Healthcare occupancy data as an early warning indicator of infectious disease surges in São Paulo

Abstract

Early detection of infectious disease surges is critical for effective public health response, yet traditional surveillance systems often suffer from delayed reporting. Here, we investigate whether passively collected occupancy data from healthcare facilities can act as an early warning indicator of epidemic activity. We analyzed occupancy trends from 17 public urgent care units in the São Paulo metropolitan area, Brazil, and compared them with laboratory-confirmed surveillance data for infectious diseases, including SARS-CoV-2 and dengue virus, using dynamic time warping and Granger causality testing, a z-based epidemic volatility index, and spatial association analysis. We show significant temporal associations between occupancy patterns and pathogen-specific indicators across three epidemic waves, with statistical lags of 1 to 5 weeks. The occupancy-based alert system detects all three waves from their onset, preceding laboratory-confirmed surge peaks. Spatial analysis identifies persistent high-occupancy hotspots in later surge stages, highlighting units where sustained healthcare monitoring remains necessary. These findings highlight that passively collected occupancy data can anticipate laboratory-confirmed infectious disease surges by up to five weeks, offering a complementary and privacy-safe layer to existing epidemic surveillance approaches. Infectious disease surges place enormous pressure on healthcare systems, yet official surveillance often takes days or weeks to confirm one is underway. This study investigated whether occupancy data from a widely used mapping application could serve as an early warning signal. We analyzed occupancy levels at 17 public urgent care units across the São Paulo metropolitan area, Brazil, and compared them with laboratory results for SARS-CoV-2, respiratory viruses, and dengue. Because people seek care before receiving a formal diagnosis, occupancy trends tend to rise before laboratory case counts do. We found that occupancy increases consistently preceded confirmed surges by up to five weeks across three epidemic waves. A simple alert system detected all three waves from their earliest stages. Passively collected occupancy data could complement existing surveillance tools and support faster public health responses. Araújo, Silva, et al. analyze passively collected occupancy data from public urgent care units in São Paulo to assess its potential as an early warning signal for infectious disease surges. Occupancy increases consistently preceded laboratory-confirmed surge peaks by up to five weeks across three epidemic waves and multiple pathogens.