Mini-symposium 4 - Implementing a Learning Health System for diagnosis in Primary Care

Talk Code: 

Aim Diagnosis is one of the primary tasks of the GP, diagnostic error is also the single greatest reason for litigation against GPs. Decision support is one possible intervention to help with this task, but existing systems lack evidence-based content, integration with the electronic health record (EHR), and have been designed with no regard for either the cognitive tasks of the GP or the human-computer interface needed to support them. Further, conducting large studies to develop diagnostic evidence is notoriously difficult and costly. Such a problem is an ideal target for the ‘Learning Health System' (LHS), describing a digital infrastructure that supports research and knowledge translation activities as part of routine health IT. FP7-TRANSFoRm, an EU large collaborative project has supported the development of a LHS for diagnosis in primary care that will have completed its evaluation by May this year.The symposium aims to:

  • 1. Discuss the wider concept of the LHS and its potential impact on patient safety, research and implementation science.
  • 2. Examine recent experimental evidence from the field of psychology on reasons for diagnostic error, and potential interventions.
  • 3. Examine the diagnostic task and the use of evidence from an informatics perspective.

Consider the technical, regulatory and policy implications of extending to an NHS-wide LHS to support diagnosis.Speakers

The problem

of diagnostic error, and why it is difficult to move to evidence-based diagnosis. Terminologies, classifications and types of evidence. What a LHS is and how it could solve the problem of collecting diagnostic evidence.Brendan Delaney, King's College London.Experimental evidence to support diagnostic decision support arising from a recent RCT of simulated cases, showing a 6% absolute improvement in diagnostic accuracy with ‘suggestions' based on the presenting problem. Impact on diagnostic accuracy of a full evaluation of the DSS in a controlled before-after study with GPs, using actors as simulated patients.Olga Kostopoulou , King's College London.A generalizable model of diagnostic knowledge. How to make evidence available in a computable format, and to curate and maintain high quality evidence.Derek Corrigan, Royal College of Surgeons of Ireland.Discussion including opportunity for hands-on with the decision support system (DSS) integrated with the InPS Vision3 EHR.Intended audience Highly multidisciplinary symposium bringing together clinicians, researchers, psychologists, statisticians and informatics experts interested in improving patient safety with respect to diagnosis.


  • Brendan Delaney