PRISM-Dengue Sri Lanka: an AI-driven, low-cost platform to close Sri Lanka's urban dengue gap

Background. Sri Lanka has reported dengue incidence between 300 and 400 cases per 100,000 population every year since 2009, three to four times the national target, despite sustained government investment in vector control (USD 12.7M in direct expenditure during the 2017 epidemic year alone). The conventional Integrated Vector Management toolkit is being squeezed by three reinforcing pressures: rising pyrethroid resistance in Aedes aegypti (kdr 1534C allele frequencies of 0.05–0.80 in surveyed populations), the operational inability of Public Health Inspectors to reach rooftop and construction-site breeding habitats in dense urban grids, and climate-driven shifts in outbreak seasonality. Proposal. This concept brief presents PRISM-Dengue, a five-layer AI-driven platform: Predict (climate-driven outbreak forecasting), Reveal (computer-vision detection of breeding micro-habitats from satellite, drone and street imagery), Inform (a two-way mobile application for PHIs and citizens, plus an anonymised public dashboard), Suppress (AI-prioritised inspections, targeted source reduction, and integration with the validated Gampaha Sterile Insect Technique programme), and Measure (independent pre-registered evaluation). The architecture augments Sri Lanka's existing dengue control workforce rather than replacing it, and is built largely on open-source software and freely available satellite imagery. Pilot design. A 12-month pilot is proposed in two high-burden MOH areas (one in Colombo Municipal Council, one in Gampaha) with matched controls, costed at approximately USD 450K. National scale-up across 25 districts over 36 months is costed at approximately USD 2.7M. Primary outcomes: ≥25% reduction in dengue incidence versus matched controls, ≥30% reduction in Container Index, ≥40% improvement in PHI inspection productivity, and a target cost-per-case-averted under USD 100 at national scale, conservative within published benchmarks for Wolbachia (~USD 1,500/DALY in LMIC settings) and genetic vector-control interventions (USD 2–30/case in modelled scenarios). Significance. The platform integrates data and biological tools that already exist in Sri Lanka, costs less than 4% of one epidemic year's vector-control spend, and is engineered for civic accountability through open data release and independent evaluation. The brief is intended for discussion with the National Dengue Control Unit, Ministry of Health, municipal authorities, academic partners, and civil society organisations.