What is it about?
We describe a generalizable AI-based algorithm for calculating disease risk using data from the types of digital health record systems that support healthcare institution in real-world settings. We also show our our algorithm functions when applied to single hospital's electronic health record (EHR) system.
Featured Image
Photo by Carlos Muza on Unsplash
Why is it important?
Traditional disease risk calculators are developed using data from a specific region, and therefore often do not generalize well to patients from disparate populations. Additionally, traditional calculators use static datasets (i.e. those that are not updated over time), resulting in risk estimates that cannot adapt to up-to-date health trends. Finally, traditional calculators solely rely on well-established factors when estimating risk, thereby ignoring much of the other (potentially valuable) information contained within patients' records. Alternatively, our strategy is designed to comprehensively leverage ever-changing real-world datasets so that risk estimates always reflect all relevant contemporary information contained within the data system.
Perspectives
Disease risk estimation plays an increasingly important role in preventative care. With our algorithm, we seek to improve real-world risk estimation in a manner that optimizes predictions by maximumly leveraging all available patient data. Our approach is unique in that it is truly data-driven, basing risk estimates on empirical evidence contained within the data system. Moreover, our strategy is general and flexible, capable of enveloping as data systems become increasingly comprehensive in the in future. s digital health data systems grow in complexity, our algorithm can be readily adapted to included new types of data (e.g., genetic or other omics), potentially enhancing risk estimates further.
Terrence J. Lee-St. John
Cleveland Clinic Abu Dhabi
Read the Original
This page is a summary of: Towards artificial intelligence-based disease prediction algorithms that comprehensively leverage and continuously learn from real-world clinical tabular data systems, PLOS Digital Health, September 2024, PLOS,
DOI: 10.1371/journal.pdig.0000589.
You can read the full text:
Contributors
The following have contributed to this page
