Related Experiment Video
Updated: Apr 9, 2026

Constructing and Visualizing Models using Mime-based Machine-learning Framework
Published on: July 22, 2025
Leveraging Google search data for predictive surveillance of Mpox: Toward active outbreak prevention
Cheng-Hsun Hsueh1, Yi-Chun Chiu2,3,4,5, Wei-Ming Cheng1,6
1Division of Urology, Department of Surgery, Taipei City Hospital, Zhongxiao Branch, Taipei, Taiwan.
Background:
Mpox (formerly monkeypox) has an incubation period of 3-17 days, creating a window when infections are present but undetected. Pre-exposure prevention depends on the two-dose Mpox vaccine, which requires about six weeks from the first injection to reach peak protection. Therefore, early and affordable surveillance capable of identifying outbreaks prior to the development of immune protection is critical. Digital traces such as Google search activity may provide early signals of epidemic dynamics, but their utility for Mpox in real-world contexts remains insufficiently characterized.
Methods:
We assembled monthly country-level Mpox cases and deaths from the World Health Organization global trends dashboard from 1 May, 2022 to 30 November, 2024 and matched them to country-specific Google Trends relative search volume (RSV) for "Monkeypox." A total of 96 countries were analyzed, including 38 Organization for Economic Cooperation and Development (OECD) members as high-income comparators. For each country we computed Pearson and Spearman correlations between RSV and (i) concurrent cases and deaths and (ii) cases and deaths one month later to assess short-term predictability.
Results:
Concurrent RSV-case associations were predominantly positive, with a third of countries showing statistically significant correlations; several OECD members exhibited strong relationships (e.g., Canada, France, Belgium, United States). Correlations with deaths were uniformly weak, consistent with sparse, delayed mortality. Predictive performance improved when RSV led cases by one month: nearly a half of countries demonstrated significant RSV and cases associations, with OECD countries showing higher correlations.
Conclusions:
Google search interest provides an informative and practical adjunct to conventional Mpox surveillance, offering a one-month early warning in most settings, particularly across OECD countries, while mortality forecasting remains unreliable due to low event counts and longer clinical latency. By leveraging Google Trends as a predictive tool, public-health authorities can anticipate rising transmission and implement preventive measures-such as targeted vaccination-before outbreaks escalate.
Related Concept Videos
Steps in Outbreak Investigation
Investigation of Disease Outbreaks
Principles of Disease Surveillance
Methods of Medium Optimization
Censoring Survival Data