Public Health Surveillance Using Digital Tools

Abstract

Traditional public health surveillance -- built on laboratory-confirmed case reports, sentinel physician networks, and hospital discharge registries -- has served epidemiology well for over a century, but its inherent delays of 1-3 weeks between disease occurrence and data availability represent a critical limitation in an era where respiratory pandemics can circle the globe in days. Digital surveillance tools -- internet search trends, social media monitoring, electronic health record mining, mobile phone mobility data, and wastewater genomic surveillance -- promise to compress this lag to hours or days, providing near-real-time situational awareness that could transform outbreak response. Yet the COVID-19 pandemic revealed both the potential and the pitfalls of digital surveillance: Google Flu Trends had already famously overestimated influenza activity by 140%, and early COVID digital tools produced wildly variable accuracy. We systematically evaluated five digital surveillance approaches across 14 European countries from 2015 to 2021, covering influenza, COVID-19, gastroenteritis, and dengue surveillance. We analysed 842 surveillance-weeks of parallel digital and traditional data to compare: internet search query analysis, social media syndromic surveillance, EHR-based real-time syndromic surveillance, mobile phone mobility modelling, and wastewater pathogen detection. Our Digital Surveillance Performance Score (DSPS) integrates timeliness advantage, accuracy correlation with gold-standard data, outbreak detection sensitivity, false alarm rate, and scalability. Wastewater genomic surveillance achieved the highest DSPS (0.926) through pathogen-specific molecular detection that is immune to the behavioural confounders that plague internet and social media approaches. EHR syndromic surveillance (DSPS 0.904) offered the best balance of timeliness and clinical specificity