Early Warning Signals for transitions between health and disease

Funding: Royal Society Exchange Grant (with Vasilis Dakos, CNRS -Montpellier)

Following a perturbation, resilient systems return to their original equilibrium state. However, some perturbations can lead systems beyond a critical threshold (or ‘tipping point’) where it is no longer possible to return to their previous state. Such shifts between states are called critical transitions. A general property of systems approaching critical transitions is ‘critical slowing down’, which can be detected by statistical metrics such as increased variance, autocorrelation, and skewness in the system's response. Collectively, these metrics have been shown to act as early warning signals (EWS) for critical transitions in a variety of ecological, financial, and physiological systems. However, EWS have rarely been applied to understand the progression of infection within individuals, even though in many diseases, health deterioration is not gradual but usually happens abruptly. We employ the statistical methodology from complex systems to investigate individual-level critical transitions between health and disease (and death) in a fruit fly model of bacterial infection. Our aim is to identify statistical indicators of three stages of disease progression: 1. A normal state: absence of infection with high resilience. 2. A pre-disease state: host is infected but is still reversible to the normal state. 3. A disease state: the infection has progressed beyond a critical transition, and the organism is no longer able to return to its normal state. To achieve this we employ statistical modelling of high-resolution data of fruit fly locomotor activity during infection, to identify early warning signals for health deterioration. 

Key publications

Detecting Infection-Related Mortality Using Dynamical Statistical Indicators of High-Resolution Activity Time Series’.

Kutzer, Megan A. M., Shima Abdullateef, Alejandro V. Cano, Iris L. Soare-Nguyen, Katy M. Montieth, Javier Escudero, Vasilis Dakos, and Pedro F. Vale. ‘ bioRxiv, 25 April 2025. https://doi.org/10.1101/2025.04.21.649753.