PROF A.I. OBI

Background: Artificial intelligence (AI) offers potential to enhance community health surveillance through early outbreak detection and improved reporting. However, adoption in low-resource primary healthcare settings remains poorly understood. This study assessed the knowledge, attitudes, uptake, utilisation, and factors influencing AI use in disease surveillance among health workers in selected Primary Health Centres (PHCs) in Benin City, Edo State, Nigeria. Methods: An analytical cross-sectional study was conducted among 230 health workers selected by stratified multistage sampling from 23 PHCs in Oredo and Egor Local Government Areas. Data were collected using a structured, self-administered questionnaire adapted from the Technology Acceptance Model and standard KAP frameworks. Knowledge,attitude, uptake, utilisation, and influencing factors were assessed using descriptive and inferential statistics (chi- square, Fisher’s exact test, and logistic regression). Results: All respondents had heard of AI (100%); the internet was the primary source (99.1%). Good knowledge of AI in disease surveillance was found in 77.8% of respondents,while 65.2% had a positive attitude towards AI use. However, actual uptake was very low(7.4%). Among the few users (n=17), the most used tools were HealthMap (76.5%) and ChatGPT (52.9%), mainly for report writing (100%) and trend analysis (52.9%). Most users reported rare use (64.7%) and had discontinued use (76.5%) due to accuracy concerns, lack of institutional support, and technical issues. Factors significantly associated with uptake included age, marital status, occupation, and knowledge level (p<0.05). Major barriers to AI use were lack of funding (100%), absence of institutional training (100%), inadequate infrastructure (99.6%), unclear ethical guidelines (69.6%), and data privacy concerns (67.0%).Unclear ethical guidelines were the only independent predictor of AI uptake (OR=0.069,p<0.001).Conclusion: While knowledge and attitudes towards AI are reasonably favourable among PHC health workers in Benin City, actual uptake and sustained use remain very low. Systemic barriers especially lack of ethical guidelines, infrastructure, training, and funding must be addressed to translate awareness into practice. Recommendations: Federal and state health authorities should develop clear ethical and operational guidelines for AI in disease surveillance, integrate AI literacy into training curricula, invest in digital infrastructure, and implement phased, supervised AI tools starting with low-risk functions such as report generation and trend analysis