The Feinstein Institutes for Medical Research

Timothy Martens

A method and a system include at least one processor of a plurality of computing devices that execute a plurality of heart disease triage (HDT) microservices operating on a scalable data container orchestration platform. The platform is configured to continuously receive a plurality of patient-specific data files that include medical data and provider visit data for each patient. Patient-specific diagnostic text of the medical history, the provider visit, and echocardiogram data in the patient-specific data files of patients identified as having echocardiogram data are tokenized using a feature-space transformation to generate a patient-specific tokenized diagnostic feature set. The patient-specific tokenized diagnostic feature set for each identified patient is inputted to an HD severity determination machine learning model to generate a heart disease (HD) severity score. A triage scoring algorithm based on the HD severity score and provider visit data are used to generate a triage score for each identified patient.

Growing community of inventors

Manhasset, NY, United States of America

Timothy Martens