Efficacy of machine learning algorithms versus conventional risk assessment tools in predicting acute kidney injury: a systematic review

Abstract

In this study, we investigated the efficacy of machine learning (ML) algorithms and compared them to conventional risk assessment tools in predicting acute kidney injury (AKI). Our evaluation encompassed a diverse array of ML models such as logistic regression, random forests, and neural networks and each model is evaluated against traditional risk assessment techniques. We documented key performance metrics such as accuracy, sensitivity, specificity, and overall predictive performances of previous studies and trials. Preliminary results reveal a notable superiority of ML algorithms over conventional tools, particularly in terms of accuracy and sensitivity. Our findings show the potential of ML models has enhanced early detection and intervention strategies for AKI and are proven a more effective approach to risk prediction. By leveraging the strengths of these innovative algorithms our healthcare providers can potentially improve patient outcomes through more precise and timely assessments. This study shows how incorporating machine learning (ML) technologies into clinical practice can change the way we identify and manage acute kidney injury (AKI) risks. It's important to further investigate how these algorithms can be practically implemented in different clinical settings to validate their effectiveness.