Dynamical Indicators of Resilience from Physiological Time Series in Geriatric Inpatients: Lessons Learned

Background The concept of physical resilience may help geriatric medicine objectively assess patients’ ability to ‘bounce back’ from future health challenges. Indicators hypothesized to forecast resilience after a stressor have been developed under two paradigms with different perspectives: Critical Slowing Down (CSD) and Loss of Complexity (LoC). This study explored if and how these indicators, based on fluctuations in physiologic signals, can validly reflect the physical resilience of geriatric inpatients. Methods Geriatric patients (n = 121, 60% female) had their heart rate and physical activity continuously monitored using a chest-worn sensor. Measures of health functioning (multimorbidity, frailty and Activities of Daily Living [ADL]) were obtained by questionnaire at admission. Indicators from both paradigms (CSD: variance, autocorrelation, cross-correlation; LoC: [multivariate] multiscale entropy) were extracted from both physiological signals. The relationships among indicators and their associations with health functioning were assessed by correlation and linear regression analyses, respectively. Results Greater complexity and higher variance in physical activity were associated with lower frailty (β = –0.28, p=.004 and β = –0.37, p<.001, respectively) and better ADL function (β = 0.23, p=.022 and β = 0.38, p<.001). The associations of physical activity variance with health functioning was not in the expected direction based on the Critical Slowing Down paradigm. Conclusions Associations between dynamical resilience indicators tested here and measures of health functioning were not all in the expected direction. In retrospect, these observations stress the importance of matching the underlying assumptions of the resilience paradigm to the homeostatic role of the variable monitored.