Hyperchat: An AI-Powered Chatbot for Hypertension Education on WhatsApp

Main Article Content

Audrey Eyram Agbeve
Richard Abbey
Sandra Adu Afriyie
Prince Ebenezer Adjei

Abstract

 Hypertension, a major cause of cardiovascular diseases like stroke and myocardial infarction, is a critical public health issue in Ghana. Limited access to healthcare professionals and resources hinders effective patient education. Patients have little understanding, and face difficulty following medication and lifestyle recommendations. This study explores the use of Artificial Intelligence to enhance personalised health education in Low- and Middle-Income Countries (LMICs) through the use of an AI chatbot integrated into a widely-used social media platform. The chatbot provides credible answers to hypertension questions and user-specific recommendations for self-management, improving health literacy and patient engagement. HyperChat relies on a Retrieval Augmented Generation pipeline developed with low-code automation services (Chatbase and Make.com) and OpenAI’s GPT-4o model. Ghanaian hypertension treatment guidelines were embedded into a vector database to enable context-specific responses to queries on hypertension management, including dietary and lifestyle modifications, medication reminders, and correct blood pressure measurement procedure. The chatbot was deployed on WhatsApp using ManyChat as the messaging interface, while Make.com (formerly Integromat) handled the automation workflow between ManyChat and Chatbase for real-time interactions. A preliminary assessment was conducted on user (n=21) feedback regarding usability, response adequacy, and health management influence. Based on user testimonials, 90% of users (n=19) found the chatbot user-friendly, as it eliminated the need for additional application downloads. 81% of users (n=17) mentioned that the recommendations were personalized. 76% (n=16) found the information comprehensive, and 90% (n=19) were willing to implement the lifestyle recommendations. Overall, users found the chatbot satisfactory because it provided accurate information on demand. This study highlights the potential of social media platforms to improve health literacy and patient engagement, by providing context-specific health information on-demand. The idea of leveraging widely used social media platforms for health education holds promise for improving public health outcomes in LMICs. 

Article Details

Section
Student Section
Author Biographies

Audrey Eyram Agbeve, Kwame Nkrumah University of Science and Technology

Department of Computer Engineering

Prince Ebenezer Adjei, Kwame Nkrumah University of Science and Technology

Department of Computer Engineering