The following is a summary of “Wearable Sensor-Based Detection of Influenza in Presymptomatic and Asymptomatic Individuals,” published in the April 2023 issue of Infectious Diseases by Temple, et al.

Timely detection of viral infections is crucial for effective clinical management and public health interventions, especially during the COVID-19 pandemic. Wearable sensors and machine learning algorithms provided a promising approach for the early detection of infections.

In the study, 20 healthy adults were challenged with the influenza A (H3N2) virus and monitored using wearable electrocardiograms and physical activity sensors from 7 days before to 10 days after inoculation. A semisupervised multivariable anomaly detection model was trained on pre-inoculation data and used to classify the post-inoculation dataset.

The virus inoculation was well-tolerated, with an 85% infection rate. The algorithm identified 16 out of 17 (94%) positive presymptomatic and asymptomatic individuals, on average 58 hours after inoculation and 23 hours before symptom onset, with no alarms recorded during the 170 healthy days.

The study demonstrated the potential of wearable sensors and machine learning algorithms for the early detection of respiratory illnesses. The algorithm can be integrated with smartwatches using optical techniques, and further validation in large heterogeneous cohorts is needed to assess its effectiveness in normal living conditions.

Reference: academic.oup.com/jid/article/227/7/864/6618641