Elderly patients with multiple lingering medical conditions habitually take several pills every day. How should effective real-time feedback sensor display systems be designed for patients who take many drugs at specific times each day? In an effort to address absentmindedness, negligence, and/or the fear of troublesome side effects, Lee and Dey built a sensor-augmented pillbox for monitoring medicine use behaviors of patients. The seven-day pillbox consists of an accelerometer to trace when it is picked up, spring switches to discover open pillbox doors, and a conventional circuitry with microcontrollers and a wireless modem. The pillbox sensor uses a wireless network to transmit data captured from each patient to a remote server. The server processes the patient behavioral data to ascertain and assess the incidents of medicine consumption, and to provide visual feedback display on how well individual patients take medications, use the phone, and prepare coffee.

A focus group experiment was performed to investigate the effective use of a feedback display in helping patients accurately take medicine. Twelve elderly patients with multiple protracted illnesses such as hypertension and diabetes were randomly assigned to feedback and control groups. The feedback group patients viewed the performance of their medication-taking behaviors in real time from a tablet display, whereas the control group patients only received a hard copy performance report for one month.

The authors evaluated the impact of a real-time feedback display on medication-taking behaviors. Adherence is the percentage of all pills taken in a time period. Correctness is the rate of accurate pills taken each day. Promptness is the percentage of pills taken on time. Time of day variance “measures how the time of day that medications were taken varied from one day to another.” Self-efficacy is the extent to which patients feel confident about surmounting the barriers to taking medications. The feedback display significantly enriched adherence, promptness, correctness, and the variance in the time of day; it had no significant effect on self-efficacy.

Although the small number of patients studied limits the generalizability of the experimental results, the authors have developed a valuable tool for monitoring the drug intake behaviors of patients with multiple chronic health problems.