evidoria

← Back to browse

Good practice Imported

AI-Based Dengue Early Warning System — Singapore NEA's Machine-Learning Outbreak Forecasting

Singapore · Singapore · See the Singapore profile · See the Singapore profile

Evidence: Quasi-experimental Top 6% 87/100 · Ask Evidence Copilot about this practice

Singapore's National Environment Agency runs a peer-reviewed machine-learning system that forecasts dengue cases up to three months ahead. In the 2013 outbreak it predicted a peak of 863 cases versus an observed 842, letting campaigns launch two months early.

22,170
2013 dengue epidemic total cases (2013)
863 forecast vs 842 observed
Forecast peak vs observed peak, week 26 (2013)
17 %
Forecast error (MAPE) at 1-week horizon
24 %
Forecast error (MAPE) at 3-month horizon
~29 %
Benchmark SARIMA/step-down regression MAPE at 3 months
~50,000
Gravitrap mosquito-surveillance devices

Details

Maturity
Established
Promoter
National Environment Agency (NEA) — Environmental Health Institute, with National University of Singapore and Nanyang Technological University
Period
Developed on 2001-2012 data, operational since 2013, ongoing
Keywords
public health surveillance, disease early warning, vector-borne disease forecasting, machine learning, decision support

Context

Singapore's National Environment Agency, with NUS and NTU, built a machine-learning (LASSO regression) system forecasting weekly dengue case counts up to three months ahead, trained on 2001-2010 data and validated on 2011-2012 data across twelve weekly submodels using over 200 predictor streams.

Results

During the severe 2013 epidemic (22,170 cases), the model forecast a week-26 peak of 863 cases against an observed peak of 842 the following week, with MAPE of 17% at one week rising to 24% at three months — outperforming benchmark SARIMA/step-down regression models (~29% MAPE at three months). The early forecast let authorities launch prevention campaigns roughly two months earlier and prioritise its ~50,000-device Gravitrap surveillance network.

Implementation

Implementation detail (cost, timeline, staffing, conditions for success) is not yet available for this practice.

Commonly funded by

National / regional programmes

Indicative funding routes for practices of this type — always check each programme's current calls and eligibility rules.

Do you run this practice? Claim it — verified implementers get a public contact pathway and can propose corrections.

Data sources

Where this practice's information was retrieved from, and when.

Similar practices you may find useful