Learning from pandemics: Feature selection and clustering for robust digital disease nowcasting
4 affiliations
- 1LIP – Laboratory for Instrumentation and Experimental Particle Physics, Avenida Prof. Gama Pinto 2, Lisboa, Portugal
- 2NOVA Medical School, Universidade NOVA de Lisboa, Campo dos Mártires da Pátria 130, Lisboa, Portugal
- 3NOVA School of Business and Economics, Universidade NOVA de Lisboa, Rua da Holanda, Carcavelos, Portugal
- 4NOVA LINCS, Faculdade de Ciências e Tecnologia, Universidade NOVA de Lisboa, Caparica, Portugal
Abstract
Disease now-casting increasingly relies on non-traditional data, such as online activity, which offer fast, inexpensive insights but are often noisy, unstructured, and influenced by behavioral confounders. Models using search engine data have frequently suffered from overfitting and unstable performance. However, pandemics may help distinguish true disease signals from behavioral noise, as both tend to surge during outbreaks. We propose a hybrid approach to improve prediction that combines model evaluation with behavioral insight. We apply simple supervised and unsupervised methods to different flu and COVID-19 datasets, using the 2009 H1N1 and 2020 COVID-19 pandemics as training periods. In both cases, selecting a curated subset of search terms improved model performance, reduced overfitting, and increased stability, consistently outperforming models that used search terms indiscriminately. These results suggest that feature curation informed by behavioral insights can enhance now-casting beyond purely statistical approaches. Our method offers a scalable and adaptable framework for integrating digital epidemiology into public health surveillance.
Related
- Spectrally Tuned Bandwidth Selection for Kernel Fuzzy Relational Clustering
- Linear-Time Multivariate Unsupervised Feature Selection Using Nyström-Based Entropy Approximation
- Universal Feature Selection with Noisy Observations and Weak Symmetry Conditions
- Variable Selection for Feature-Based Newsvendor
- Robust LassoNet: Enhancing Feature Selection in Neural Networks via Robust Loss Functions
BibTeX
@article{mesquita2026learning,
title = {{Learning from pandemics: Feature selection and clustering for robust digital disease nowcasting}},
author = {Sara Mesquita and João Loureiro and Cláudio Haupt Vieira and Lília Perfeito and Joana Gonçalves-Sá},
journal = {PLOS One},
year = {2026},
month = oct,
doi = {10.1371/journal.pone.0358063},
url = {https://doi.org/10.1371/journal.pone.0358063},
note = {CC BY 4.0},
}
RIS
TY - JOUR
TI - Learning from pandemics: Feature selection and clustering for robust digital disease nowcasting
AU - Sara Mesquita
AU - João Loureiro
AU - Cláudio Haupt Vieira
AU - Lília Perfeito
AU - Joana Gonçalves-Sá
PY - 2026
DA - 2026/10/01
JO - PLOS One
DO - 10.1371/journal.pone.0358063
UR - https://doi.org/10.1371/journal.pone.0358063
AB - Disease now-casting increasingly relies on non-traditional data, such as online activity, which offer fast, inexpensive insights but are often noisy, unstructured, and influenced by behavioral confounders. Models using search engine data have frequently suffered from overfitting and unstable performance. However, pandemics may help distinguish true disease signals from behavioral noise, as both tend to surge during outbreaks. We propose a hybrid approach to improve prediction that combines model evaluation with behavioral insight. We apply simple supervised and unsupervised methods to different flu and COVID-19 datasets, using the 2009 H1N1 and 2020 COVID-19 pandemics as training periods. In both cases, selecting a curated subset of search terms improved model performance, reduced overfitting, and increased stability, consistently outperforming models that used search terms indiscriminately. These results suggest that feature curation informed by behavioral insights can enhance now-casting beyond purely statistical approaches. Our method offers a scalable and adaptable framework for integrating digital epidemiology into public health surveillance.
SN - 1932-6203
ER -