Robust language-based mental health assessments in time and space through social media

• DOI: 10.1038/s41746-024-01100-0
• Publication Date: 2024-05-02T19:01:56Z
• Authors: Siddharth Mangalik; Johannes C. Eichs...; Salvatore Giorgi; Jihu Mun; Farhan Ahmed; Gilvir Gill; Adithya V. Ganesan; Shashanka Subrahm...; Nikita Soni; Sean A. P. Clouston; H. Andrew Schwartz
• Keywords: FOS: Computer and information sciences; 0303 health sciences; Computer Science - Computation and Language; J.4; I.2.7; Computer applications to medicine. Medical informatics; R858-859.7; Article; J.4; I.2.7; 3. Good health; 03 medical and health sciences; 0302 clinical medicine; Computation and Language (cs.CL)

Abstract In the most comprehensive population surveys, mental health is only broadly captured through questionnaires asking about “mentally unhealthy days” or feelings of “sadness.” Further, estimates are predominantly consolidated to yearly at state level, which considerably coarser than best physical health. Through large-scale analysis social media, robust estimation feasible finer resolutions. this study, we created a pipeline that used ~1 billion Tweets from 2 million geo-located users estimate levels and changes for depression anxiety, two leading conditions. Language-based assessments (LBMHAs) had substantially higher reliability across space time available survey measures. This work presents reliable anxiety down county-weeks level. Where surveys were available, found moderate strong associations between LBMHAs scores multiple granularity, national level weekly county measurements (fixed effects β = 0.34 1.82 ; p < 0.001). demonstrated temporal validity, showing clear absolute increases after list major societal events (+23% change assessments). showed improved external evidenced by stronger correlations with measures socioeconomic status surveys. study shows careful aggregation media data yields spatiotemporal exceed granularity achievable existing does so generally greater validity.