Self-Directed Learning for Community Health Workers in Malawi Through Generative AI

Abstract

In many lower and middle-income countries, a lack of resources affects the availability and quality of education and training. In the healthcare domain, access to knowledge can make the difference between life and death. Timely access to technical and clinical guidelines to support decisions is crucial. Healthcare workers need access to up-to-date guidelines on case definitions for surveillance, treatment protocols, and relevant clinical and medical knowledge. However, guidelines documents tend to be bulky and complex and may change over time in response to health priorities, research, or public health emergencies. Generative AI has proven to be a disruptive technology in health care, but its limitations and applicability are subject to experimentation. We present evidence that Large Language Models (LLMs) can be leveraged to facilitate needs-driven and self-directed learning regarding guidelines for healthcare professionals in Malawi. We developed an application called IntelSurv that uses GPT -4 to achieve a ‘chat’ -like functionality where users ask questions about priority diseases, seek clarification on the use of case identification forms, and have access to technical guidelines published by the Ministry of Health. IntelSurv is both a web app and a mobile app and can run either online or offline. Healthcare professionals engaged in disease surveillance and community health in two major cities in Malawi tested the tool and gave positive feedback on its impact. We report on the development of the tool, and its use of GPT -4. We discuss choices of features and functionalities in response to testing and feedback from users.