Systematic Review and Meta-Analysis of the Diagnostic Accuracy of Machine Learning and Deep Learning Models to Detect Malaria: A Protocol

Abstract

Abstract Background: Malaria continues to be a serious public health problem, particularly in tropical and subtropical nations, with an estimated 247 million malaria cases and 619,000 malaria deaths worldwide. Despite early and accurate diagnosis being essential for limiting public health challenges, existing conventional diagnostic techniques face many overwhelming challenges. However, recent advances in machine learning and deep learning models have shown promising results in overcoming the challenges of malaria diagnosis. Therefore, this study aims to conduct a systematic review and meta-analysis to synthesize the existing literature and evaluate the overall performance and reliability of machine learning and deep learning models. Methods: A systematic literature search will be conducted to identify all relevant studies from electronic databases including PubMed, Medline, Embase, Cochrane Library, Web of Science, Scopus, Google Scholar and Science Direct until November 15, 2023. Study selection will be rigorously carried out independently using the established eligibility criteria and strictly following the PRISMA guidelines. Then two authors will extract data files from a full-text article using a predefined extraction checklist. The methodological qualities of the included studies will be assessed using the QUADAS-2 tool. Pooled estimates of key diagnostic performance measures will be calculated, and the synthesis outcomes will be presented using forest plots. Random effects models will be used to account for the expected variation among studies. The presence of heterogeneity among studies will be assessed using the I² statistic. If significant heterogeneity is observed, we will explore the source of heterogeneity using subgroup analyses meta-regression, and sensitivity analysis. Discussion: The rigorous meta-analysis approach used will ensure a robust knowledge synthesis from the extracted data files. The result of the study will provide empirical evidence for policymakers, researchers, and other stakeholders who would like to use the data to establish the appropriate strategy. Systematic review registration: This review protocol has been submitted to PROSPERO for registration.