Monitoring Soldiers' Vital Signs
Probabilistic Unit Life Status Estimation (PULSE)
The Situation
The growing digitization of the battlefield is on the threshold of producing a quantum leap in the quality of medical care afforded to thousands of soldiers in the field. This important advance is the objective of the U.S. Army's Warfighter Physiological Status Monitoring (WPSM) system, which aims to provide real-time health monitoring of warfighters, using an array of personal biomedical sensors worn by each soldier. The array will provide instant notification of medical events or trauma to battalion medical staff, enabling remotely performed casualty detection, diagnosis, and triage, hence delivering critical medical care to warfighters much more rapidly than is now possible with current equipment and methods. To be effective, the WPSM system requires sophisticated sensor processing algorithms for robust onboard sensor data fusion and automated clinical estimates, to produce timely analyses and alerts, and to reduce the burden of routine diagnosis by medical staff.
The Charles River Analytics Solution
Charles River Analytics developed the Probabilistic Unit Life Status Estimation (PULSE) suite of algorithms for clinical estimates. The PULSE algorithms were based on Temporal Bayesian Belief Networks (TBNs), an extension to standard Bayesian Belief Networks (BBNs) that add temporal reasoning to the probabilistic reasoning of BBNs. We used TBNs primarily to model the subject's clinical state, but also to model the sensor array hardware in parallel, obtaining metrics of sensor reliability which improved the clinical state estimates. A sample application of TBNs to monitoring of heart rates is shown below.
The PULSE system was designed for flexibility and extensibility; the overall PULSE architecture is illustrated below, as provided for use with a sensor simulation platform.
The Benefit
The PULSE system was delivered to the U.S. Army for further evaluation and possible inclusion (in whole or part) within the WPSM system. It forms the basis for a system that will enable medical staff to not only to identify casualties as they happen, but also prioritize their responses and better deliver needed trauma care within the "golden hour" after a serious wound or other battlefield trauma.
Links:
U.S. Army Medical Research and Materiel Command, Combat Casualty Care Research Program (http://www.usaccc.org)
Related Papers:
Das, S. K., Lawless, D., Hanson, M., & Zacharias, G. (2001). "Model-based Reasoning with Temporal Belief Networks", NASA Contract No. NAS2-01048, Final Report. Cambridge, MA, Charles River Analytics
Das, S.K., Introne, J., & Lawless, D. (2003). "Probabilistic Unit Life Status Estimation (PULSE)", Government Contract No. DAMD17-03-C-0060, Final Report. Cambridge, MA, Charles River Analytics
Das, S.K., et al. "Probabilistic Unit Life Status Estimation (PULSE)". Proceedings of the 7th International Conference on Information Fusion, Stockholm, Sweden. July (2004).
