Clinical Decision Support
•Download as PPTX, PDF•
1 like•440 views
OpenClinical is an open source clinical decision support (CDS) platform that provides pathways and computer interpretable guidelines to help clinicians make better healthcare decisions. It aims to be intelligible, personalizable, justifiable, and give clinicians control. The platform can integrate with electronic medical records and be used for automated tasks and research databases. It focuses on usability, risk communication, visualizing patient journeys, and mobile access. Potential risks include maintaining high quality evidence over time, siloed care processes, reduced opportunities for discovery, and communication errors that can impact cognitive framing.
Recommended
Recommended
More Related Content
Viewers also liked
Viewers also liked (15)
Similar to Clinical Decision Support
Similar to Clinical Decision Support (20)
More from Health Informatics New Zealand
More from Health Informatics New Zealand (20)
Recently uploaded
Recently uploaded (20)
Clinical Decision Support
- 1. OPENCLINICAL.NET CLINICAL DECISION SUPPORT Dr Marc Gutenstein FACEM
- 5. Tallis
- 23. OpenClinical A Complete CDSS platform Based on robust decision-making theory Open source for research into CDSS Platform for sharing and collaboration ! Usability ! Implementation into clinical systems
- 25. Patients and Clinicians Intelligible Personal Justifiable Controllable
- 26. Clinical Use
- 27. Workflow Better Decisions EMR Research Database Automated tasks
- 28. Usability Best UX Risk Communication Journey Visualization Mobile
- 29. Risks of CDSS • Evidence quality & maintenance • Silos of process • Loss of discovery space • Communication and linguistic errors • Cognitive framing errors • Loss of learning & desirable difficulty
- 30. Conclusion
- 31. Conclusion
Editor's Notes
- In this talk an introduction to a CDSS open tool called openclinical along with my reflections as a working clinician into how to make CDSS tools succeed in the clinical environment
- Introduction: Personal First I should state my particular perspective & my journey in e-health. I have no conflict of interest in sharing this talk with you. Emergency Medicine environment is rich in workload, decision intensity, IT use, communication and interruption, and clinical scenarios are often life threatening and time-critical. A subject-matter expert in Emergency Medicine, decision-making, communication and workflow. The decision support agenda is interesting because it presupposes that decisions require support and that e-tools will help patients. But in the rest of medicine we demand evidence for common sense interventions
- Pathways CDHB, ACC, NICE are just a few example organisations with published clinical pathways. Pathways have evolved as the movement of evidence-based medicine has grown. They are more than just clinical guidelines rewritten. Although this function as a reminder of best practice is central, they crucially also act as agreements between services, reducing variability in care, obviating the need for renegotiation, and justifying resource use. This is no doubt a good thing for busy clinicians in complex work environments. ----- Meeting Notes (13/10/15 18:03) ----- 5 mins max.
- OpenClinical – Website Screenshot OpenClinical is an effort to bridge the evidence gap Evidence based assessment of DSS as an intervention along with an opportunity to create a commons for CDSS tools
- Tallis - Screenshot Is a Java based application that enables the user to create a decision support pathway using both graphical tools to schedule the skeleton of the decision process, and logical expressions to code the decision arguments and conditions within each decision step. The logic is based on a language called pro-forma, which was developed by the Oxford team to emulate the process of human decision-making. As such it has a robust and versatile logical foundation
- Pathways translated to CDSS applications With Tallis tools can then be published to the website If medical evidence is the written direction of care, then pathways are the evolution of medical evidence into map form, and encoding pathways into CDSS tools is the further evolution into the GPS application. The primary point of the CDSS application is to actively guide the user and therefore reduce the chances of taking a wrong turn. Here follows some example applications I created in Openclinical.
- OC Demonstrations ----- Meeting Notes (13/10/15 18:03) -----
- Openclinical as research base for CDSS The site also aims to be a library of resources and a research centre for trials of CDSS applications. Importantly the licence is a research licence for use to trial at this meta-level of proving the decision support technology, not prove actual clinical care. This is an interesting concept, as trials of CDSS are not great in number and clinical outcomes not dramatic. An interesting question arises, akin to a parachute trial – if we truly believe in our clinical research and stated aims to implement this, then do we need to trial it? The answer is complex! However despite no trials of parachute use when jumping from planes, we firmly believe in their efficacy and promote their use. ----- Meeting Notes (13/10/15 18:03) ----- 10 mins
- OC pros and cons The inextricable link between the content of what we’re deciding… And the process of making a decision and the commitment to follow that pathway… So the best we might achieve with this system is to trial complex decision flows outside of the clinical environment Or perhaps use as educational aide memoires Consider the analogy A portable computer vs. A iPhone A mapping application vs. A GPS watch
- The gap So if we’re really going to leverage CDSS we need to step up a notch… Assuming decision quality needs to be optimised is fine… But creating technically accurate but poorly useable applications will not show a benefit The usefulness problem – is it helpful The usability problem – is it easy The validity problem – is it relevant The reason spectacular M&M gains have not been seen so far The control group do pretty well The incremental gains in efficiency do not often translate to life or death results in the short term (but rather population gains in the long term) We mistakenly think of the content of decisions being all important where the process of decision making is key The gap between clinicians and solution architects You think we know the best way to treat people vs. medicine based on transitory best practice in a complex environment You think the content of our decisions is separate from the way in which we make them You think all we have is a technical challenge So the intent to tidy up decision making and standardise as a technical challenge is not the whole solution (image: sbts.edu)
- Implementing CDSS – Clinicians and patients As CDSS trickle into our work environment it is worth being cognizant of clinician and patient factors that are crucial to a projects success. Firstly healthcare = a combination of competence and care. Or put another way both technically rational decision making and an emotionally relevant exchange. Secondarily that HCI is more central that most would believe (as if somehow the UX is superfluous because the technical content is so important) The third man problem CDSS tools can be like an annoying third person in the room, somewhat unwanted. Stifling a good conversation like an iphone on the table Patient Pushback “I went to the doctor and all he did was look at his screen, and recommended me a statin. He didn’t even check my cholesterol or talk to me about it!” Patients as individuals… Clinician Pushback “I hate this tool, it’s too prescriptive and medicine is not like that!” Doctors as artists… Linguistic Dangers – The communication space Non-expert operation of algorithmic medicine may increase demand (Image - http://observatorioesclerosismultiple.com)
- Human Factors approach – watch what people actually do in an effort to do their job well These tools have to feel helpful, and then rather than demand extra work from us, reduce our work Consider 1 click and 90,000 patients on a 2 second delay = 50 hours of spinning wheels per annum No-one designing a new messaging iPhone app would think functionality will sell in isolation of usability, speed and aesthetics. A Human Factors Approach to what really happens in a hospital Take photos on camera phones because hospital tech too clunky Send texts from private phones because hospital paging and messaging systems have been too slow in development Check drug doses on privately bought apps because the UX of formal guidelines difficult Open private browsing windows because 2 windows is better than one Login on other peoples cards because the privacy policy too strict (worse than poor accountability is wrong accountability)
- Leveraging CDSS – Our intended project To make this work it is more than decisions that matter To maximize value from a CDSS project it is best to leverage what we’re doing by running the workflow. The cardiac chest pain pathway is a successful clinical pathway that is being rolled out across the South Island NZ. We are aiming to leverage the pathway so that invisibly it is doing much more than help decisions. Enhance quality decision making Create a clinical record Add to a database of patient journeys for audit, research and data visualisation Automate actions and tasks We have to make these tools feel useful primarily, then hide the functionality that really works! ----- Meeting Notes (13/10/15 18:03) ----- 15 mins
- Leveraging CDSS – Our intended project To make this work it is more than decisions that matter UX is crucial. Lessons from web design and mobile technology How do people understand risk? Statistics, Charts, Stick figures…. NOT the same impression People focus more on humanoid individuals People have more aversion to risk than attraction to reward So an easy calculus between benefit and harm is not necessarily easy to do UX and visualisation key to adoption of technology
- Risks of CDSS Evidence quality & maintenance (parachute analogy) Silos of processes Loss of discovery space Communication and linguistic errors Cognitive framing errors Loss of learning & desirable difficulty The Dangerously Easy, The Desirably Difficult and the Dastardly CDSS that is too easy can compel us to do things that don’t add value, and are probably dangerous. For example a mere mention of chest pain can compel an ambulance to be dispatched, an ED to organize tests, and possibly intervention to occur, with the risk that this involves. In the sweet spot lies the desirably difficult. Requiring human work within the communication space, forcing thought and conscious action and then improving decision making in difficult domains, and allowing people to learn while they connect. This adds value in a non-obtrusive way and will be welcomed. The dastardly is the arena for technically adequate but effectively useless IT as it is too hard, too parallel or too disruptive to workflow. If we want to permanently shift to a new space in medicine that’s fine. But is it what we want? Star Trek: “When you create a machine to do the work of a man, you take something away from the man”
- Grand Challenges (from paper) D.F. Sittig et al. / Journal of Biomedical Informatics 41 (2008) 387–392 Improve the Human Computer Interface Summarize patient level information Prioritise and Filter recommendations Combine recommendations for patients with comorbidities Use freetext information Prioritise CDS content development and implementation Mine databases to create new CDS Disseminate best practice Create an architecture for sharing executable CDS Create internet accessible repositories Gutenstein Golden rules of CDSS Enhance decision quality Part of a clinical workflow (invisible to clinical workflow) Easy and compelling UX Automates tasks Database pathway journeys for research & visualisation Create Clinical Record Allow freetext variation Maintain clinician decision autonomy Maintain patient individuality of values Permits discovery moments Care with linguistically rich decisions (unless supported by Natural Language Processing or alternate analyses)
- Grand Challenges (from paper) D.F. Sittig et al. / Journal of Biomedical Informatics 41 (2008) 387–392 Improve the Human Computer Interface Summarize patient level information Prioritise and Filter recommendations Combine recommendations for patients with comorbidities Use freetext information Prioritise CDS content development and implementation Mine databases to create new CDS Disseminate best practice Create an architecture for sharing executable CDS Create internet accessible repositories Gutenstein Golden rules of CDSS Enhance decision quality Part of a clinical workflow (invisible to clinical workflow) Easy and compelling UX Automates tasks Database pathway journeys for research & visualisation Create Clinical Record Allow freetext variation Maintain clinician decision autonomy Maintain patient individuality of values Permits discovery moments Care with linguistically rich decisions (unless supported by Natural Language Processing or alternate analyses)
- ----- Meeting Notes (13/10/15 18:03) ----- 20 mins ideally
- Conclusion A tour of Openclinical Clinican level application development Research into CDSS application of clinical pathways Sharing platform BUT… Even if we move towards algorithmic medicine, both doctors and patients have non-algorithmic needs. We all take a step and a half towards perfect rationality but not the whole way. My rules for implementation Make CDSS a reliable and empathic friend, that not only tells us what to do, but quietly and delightfully helps us deliver it Psychology of patients and doctors UX considerations Clinical workflow Just like cartography led to maps which led to digital maps, satnav and eventually GPS to improve the reliability of our journeying… …It is only when you can fit it unobtrusively in your pocket or on on your dashboard, it suggests which turn to take but can cope with variation to this, and even helpfully books your tickets …then we will really take off. ----- Meeting Notes (13/10/15 18:03) ----- 25 mins max