AI Impact Summit

Global AI Impact Commons

LLM-Powered Chatbot to Support the Informational Needs of Community Health Workers

LLM-Powered Chatbot to Support the Informational Needs of Community Health Workers

India flag

India

Healthcare

High replicability and adaption

Implementing Organisation

Khushi Baby

India, Rajasthan, Udaipur

Civil Society

Implementing Point of Contact

Saket Kumar

R&D Lead

Contributor of the Impact Story

Khushi Baby

Year of implementation

2024

Problem statement

India-s public health system, which delivers last-mile services to over 900 million people, faces a persistent -data-to-action- gap that limits timely and effective responses to community health needs. A key contributor to this gap is the strain on frontline health workers and the constraints of existing support mechanisms. Accredited Social Health Activists (ASHAs) form the backbone of community health delivery, yet they receive only limited initial training and have few opportunities for ongoing upskilling. The reference materials available to them are often insufficient for addressing complex, context-specific, or sensitive questions that arise during fieldwork. Although ASHAs attend periodic group training and meetings, these sessions frequently lack adequate time to address individual queries comprehensively. As a result, many questions remain unanswered and misunderstandings persist. ASHAs rely heavily on Auxiliary Nurse Midwives (ANMs) for guidance, but a single ANM typically supervises ASHAs across multiple villages, making coordination through phone calls or in-person visits difficult. ANMs themselves may not always have immediate answers and are often overburdened, further delaying resolution. These delays affect the quality of care delivered at the doorstep and can reduce ASHAs- ability to complete tasks efficiently, with downstream implications for their performance-based incentives. Additionally, ASHA often hesitate to ask rudimentary or sensitive questions due to fear of judgment, limiting open knowledge-seeking and reinforcing gaps in frontline decision-making.

Impact story details

Khushi Baby (KB) is a non-profit organization working as a systems enabler to public health departments in India. Established in 2016, KB-s team of 130 members include expertise from public health, product design, engineering, field implementation, and data science. Co-created with 250,000 hours of community health workers engagement, KB-s solutions have been used to track the health of over 50M beneficiaries across Rajasthan, Maharashtra, and Karnataka. Through enabling high quality data, timely insights, and effective actions, KB aims to close the loop for the public health system at the last mile.

AI Technology Used

Machine Learning

Key Outcomes

Efficiency

Productivity: Reducing time insights

actions Access

Reach: Providing immediate health info to ASHAs in underserved areas Accuracy

Quality Improvement of the AI responses Knowledge

Skills Impact: Improving the knowledge

digital literacy of frontline workers Economic Value Creation: Increasing ASHA earnings through increased awareness

Impact Metrics

Implementation Context

Pilot

India - Rajasthan, Maharashtra

Current: 8,000 ASHAs Scale: 50,000 ASHAs

Key Partnerships

Government: Rajasthan District administration, Maharashtra District Administration Academic/Research: IIIT Bangalore, Microsoft Research India

Replicability & Adaptation

High

Adapt the knowledge base to align with local clinical guidelines, government protocols, and language variants, including regional dialects, to ensure accuracy and trust. Invest in iterative pilot testing to identify gaps in the knowledge base and reduce -I don-t know- responses before scale-up. Incorporate clear onboarding and AI literacy training to set expectations about the system-s limitations and reduce over-reliance on AI-generated responses. Design expert-in-the-loop workflows that respect supervisors- workload and accountability concerns, potentially by limiting escalation frequency or batching queries.

* The data presented is self-reported by the respective organisations. Readers should consult the original sources for further details.