Smartwatch-based algorithm for early detection of pulmonary infection: Validation and performance evaluation

Area of Science:

• Biomedical Engineering
• Digital Health
• Infectious Disease Monitoring

Background:

• Smart devices offer potential for early detection of pulmonary infections (PI).
• This study focuses on validating a smartwatch-based algorithm for PI risk monitoring in adults.

Purpose of the Study:

• To validate a smartwatch algorithm for early detection of pulmonary infections (PI).
• To assess the algorithm's accuracy, sensitivity, and specificity in identifying PI risk.

Main Methods:

• An algorithm was developed and tested on smartwatches, analyzing heart rate variability, respiratory rate, oxygen saturation, body temperature, and cough sound.
• The algorithm was embedded in the Respiratory Health Study app and validated against clinical diagnosis (gold standard).

Main Results:

• The algorithm achieved an area under the curve of 0.86 for predicting PI (P < 0.001).
• Overall accuracy was 85.9%, with 81.4% sensitivity and 90.4% specificity.
• Incorporating cough sound improved accuracy to 82.6% compared to other vital signs alone (68.2%).

Conclusions:

• A wearable system using smartwatches can facilitate early detection of pulmonary infection risk.
• Multi-source features significantly enhance the performance of lung infection screening algorithms.