Climate-related weather events and social media-derived anxiety and depression signals in Florida: A spatiotemporal analysis - 17/11/26
, Melanie Stearns a, Md Tareq Ferdous Khan b, Xiaoli Zhang a, Fazlay Faruque c, Ukamaka Oruche a, ⁎ 
Abstract |
Introduction |
Weather events may affect population mental health, but temporal and geographic patterns in related discourse remain unclear. We examined associations between National Oceanic and Atmospheric Administration (NOAA)-documented weather-events and anxiety- and depression-related social media signals across Florida.
Methods |
We conducted a spatiotemporal ecological study of geolocated social media posts (June 2022-September 2025) linked to NOAA-documented weather-events by county and date. Anxiety- and depression-related signals were identified using keyword-based classification. Multivariable logistic regression estimated overall and event-specific associations with 7-day weather exposure. Temporal, spatial, subgroup, sensitivity, and independent validation analyses further assessed robustness of the findings.
Results |
Among 39,770 posts (mean age, 33.1 ± 8.7 years; 61.6% male), 13.4% contained anxiety-related signals, 12.8% depression-related signals, and 9.5% had 7-day NOAA-documented weather-events exposure, which was associated with 25% higher odds of anxiety (AOR:1.25; 95%CI:1.04–1.52) but not depression-related signals (AOR:1.17; 95%CI:0.97–1.42). Tornado exposure showed the strongest event-specific anxiety association (AOR:1.65; 95%CI:1.19–2.28), with associations strongest on the event day for anxiety (AOR:1.59; 95%CI:1.18–2.14) and 1–3 days post-event for depression (AOR:1.49; 95%CI:1.13–1.96); no other event-specific association remained significant for depression. Spatial autocorrelation was significant for anxiety (Moran’s I =.338; P =.041), but not depression (Moran’s I =.280; P =.052), while classification across five validation samples showed Cohen’s κ of .749 and .910, respectively. Sensitivity, subgroup, and secondary analyses were consistent with the primary findings.
Conclusions |
Weather events were associated with short-term and geographic variation in mental health-related social media signals, with more consistent associations for anxiety than depression. Linking NOAA-documented weather-events with geolocated social media signals may complement surveillance of temporal and geographic patterns.
Le texte complet de cet article est disponible en PDF.Keywords : Weather events, Mental health signals, Text-mining, Temporal trend analysis, Geospatial modeling, Social media analytics
Plan
Vol 32
Article 100775- novembre 2026 Retour au numéroBienvenue sur EM-consulte, la référence des professionnels de santé.
L’accès au texte intégral de cet article nécessite un abonnement.
Déjà abonné à cette revue ?
