AI-enabled occult bleeding detection and monitoring

Project Summary/Abstract Internal hemorrhage is the second leading cause of death following trauma, with missed abdominal bleeding contributing to significant late-stage mortality and morbidity. Current diagnostic approaches, such as Focused Assessment with Sonography for Trauma (FAST) and computed tomography (CT), require trained personnel, are not portable, and provide only intermittent assessments. These limitations are particularly acute in field and pre-hospital environments, where rapid diagnosis and triage are critical, but expertise and imaging capabilities are limited or unavailable. This STTR Phase I project proposes to develop and evaluate a novel point-of-care diagnostic device based on electrical impedance tomography (EIT) that enables automated detection and monitoring of internal bleeding. The proposed system uses a compact array of electrodes and machine learning-enhanced imaging algorithms to provide real-time information on bleeding volume and rate. It is designed for immediate use by non-expert personnel and aims to support timely triage and transport decisions in both civilian and military trauma settings. Specific Aim 1: Optimize machine learning algorithms for accurate bleeding detection using reduced- sensor EIT arrays. We will train deep neural networks to mitigate movement and breathing artifacts and validate performance using a cadaver model capable of simulating realistic bleeding and respiratory motion. Specific Aim 2: Develop and test a candidate housing for the prototype device in volunteer firefighter subjects. We will evaluate sensor application, ease of use, and device performance in fluid monitoring scenarios, alongside feedback from users on form factor and usability. This project leverages the academic team’s extensive background in electrical imaging and the small business partner’s expertise in medical device development. Success in Phase I will establish the technical foundation and user requirements needed to advance to a Phase II effort focused on ruggedization, expanded clinical testing, and regulatory preparation. The long-term vision is to produce a lightweight, self-interpreting hemorrhage detection system that can be integrated into trauma workflows across EMS, battlefield, and rural settings—ultimately improving survival outcomes through earlier, more accurate identification of internal bleeding.