Feasibility of Using Smartwatches and Smartphones to Remotely Monitor Patients with COPD
Andrea Gershon

The use of remote sensing to detect early exacerbations of Chronic Obstructive Pulmonary Disease (COPD) coupled with early  intervention has the potential to transform COPD care from being reactive to proactive. This will reduce the need for hospital admissions and emergency department visits. The use of low cost hardware (which patients might already have) and software means that it can be easily adopted by other centres to improve patient outcomes.  The use of a passive smartphone applicaton provides a unique approach to health care delivery.  Evaluation of the application is in early stages so patient outcome data is not yet available. Nonetheless, we have found that patients with COPD are willing and able to use smartwatches, and these mobile devices can gather important health information such as heart rate, physical activity and frequency of coughing.

Background

Acute exacerbations of COPD cause significant mortality and decrease patients’ quality of life. These outcomes may be avoided with early detection and treatment.  We aimed to develop a smartwatch and smartphone application that passively detects early exacerbations of COPD. The current study was conducted to determine if it was feasible to obtain consistent, useful physiologic data from a smartwatch application worn by COPD patients who are often older and can be quite ill.

Methods

Hospitalized patients with acute exacerbations of COPD and COPD clinic patients were recruited from University Health Network and Sunnybrook Health Sciences Centre. Patients were asked to wear an Android smartwatch which passively monitored and recorded physiologic signs including heart rate, step count, raw acceleration measurements as an indication of physical activity, and audio recordings to detect cough, breathlessness and wheezing.They were also asked to complete a daily symptom questionnaire on a smartphone.

Results

Fourteen patients with COPD have been enrolled in the study for a combined total of 477 days. The smartwatch was worn by patients as directed and both the smartwatch and smartphone were charged appropriately. Over 530 hours of audio recording, 815,000 heart rate samples and 8.6 million accelerometer samples were obtained. Symptom questionnaires were completed daily.

Conclusion

Using a smartphone and smartwatch to record physiologic and audio data in COPD patients appears feasible. Future work includes the analysis of sounds for coughing, wheezing and breathlessness. Machine learning algorithms to automatically detect the sounds will then be added to the application. We will also correlate sensing data (heart rate, activity, coughing) to daily symptoms to determine the accuracy in which early acute exacerbations of COPD can be detected.