Predicting an Individual's Physiologic State without a Crystal Ball
Abstract
This talk illustrates two approaches by which biomathematical models can be developed to construct individualized, i.e. subject specific, physiologic predictive algorithms. We describe, in layman's terms, the pros and cons of first-principles physiology-based algorithms and data-driven, autoregressive algorithms. We also discuss how these algorithms may be customized to predict the physiologic state of specific individuals some time into the future. We illustrate the predictive power of these approaches in the prediction of: (i) performance impairment due to total sleep, (ii) body core temperature during physical activity, and (iii) glucose levels of type 1 and type 2 diabetes patients. In conjunction with real-time physiologic monitoring devices, such predictive algorithms may be used to optimize the timing and dosing of fatigue countermeasures, e.g. naps and caffeine, so that performance peaks and is maintained during desired times of day, minimize the occurrence of heat-related injuries, such as heat strokes, and allow for proactive glucose regulatory interventions be for glycemic levels drift from the desired range. [TATRC website, 15 Dec 2008]
Document Details
- Document Type
- Technical Report
- Publication Date
- Apr 05, 2008
- Accession Number
- ADA490361
Entities
People
- Jaques Reifman
Organizations
- United States Army Medical Research and Development Command