Information delivery in tetherless healthcare

Tetherless care was proposed to help address the costly burden of chronic conditions and diseases like diabetes, hypertension, and heart disease. In support of this vision, this work presents a solution for the intelligent delivery of realtime messages given intermittent connectivity and limited energy. It employs a O(1) Markov predictor operating over a history of network sessions to predict likely future low power opportunities to transfer data while attending to realtime delivery deadlines. The algorithm was deployed to the smartphones of several volunteers for two months and was tasked with managing the transfer of test data and statistics of its operation. Results show: • Predictions of the duration or start time of a given session have 80% accuracy to within six minutes. • Where delays between successiveWiFi sessions are less than nine minutes with 81% probability, the system is capable of supporting deadlines on the order of minutes with WiFi-based sessions only, falling back on more costly cellular technology to cover the final 19% of delays. • With a fixed 24 hour deadline for all messages, the solution can often introduce further delay to conserve energy, waiting for the advent of some future session before initiating transmission.