Bayesian Networks-Based Interval Training Guidance System for Cancer Rehabilitation

The number of cancer patients who live more than 5 years after surgery exceeds 53.9% over the period of 1974 and 1990; Treatments for cancer patients are important during the recovery period, as physical pain and cancer fatigue affect cancer patients’ psychological and social functions. Researchers have shown that interval training improves the physical performance in terms of fatigue level, cardiovascular build-up, and hemoglobin concentration, the feelings of control, independence, self-esteem, and social relationship during cancer rehabilitation and chemotherapy periods. The lack of proper individual motivation levels and the difficulty in following given interval training protocols results in patients stopping interval training sessions before reaching proper exhaustion levels. In this work, we use behavioral cueing using music and performance feedback, combined with a social network interface, to provide motivation during interval training exercise sessions. We have developed an application program on the popular lightweight iPhone platform, embedded with several leveraged sensors. By measuring the exercise accuracy of the user through sensor readings, specifically accelerometers embedded in the iPhone, we are able to play suitable songs to match the user’s workout plan. A hybrid of a content-based, context-aware, and collaborative filtering methods using Bayesian networks incorporates the user’s music preferences and the exercise speed that will enhance performance. Additionally, exercise information such as the amount of calorie burned, exercise time, and the exercise accuracy, etc. are sent to the user’s social network group by analyzing contents of the web database and contact lists in the user’s iPhone. Keywords-component; exercise guidance system, interval training, music recommendation, social networks, rehabilitation