Using Deep Learning and Natural Language Processing to Detect and Optimize 911 Stroke Cases
Lehansa Marambage
Jose Marti STEM Academy, Union City, United States
Publication date: November 20, 2025
Jose Marti STEM Academy, Union City, United States
Publication date: November 20, 2025
DOI: http://doi.org/10.34614/JIYRC2025II60
ABSTRACT
Stroke, the fifth leading cause of death and a major cause of disability in the US, is a time-critical emergency handled by EMS agencies—20% of which manage 80% of service calls (Krohmer & Elkins, 2020). Rising unanswered 911 calls and outdated CAD systems further hinder response efficiency. This study aims to reduce stroke-related mortality and rehabilitation costs from 911 CAD delays by enabling faster, more accurate stroke predictions during 911 calls using deep-learning and natural language processing (NLP). A Python mel-spectrogram-based audio machine learning model, built with TensorFlow and YAMNet transfer learning, achieved 61.1% accuracy in categorizing calls into fire, police, or EMS departments during the study's initial phase. An NLP program records and provides summarized 911 call reports for first responders. Ongoing efforts aim to enhance EMS-related medical-prediction models. After further validation with larger datasets, integration of this system into 911 CAD could improve response times and potentially save lives.
Stroke, the fifth leading cause of death and a major cause of disability in the US, is a time-critical emergency handled by EMS agencies—20% of which manage 80% of service calls (Krohmer & Elkins, 2020). Rising unanswered 911 calls and outdated CAD systems further hinder response efficiency. This study aims to reduce stroke-related mortality and rehabilitation costs from 911 CAD delays by enabling faster, more accurate stroke predictions during 911 calls using deep-learning and natural language processing (NLP). A Python mel-spectrogram-based audio machine learning model, built with TensorFlow and YAMNet transfer learning, achieved 61.1% accuracy in categorizing calls into fire, police, or EMS departments during the study's initial phase. An NLP program records and provides summarized 911 call reports for first responders. Ongoing efforts aim to enhance EMS-related medical-prediction models. After further validation with larger datasets, integration of this system into 911 CAD could improve response times and potentially save lives.
