Artificial intelligence in zoonotic disease surveillance: A systematic review with One Health implications for low- and middle-income countries

Zoonotic diseases account for approximately ۶۰% of known infectious diseases and ۷۵% of emerging threats, disproportionately affecting low- and middle-income countries (LMICs). This systematic review synthesises evidence on artificial intelligence (AI) applications in veterinary surveillance relevant to zoonotic disease monitoring, with attention to LMIC implementation contexts. Following PRISMA ۲۰۲۰ guidelines and JBI methodology, we searched PubMed, Scopus, and Web of Science (January ۲۰۱۵–December ۲۰۲۵). Two reviewers independently performed screening, full-text assessment, and data extraction. Methodological quality was appraised using a six-item framework integrating JBI and PROBAST elements. Thirty empirical studies met inclusion criteria (۲۰ primary empirical + ۱۰ descriptive operational evaluations); ۵۰.۰% addressed LMIC settings. Four application domains were identified: predictive surveillance (۳۰.۰%), event-based surveillance (۲۳.۳%), automated monitoring (۲۳.۳%), and disease detection (۲۳.۳%). Fourteen studies (۴۶.۷%) demonstrated direct zoonotic relevance. A machine learning rabies risk-stratification approach yielded an approximately three-fold increase in epidemiologically useful surveillance data, and event-based syndromic classification achieved F۱-macro of ۰.۹۹۹۵ under retrospective validation. Edge AI and transfer learning approaches demonstrated feasibility for resource-constrained settings. AI-enabled approaches may strengthen zoonotic disease surveillance within One Health frameworks, particularly in LMIC settings, though most evidence reflects proof-of-concept and retrospective validation; prospective operational research is needed.

کلیدواژه ها:

Artificial intelligence ، machine learning ، Zoonotic disease ، One health