Abstract
The rapid advancement of artificial intelligence (AI), particularly machine learning (ML), has brought significant transformations in healthcare systems globally. Advanced ML models enable predictive analytics to improve clinical decision-making, predict disease outbreaks and risk assessment and improve health outcomes. However, the integration of machine learning models within a structured problem-solving paradigm such as design science research methodology (DSRM) remains underexplored. This limits the practical implementation and deployment of data-driven applications in healthcare. Therefore, this study proposes a methodology called ML-DSRM, which integrates machine learning and design science research methodology to systematically design , develop and refine healthcare predictive analytics applications. The proposed ML-DSRM methodology embeds ML into the iterative DSRM cycle to ensure that predictive models are contextually relevant, accurate and practically applicable. The methodology outlines methodological principles, from problem identification to system design, validation and deployment. The proposed methodology combines ML's predictive capabilities with DSRM's iterative problem solving approach to improve evidence-based decision-making. This assists in ensuring the effective integration of data-driven applications in healthcare while ensuring the interpretability, reliability and usability of ML-driven pre-dictive analytics. The methodology guides data scientists, healthcare practitioners and system developers to develop and integrate data-driven applications with existing health information systems to improve health outcomes and address real-world healthcare challenges.