Oral Presentations

97 PICTURE-Risk – A method for patient risk scoring using Electronic Health Record data

Abstract

Background Risk scores are widely used tools that help clinicians assess risk for a patient. Benefits of such tools range from disease prevention and management to more efficient clinical resource utilisation. Developing and using a risk score tool often involves many steps from data extraction to rigorous testing and validation of its clinical usefulness and safety.

Methods We have worked with clinicians to create a tool that simplifies the development, validation and deployment of risk scores and supports their use in clinical settings. We present PICTURE-Risk, a generalisable component of PICTURE, a clinical informatics platform developed at Great Ormond Street Hospital (GOSH).

As a PICTURE component, PICTURE-Risk benefits from the standard electronic health record (EHR) extraction processes developed by the GOSH Digital Research Environment (DRE) and PICTURE’s cohort builder and analytics functionality. A new risk score can be configured by specifying a set of items (risk factors) and their contribution to the total score. PICTURE-Risk will then calculate and present on demand the total risk for a given patient or cohort in an interactive user interface.

Results We have developed proof-of-concept (POC) risk scores for diabetes and sudden cardiac death in collaboration with clinicians. The total risk, a breakdown of the risk factors’ contribution and various analyses of the cohort risk scores were generated on demand for a patient or cohort using EHR data. The POC version has been developed in R and Shiny and most of the user interface was built using TypeScript and React.

Conclusion The POC application of PICTURE-Risk has demonstrated that it is feasible to make a wide range of risk scores available for research and clinical use with a high-level specification of the risk score being the only required input. We envisage this component will accelerate the development and use of risk scores at GOSH.

Acknowledgements for funding or support This work is supported by the NIHR GOSH BRC. The views expressed are those of the author(s) and not necessarily those of the NHS, the NIHR or the Department of Health. This work is supported by the Great Ormond Street Hospital Children’s Charity.

Article metrics
Altmetric data not available for this article.
Dimensionsopen-url