Smartwatch-derived versus self-reported outcomes of physiological recovery after COVID-19, influenza, and group A streptococcus: a 2-year prospective cohort study.

backgroundAccurate detection of recovery from communicable diseases enables timely care and helps prevent complications and chronic conditions. We aimed to investigate the time for self-reported symptom recovery and digital recovery based on physiological data from smartwatches after infection with COVID-19, influenza, and group A streptococcus (GAS), and how activity levels changed throughout recovery.

methodsWe analysed data on COVID-19, influenza, and GAS from a 2-year prospective cohort study in Israel, incorporating smartwatch data, self-reported symptoms, and medical records. Eligible individuals were aged at least 18 years, Maccabi Healthcare Services members for at least 2 years, using their own smartphone, and could provide informed consent. Participants were recruited through social media and word-of-mouth. Controls were matched by age and sex, with post-hoc validation confirming similar baseline health profiles. At enrolment, participants completed a one-time questionnaire, received a smartwatch, and downloaded two applications to complete daily questionnaires and self-reports; passive data (eg, daily steps, distance walked, active time, active calories) and physiological measures (eg, heart rate and heart rate variability-based stress) were also collected from the smartwatches. Positive diagnoses for COVID-19, influenza, or GAS were identified from electronic medical records or self-reported through the app after home testing. The primary outcome measure was the duration of lag between self-reported symptom resolution and digital recovery, defined as the return of smartwatch-detected physiology (heart rate and heart rate variability-based stress during sedentary periods) to baseline levels. Digital recovery was assessed in all participants who tested positive for the disease, and valid smartwatch and questionnaire data were available. We examined this lag across illnesses and by severity (mild or moderate-to-severe). We also analysed behavioural and activity measures, such as daily steps, active calories, active time, and total distance, from smartwatch data to contextualise recovery trajectories.

findingsDuring the study period Nov 16, 2020, to May 11, 2023, involving 4795 participants, 3097 COVID-19 cases, 633 influenza cases, and 380 GAS cases occurred. 2742 participants had COVID-19 at least once during the 2-year follow up, for which 1421 (51·8%) were female and 1321 (48·2%) were male, with a median age of 44·0 years (IQR 33·0-56·0). Likewise, of 531 participants who had influenza at least once, 305 (57·4%) were female and 226 (42·6%) were male, with a median age of 51·0 years (IQR 38·0-61·0). For 334 participants who had GAS at least once, 191 (57·2%) were female and 143 (42·8%) were male, with a median age of 38·0 years (IQR 32·0-47·0). Digital recovery (measured by smartwatches) lagged substantially behind self-reported symptom resolution in most cases and both varied by illness severity and type: COVID-19 mild cases showed 8·53 days (95% CI 7·89 to 9·17) to self-reported symptom resolution, with an additional 7·14 days (6·49 to 7·78) to digital recovery, and moderate-to-severe cases showed 12·05 days (11·38 to 12·72) to symptom resolution with an additional 60·23 days (59·58 to 60·89) to digital recovery; influenza mild cases showed 7·98 days (6·66 to 9·29) to symptom resolution with an additional 2·51 days (1·21 to 3·81) to digital recovery, and moderate-to-severe cases showed 12·06 days (10·81 to 13·31) with an additional 7·85 days (6·62 to 9·08) to digital recovery; and GAS mild cases showed 7·95 days (6·45 to 9·45) to symptom resolution with an additional -1·12 days (-2·59 to 0·34) to digital recovery and moderate-to-severe cases showed 9·75 days (8·32 to 11·19) with an additional 4·37 days (2·97 to 5·76) to digital recovery. Markedly, when participants reported being symptom-free, their daily steps, distance walked, active time, and active calories returned to baseline levels, indicating a restoration of their regular daily routines.

interpretationPublic health recommendations suggest individuals can resume normal activities 5 days after symptom cessation. However, our study finds that full recovery can take longer, indicating that delayed return to normal activities might be warranted for those who had fever or other severe symptoms. Smartwatches can help identify patients who have not fully recovered from infection. Future research should determine whether identifying these people, and modifying their behaviour appropriately, leads to improved outcomes and can reduce the burden of communicable diseases.

fundingEuropean Research Council, the Israel Science Foundation within the Israel Precision Medicine Partnership programme, and a Koret Foundation gift for Smart Cities and Digital Living.