Machine Learning Algorithms for Predicting Hospital Readmission and Mortality Rates in Patients with Heart Failure

Purpose: The potential of predictive analytics in enhancing resource allocation and patient care for Heart failure (HF) outcomes is significant. This review aims to highlight this potential by analyzing existing studies and identifying the main barriers and challenges to applicability in all settings. Design/ Methodology/ Approach: A comprehensive search of related articles was meticulously conducted across electronic databases, including Google Scholar, Web of Science, and PubMed. Using precise search phrases and keywords, 1,835 scholarly articles published between 1 January 2017 and 14 May 2024 were retrieved. Only 23 articles that met the strict inclusion criteria were considered, ensuring the validity of the findings. A quantitative meta-analysis approach was utilised. Research Limitation/Implication: This research offers insights into enhancing healthcare outcomes as we analyse the challenges and feasibility of applying ML algorithms to predict heart failure outcomes in low-income settings. Findings: The challenges include scalability, ethical and legal issues, the choice of appropriate ML model, interpretability, data availability, and healthcare professional mistrust of these ML algorithms. Practical implications: This study offers practical strategies to bridge the gap between clinical practice and predictive analytics in these regions. These strategies should inspire and motivate healthcare professionals, researchers, and policymakers to consider and implement them. Social implication: This study provides insights that may improve HF outcomes and healthcare delivery. Originality/Value: The review identifies current gaps in the research, such as the need for more robust validation studies, the challenge of model interpretability, and the necessity for models that can be easily integrated into clinical workflows.