Federal Project Grant Award Summary

The National Heart, Lung, and Blood Institute (NHLBI) awarded the University of Pittsburgh a $238,500 Project Grant effective May 1, 2026 through April 30, 2028, under the Cardiovascular Diseases Research program (CFDA 93.837). This award supports research to develop and validate natural language processing (NLP) and machine learning (ML) algorithms that enhance prehospital triage decision-making by analyzing free-text clinical observations. The project addresses critical gaps in emergency medical services (EMS) triage accuracy, where current misclassification rates of illness severity reach 11-72% underestimation and 10-48% overestimation, resulting in thousands of excess deaths annually and hundreds of millions in preventable healthcare costs. The research will leverage lexical data from prehospital clinician impressions and 9-1-1 telecommunicator notes to develop predictive models for illness severity indicators such as injury severity scores, patient mortality, and hospital length of stay, as well as identification of patients requiring prehospital life-saving interventions.

The project represents the first comprehensive assessment of whether and how NLP and ML methodologies can systematically extract diagnostic and predictive information from unstructured prehospital narrative data to support algorithmic decision support tools for the approximately 1 million EMS clinicians across 23,000 agencies who annually assess the 25 million U.S. patients transported to emergency departments. Research deliverables will include validated prediction models and algorithmic tools designed to synthesize diagnostic information and provider intuition to improve triage accuracy and patient outcomes in the prehospital care setting.