Predicting mortality and hospitalization in heart failure using machine learning: A systematic literature review

• DOI: 10.1016/j.ijcha.2021.100773
• Publication Date: 2021-04-14T05:27:33Z
• Authors: Dineo Mpanya; Turgay Celik; Eric Klug; Hopewell Ntsinjana
• Keywords: Interquartile range; Predictive modelling; Distrust

The partnership between humans and machines can enhance clinical decisions accuracy, leading to improved patient outcomes. Despite this, the application of machine learning techniques in healthcare sector, particularly guiding heart failure management, remains unpopular. This systematic review aims identify factors restricting integration derived risk scores into practice when treating adults with acute chronic failure.Four academic research databases Google Scholar were searched original studies where data was used build models predicting all-cause mortality, cardiac death, failure-related hospitalization.Thirty met inclusion criteria. selected studies' sample size ranged 71 716 790 patients, median age 72.1 (interquartile range: 61.1-76.8) years. minimum maximum area under receiver operating characteristic curve (AUC) for mortality 0.48 0.92, respectively. Models hospitalization had an AUC 0.47 0.84. Nineteen (63%) logistic regression, 53% random forests, 37% decision trees predictive models. None built or externally validated using originating from Africa Middle-East.The variation aetiologies failure, limited access structured health data, distrust among clinicians modest accuracy existing are some precluding widespread use calculators.