An Adaptive game-based educational tool for maternal and neonatal health

Abstract

Globally, Maternal and Neonatal Health has become a critical topic of discussion due to high mortality rates in different countries. Despite ongoing efforts to handle the situation, preventable challenges persist in contributing to high rates of Infant and Maternal Mortality Rate (IMMR). In Kenya, the Neonatal Mortality Rate (NMR) stands at 21 deaths per 1000 live births, while the Maternal Mortality Rate (MMR) is reported at 530 deaths per 100,000 live births. In contrast, countries like Italy report much lower rates, with 1.76 neonatal deaths per 1,000 live births and only 12 maternal deaths per 100,000 live births. Education emerges as a critical factor in addressing these challenges, with studies indicating substantial differences in mortality rates between educated and uneducated women. One study highlights that Maternal Health Education (MHE) significantly influences child mortality, with uneducated mothers facing higher risks. Currently, MHE is on a need-to-know basis, either in accidental or planned pregnancies, school-based research or in professional setting. This study comes up with a proof-of-concept that utilizes gamification and artificial intelligence in the simplification of MHE delivery. The study designs an adaptive, game-based educational tool that is engaging, interactive, and context-specific. By increasing the accessibility and reducing the complexity of MHE content, the tool seeks to make maternal health education more widely available and easily understandable for the public. The research provides an overview of Maternal Health highlighting the challenges faced in addressing maternal and neonatal health issues. Additionally, the study explores the principles of game-based learning, examining how these strategies are currently being used in education and their potential to enhance MHE delivery. A methodology section has been outlined detailing how the study has been conducted. A design-based research (DBR) approach has been employed. The study focuses on the iterative development of a culturally adaptive, game-based educational tool for Maternal Health Education (MHE). The use of secondary data has been used to inform the design process, followed by prototyping and user testing to refine the solution. The study includes an implementation and analysis section that clearly explains how the pretrained LLaMA model was integrated to support the system. Additionally, the results section presents the outcomes of using FAISS as a Retrieval-Augmented Generation (RAG) methodology, along with an evaluation report where performance metrics such as precision, recall, and F1-score were each recorded as 1.0.