Early warning signals of emerging infectious diseases

Early warning signals (EWS) provide a crucial opportunity to predict and mitigate emerging infectious disease outbreaks before they escalate ^1–3 . However, their practical application remains largely theoretical, often limited to single diseases and specific locations, with little systematic evaluation of their predictive accuracy across different contexts ¹ . Additionally, traditional EWS approaches fail to directly estimate outbreak timing or lead time-the period between detection of warning signals and outbreak onset-which is critical for effective intervention ¹ . To address these gaps, we integrated EWS with time-to-event analysis to assess the predictive performance of 19 resilience indicators (model-free statistical properties of time series, RIs) across 184 time series encompassing 390 outbreaks of 31 infectious diseases in 134 regions. Our findings demonstrate that both traditional RI and time-to-event analyses reliably warn of upcoming outbreaks and combining the most accurate RIs significantly improves outbreak prediction, yielding an average lead time of 18-21 days-a window sufficient for implementing disease control and prevention measures. Lead time and outbreak timing were influenced by pathogen type, climate, and socio-economic factors. Temperature and precipitation exhibited unimodal effects on outbreak risk and lead time for vector-borne and viral diseases, while pathogens with longer incubation times and regions with higher human development index (an estimate of development and per capita income) experienced longer lead times and lower outbreak risk, especially for vector-borne diseases. These findings underscore the importance of integrating socio-environmental factors into outbreak prediction models and suggest that early initiation of monitoring and case reporting and an enhanced commitment to outbreak surveillance might extend lead times before an outbreak. Our study advances the field of epidemic forecasting by providing a robust framework for applying RI-based EWSs to diverse pathogens and geographies, laying the foundation for a global disease early warning system that can guide proactive public health interventions.