Building an Intelligent Platform for Childhood Pneumonia and Tuberculosis Surveillance

Context & Background

Pneumonia and Tuberculosis (TB) are leading causes of pediatric mortality in India, with India carrying over 22% of the global childhood TB burden. Traditional epidemiological reporting is slow and manually driven, failing to identify local outbreaks in time to save lives.

Problems to be Addressed

Rural healthcare networks lack coordinated surveillance platforms that can predict regional disease spikes. Missing data and unstandardized recording at primary clinics make epidemiological modeling unreliable.

Aims and Objectives

1. Build an AI-enabled digital platform (DICEFlow) for primary health surveillance.
2. Monitor 2500 mothers and children in Gurgaon, Haryana.
3. Formulate predictive models for disease spikes and drug-resistant TB variants.

Methodology

In collaboration with NGO Swasti, the project deploys a mobile interface (DICEFlow) to map workflows for community health workers. Local data is gathered, cleaned, and processed using causal temporal machine learning. Changepoint detection models analyze spikes in respiratory illness, and custom libraries handle missing data.

Expected Outcomes

Deployment of DICEFlow platform, prediction models for respiratory outbreaks, and integration-ready APIs for India's National Digital Health Mission.