Monitoring Resilience in Bursts

Abstract

AbstractThe possibility to anticipate critical transitions through detecting loss of resilience has attracted attention in a variety of fields. Resilience indicators rely on the mathematical concept of critical slowing down, which means that a system recovers increasingly slowly from external perturbations when approaching a tipping point. This decrease in recovery rate can be reflected in rising autocorrelation and variance in data. To test whether resilience is changing, resilience indicators are often calculated using a moving window in long, continuous time series of the system. However, for some systems it may be more feasible to collect several high-resolution time series in short periods of time, i.e. in bursts. Resilience indicators can then be calculated to detect a change of resilience in a system between such bursts. Here, we compare the performance of both methods using simulated data, and showcase possible use of bursts in a case-study using mood data to anticipate depression in a patient. Using the same number of data points, the burst approach outperformed the moving window method, suggesting that it is possible to down-sample the continuous time series and still signal of an upcoming transition. We suggest guidelines to design an optimal sampling strategy. Our results imply that using bursts of data instead of continuous time series may improve the capacity to detect changes in systems’ resilience. This method is promising for a variety of fields, such as human health, epidemiology, or ecology, where continuous monitoring is costly or unfeasible.Significance statementGauging the risk of tipping points is of great relevance in complex systems ranging from health to climate, and ecosystems. For this purpose, dynamical indicators of resilience are being derived from long continuous time series to monitor the system and obtain early warning signals. However, gathering such data is often prohibitively expensive or practically unfeasible. Here we show that collecting data in brief, intense bursts may often solve the problem, making it possible to estimate change in resilience between the bursts withrelatively high precision. This may be particularly useful for monitoring resilience of humans or animals, where brief time series of blood pressure, balance, mood or other relevant markers may be collected relatively easily to help estimating systemic resilience.