人工知能コミュニティ分析とナノポアメタゲノムサーベイランスを活用した早期腸管病原体アウトブレイク監視

Foodborne enteric infections are a major public health and economical burden, yet their surveillance often relies on latent indicators that delay containment efforts by several days and weeks. Conversely, whole metagenome shotgun sequencing of communal wastewater allows continuous monitoring of enteric pathogens. Spikes in abundance can be observed several weeks before the first case reports emerge. In addition, AI-driven social media mining, already in use for public opinion analytics, could be repurposed for predicting outbreaks at the community level by predicting the number of people experiencing symptoms in the population given their social media activity. Here we report how AI-driven community analytics and high-throughput long-read metagenomic surveillance of communal wastewater microbiota were combined to monitor non-typhoidal salmonellosis in Quebec City, Canada, from August 2023 to February 2024. Both approaches indicated similar fluctuations over time for: (i) people experiencing salmonellosis symptoms, and (ii) Salmonella enterica relative abundance in wastewater, with predicted cases leading metagenomic peaks by a week. Moreover, both approaches detected a maximum around September 13th, 2023, 5 weeks before a Salmonella food recall for the Quebec and Ontario provinces was made by the Public Health Agency of Canada. We therefore suggest that continuous AI-driven analytics and wastewater metagenomics monitoring could become part of a nationwide surveillance pipeline from the community scale to the molecular level.