Real-time Prediction of Next Generation Sequencing Results: Application in Clinical Settings Using FHIR API

Details

This report details a methodology published in npj Digital Medicine for predicting Next Generation Sequencing (NGS) results in real time. The Heme-STAMP project aims to derive clinical insights from NGS data by employing machine learning, specifically integrating this process with Electronic Health Record (EHR) systems. By utilizing the FHIR API to acquire and leverage EHR data, it becomes possible to assist in disease diagnosis and prognosis prediction. The model showed potential for improving diagnostic accuracy for specific diseases, achieving an AUC value of 0.77 (compared to 0.78 in control groups). Furthermore, the study reported that integrating and analyzing both EHR and NGS data provides higher predictive performance than single-source analysis. Technically, the system is built using platforms like Epic API and Azure, with FHIR standard-based API integration being crucial. This enables real-time utilization of data in clinical settings and application to Clinical Decision Support Systems (CDSS).