Fighter pilots face unique occupational stressors, including extreme gravitational forces, long and stressful missions, and rigorous physical and mental training. The physical demands can take their toll on the pilot’s body, as evidenced by the numerous health issues such as neck and back problems that have plagued the fighter pilot population for years. To help address this problem, several groups within the United States Air Force (USAF) are implementing wellness and human performance optimization programs. Most approaches are centered around increasing physical therapy and wellness efforts. While implementing these programs will undoubtedly make a positive impact towards the goal of rapid recovery and full operating capacity, there is an opportunity to create and implement artificial intelligence (AI) algorithms that can collect, analyze, store and present objective pain data. This would deliver critical information to provide better insights and enhance fighter pilots' rehabilitation decisions while also providing a unique and personalized approach for each fighter pilot.
An unobtrusive physiological pain classifier was created for use by the USAF fighter pilot community by first collecting physiological measures (electrocardiogram [ECG]) from healthy, adult (N=41) participants during baseline and pain-induction tasks. These raw ECG signals were used to derive a series of cardiovascular features including time domain, frequency domain, and non-linear features. Using logistic regression, these features classified pain at an accuracy level of 79.6%. Field data collection is currently underway with the 56th Training Squadron at Luke Air Force Base (AFB) to determine classification accuracy and ruggedness in operational environments. For this effort, the classifier was integrated with a smartwatch and mobile application for classification in on- and off-duty environments. This data will ultimately assist the pilots and their medical staff in building a more robust, individualized physical therapy program.