2026-10-11T22:08:37Z https://api.figshare.com/v2/oai

oai:figshare.com:article/34048441 2026-10-01T17:33:30Z category_15 category_734 category_931 category_811 category_64 category_132 portal_5 item_type_1 month_year_10_2026

Correlation between search terms, infection cases, and media mentions. Sara Mesquita (17691590) João Loureiro (480403) Cláudio Haupt Vieira (25157779) Lília Perfeito (20553240) Joana Gonçalves-Sá (3173976) Neuroscience Biological Sciences not elsewhere classified Information Systems not elsewhere classified Mathematical Sciences not elsewhere classified Cancer Infectious Diseases xlink "> disease integrating digital epidemiology consistently outperforming models combines model evaluation casting increasingly relies apply simple supervised feature curation informed feature selection unsupervised methods unstable performance training periods traditional data results suggest online activity often noisy offer fast method offers inexpensive insights increased stability improve prediction hybrid approach frequently suffered different flu curated subset behavioral noise behavioral insights behavioral insight behavioral confounders adaptable framework 2009 h1n1 19 datasets a) and b) show the Pearson correlation between each cluster and infection cases (left) and media mentions (right) during the pandemic period for both the flu (top) and COVID-19 (bottom). In contrast, c) and d) present the Linear Regression model’s R2 performance. The asterisks represent the p-value of a t-test with the null hypothesis that mean correlation is zero (*0.01 < p < 0.05; **0.001 < p < 0.01; ***p < 0.001; ns p > 0.05). In a) and b), correlations can be both positive and negative, indicating whether search volumes increase or decrease with cases or media mentions. In c) and d), however, only the absolute correlation values are considered, as predictive power depends on correlation strength rather than direction. Also, c) and d) illustrate the cumulative contribution of individual search terms to the model’s performance. Terms are added sequentially from the most to the least correlated with infection cases (left to right, in Cases panels) and from the least to the most correlated with media mentions (left to right, in Media panels). The black line represents the Linear Regression R2, showing how predictive performance evolves as new terms are added, while the dashed line shows the R2 obtained for C1 as a reference. We focus on the Linear Regression model here as it provides a more interpretable baseline for assessing the contribution of different feature sets. The colors of the vertical bars correspond to the clusters each term belongs to, allowing a comparison of cluster contributions. 2026-10-01T17:32:53Z Image Figure 10.1371/journal.pone.0358063.g003 https://figshare.com/articles/figure/_p_Correlation_between_search_terms_infection_cases_and_media_mentions_p_/34048441 CC BY 4.0