Back

Integrating respiratory infection surveillance and temperature data improves all-season mortality reconstruction

Wang, J.; an der Heiden, M.; Irrgang, C.

2026-08-17 public and global health
10.64898/2026.08.15.26359980 medRxiv
Show abstract

All-season mortality surveillance can inform public-health planning under climate change, but attribution is complicated by overlapping effects of temperature, respiratory infections and other non-environmental factors. Here we develop a component-based neural-network model for daily all-cause mortality that combines high-resolution temperature and dew-point data, severe acute respiratory infection (SARI) hospitalization incidence, demographic structure and an adaptive mortality baseline. In Germany, inputs included non-COVID SARI and COVID-19-associated SARI. Using 2014-2025 mortality data, the hybrid model reduced daily root-mean-square error to 80.2 deaths, compared with 170.2 for a weather-only model and 117.1 for an infection-only model, and enabled age- and sex-resolved examination of fitted components. Including respiratory-infection indicators reduced the increase in risk assigned by the model to cold exposure, suggesting that temperature-only models may partly assign winter infection-related variation to cold. Component magnitudes represent conditional model decompositions rather than causal estimates of deaths due to SARI or temperature.

Matching journals

The top 3 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.