaboration across health their health care, and we have known health services do rather poorly). These and bio-science. Cross-disciplinary se- for years that individuals can acquire are challenges for which the semantic mantic interoperability is now needed. considerable expertise in managing ill- web seems ideally suited: patients and The acceptance of new ways of man- ness and preventive health if they are healthy citizens will be able to provide aging conditions, and of making stra- given useful and appropriate material data or pose questions to services that tegic decisions (such as a chemo- with which to educate themselves and provide intelligent feedback on their therapy agent choice based on the tools with which to participate (e.g. data and measurements, perhaps based bio-markers, or an infusion rate based [15]). In many countries patients are on simulation models that take account on cardiac modelling) needs to recog- finally being encouraged (or are them- of their past history and present health, nise that clinical decisions are often selves pushing) to participate more medication, lifestyle and environmen- based on a portfolio of disease/stage/ strongly in the management of their tal factors, and genetic profile. Smart severity/comorbidity and person/pref- health and healthcare. services might provide significant sup- erence/logistic influences. The ad- However, there is a critical gap in port for self-care, only occasionally vanced modelling research communi- the design and delivery of health care escalating a situation to a healthcare ties need to understand better the kinds services that fails to harness the im- professional. of diverse inputs that different profes- mense contributions that patients and However, the success of patient en- sionals and specialities have to inter- citizens can make. These contributions gagement also critically depends upon pret, the kinds and quality and time are now vital as healthcare needs and attitudes and culture change: policy, or- spans of data that need to be co-inter- costs make existing models of service ganisational, insurance and reimburse- preted, the nature and criticality of the unsustainable, and as societal pressure ment changes, and it has ethical im- decisions being made, and how accu- and the growth of health consumerism plications. These changes are, in part, rate the modelling projections would demands the respect and participation needed because patient centred serv- have to be in order to be useful. Re- of individuals in healthcare decisions ices will include actors whose contri- ciprocally, clinical communities need and delivery. There needs to be mean- butions conventional health services to understand better what future op- ingful sharing of knowledge between today largely ignore: social care, domi- portunities and solutions are in the clinicians and patients. ciliary care, health charities, comple- pipeline, how these might impact on A major change is taking place in mentary therapists, families etc. In our care decisions, any adaptations to systems and services to support patients design of future systems we also need

IMIA Yearbook of Medical Informatics 2011 13 Health informatics 3.0

to recognise that people are part of of trust are less tangible, and hinge grate medical school knowledge and families and communities, and do upon the evidence basis for clinical experience with longstanding famili- sometimes share and collaborate with assertions documented within the EHR arity with the patient, glancing occa- others (for example, Patients Like Me and the seniority/expertise of the per- sionally at some scrappy paper notes [16]). Healthcare and eHealth services son making the assertions. as an aide memoire. Personalised medi- probably have a lot to learn from the Many clinicians still need to learn cine would not be new to him either, social social computing paradigm about how to trust patients and citizens with since every decision would be made in ways in which people interact, share the data they provide, for example in full awareness of the patient as a per- information and collaborate, and their a PHR or (if permitted) into an EHR. son and as a member of a family and attitudes to privacy in these contexts. The paradox of this is that clinicians community, and a holistic knowledge have little difficulty in trusting a pa- of all of the patient’s health conditions tient’s symptom history (e.g. a diag- and lifestyle risk factors. nosis of migraine, that is almost im- What has changed is the scale, com- Trusting Clinical Information possible to confirm except through the plexity, mobility and quality expecta- history). With the proliferation of in- tions of modern health care, the rapid Investments in interoperability stand- creasingly portable, wearable, friendly pace of change in knowledge, and ards, interoperable EHR systems, na- and connected devices playing vital growing needs for the re-use of health tional eHealth architectures, patient roles in next-generation tele-care and information for many, valuable, non- and provider registries and progres- self-care, this barrier to trust will have clinical purposes. sively more coherent confidentiality to be overcome. As the data volumes have increased and security policies are making it It is less clear still how clinicians we have elaborated and later outgrown possible for health record information will trust information that has been sig- paper records. We are making progress to genuinely follow the patient be- nificantly pre-processed by computers, on electronic health records as reposi- tween care providers. However, most for example a treatment recommenda- tories of data but not yet into shaping professionals are used to receiving in- tion. The cumulative experience in the EHR as a knowledge management formation from external colleagues in decision support to date suggests that and knowledge discovery tool. Our no- quite formalised ways: reports (e.g. clinicians are happier to be prompted tion of shared records is shared access from a lab or radiology department) or alerted about things they might have to each of our own electronic records, or letters (referral, discharge, outpa- missed in a health record or in a drug not yet to collaborate on a systemati- tients etc.). Trust in the information formulary than they are being given cally organised collective record. We provided by colleagues is mixed: there “clever” guidance about a clinical are still cautious about trusting and us- is quite a strong tradition of repeating judgement. The vision of 3.0 is going ing information from outside of our tests that have just been carried out in to challenge clinicians with informa- own team or institution, and perhaps other hospitals, re-checking details tion that has been derived from a com- even more cautious about trusting in- such as the present and past history with plexity of information and knowledge formation entered by patients. We need patients, and operating a tacit "hierar- sources which they could not practi- to progress from integrated electronic chy of evidence" in which more lo- cally recreate (and therefore validate) health records to collaborative health cally sourced information is trusted manually i.e. there may be limited records: semantically interoperable and more than information from far away transparency. Heath informatics needs supporting virtual teamwork - includ- and unfamiliar or overseas organisa- to drill into this issue of trust, and seek ing the patient’s care network. tions. Information provided in less for- out ways in which evidence of trust- The pace of medical discoveries and malised ways (without headed note-pa- worthiness can be compiled by col- bio-science advances has not only per, undated, unsigned etc.) is usually laborating services and be available for made it impossible to memorise all disregarded. scrutiny before a recommendation is of what needs to be known (antici- Professionals are certainly not used accepted. patory learning), but impossible to to sharing original EHR data on a wide look up what is relevant on a need to scale: this capability is relatively new. know basis (just in time learning). It They will need to learn how and when is also no longer good enough for they should trust remote EHR infor- Conclusion knowledge to be incorporated (locked) mation: its provenance and its cer- into specific applications. Facts, evi- tainty. EHR architectures contain prop- An integrated information and knowl- dence, decision logic, care pathway erties that capture and communicate edge environment is not new to steps, alerts, education need to be em- some of this provenance information healthcare: an old style village GP bedded within components that can in- to EHR data recipients. Other aspects would have used his memory to inte- teract with, and thereby serve, multi-

IMIA Yearbook of Medical Informatics 2011 14 Kalra

ple systems and services consistently, (August 14, 2007) Vol. 104 No. 33. p. 13217-8. Semantic Interoperability. ARGOS is a Transatlantic and be more easily updated. 3. The UK Collaborative Group for HIV and STI Observatory for Meeting Global Health Policy Challenges through ICT-Enabled Solutions. (To be Health informatics 2.0 has delivered Surveillance. Testing Times. HIV and other Sexually Transmitted Infections in the United Kingdom: published summer 2011.) Please see http:// many significant research, standardi- 2007. London: Health Protection Agency, Centre www.eurorec.org/RD/ARGOS.cfm sation and implementation building for Infections; November 2007. 11. Qamar R, Rector A. Semantic Issues in Integrating blocks for global eHealth. The mission 4. The European Society for Immunodeficiencies Data from Different Models to Achieve Data registry. Please see http://www.esid.org/ Interoperability. Proc Medinfo 2007 Kuhn K, Warren for Health Informatics 3.0 is to enable JR, Leong TY, editors. IOS Press; 2007. workingparty.php?party=2&sub=2&id=233 [Last citizens, patients and professionals to 12. Veli N. Stroetmann, Jean-Pierre Thierry, Karl A. accessed 30 April 2011] work together within a knowledge so- Stroetmann, Alexander Dobrev. eHealth for Safety 5. Stroetman,V., Kalra,D.,Lewalle,P.,Rector,A., - Impact of ICT on Patient Safety and Risk ciety, informed and advised and em- Rodrigues,J., Stroetman,K., Surjan,G., Ustun,B., Management: European Commission. eHealth for powered by Web 3.0 services in Virtanen,M., Zanstra,P. (2009), Semantic Safety Report, October 2007. Luxembourg, Office which patient specific information Interoperability for Better health and Safer Healthcare for Official Publications of the European and personalised real-time evidence [34 pages]. The European Commission. ISBN-13: Communities; 2007. 978-92-79-11139-6. DOI: 10.2759/38514. http:// 13. Study participant Recruitment and Retention in are seamlessly interwoven. ec.europa.eu/information_society/activities/health/ Clinical Trials: Emerging strategies in Europe, the These are primarily challenges for docs/publications/2009/ 2009semantic-health- US and Asia. Business Insights; June 2007. health informatics to deliver, but we report.pdf 14. Please see http://www.vph-noe.eu/ also need to recognise these (posi- 6. Rogers J, Jain NL, Hayes GM. Evaluation of an 15. Jones R, Pearson J, McGregor S, Cawsey AJ, Barrett A, Craig N, et al. Randomised trial of personalised tively) disruptive innovations will re- implementation of PRODIGY phase two. Proceedings / AMIA Annual Symposium. 1999; computer based information for cancer patients. BMJ quire acceptance (trust) and adaptation 604-8. Nov 1999; 319(7219):1241-7. by health professionals and health sys- 7. Kalra D., Editor. The Synapses Object Model and 16. Please see http://www.patientslikeme.com tems if their benefits are to be fully Object Dictionary. EU Telematics Application realised. Programme, Brussels; 1997; The Synapses Project: Deliverable USER 1.3.2. (Copy available from the author on request) 8. Lovis C, Douglas T, Pasche E, Ruch P, Colaert D, Correspondence to:

Stroetmann K. DebugIT: building a European Dipak Kalra References distributed clinical data mining network to foster Clinical Professor of Health Informatics 1. Berners-Lee T, Hendler J, Lassila O. The Semantic the fight against microbial diseases. Stud Health Director, Centre for Health Informatics and Multiprofessional Education Web - A new form of Web content that is meaningful Technol Inform 2009;148:50-9. University College London to computers will unleash a revolution of new 9. Krog R, Markwell D, Dolin R, Gabriel D, Cheetham Holborn Union Building, Highgate Hill, London N19 5LW possibilities. Scientific American, May 17, 2001 E, Spackman K, et al. Using SNOMED CT in HL7 Honorary Consultant, The Whittington Hospital NHS Trust, London 2. Dora L. Costa. The economics and demography of Version 3; Implementation Guide, Draft Standard United Kingdom aging. In PNAS - Proceedings of the National for Trial Use. Health Level Seven; 2007. Tel: +44 20 7288 5966 Academy of Sciences of the United States of America 10. Kalra D, Musen M. ARGOS Policy Brief on E-mail: [email protected]

IMIA Yearbook of Medical Informatics 2011