Review on enhancing clinical decision support system using machine learning

Clinical decision-making is a complex patient-centred process. For an informed clinical decision, the input data is very thorough ranging from detailed family history, environmental history, social history, health-risk assessments, and prior relevant medical cases. Identifying the need for structured input data to enable clinical decision-making and quality reporting, such that it is crucial for the end-users is still a challenge. The Clinical Decision Support Systems (CDSS) enhanced using Machine Learning (ML) approaches are described. CDSS aids in the detection and classification of various diseases but they cannot fully capture the environmental, clinical, and social constraints that are taken into consideration by the clinician in the diagnosis process. The authors provide an overview of state-of-the-art healthcare CDSS. The authors initially collected 3165 research articles for this review out of which approximately 3148 records were identified from databases while 17 records were from other sources. A total of 1309 unique articles obtained from the searches were included in the study which was further rigorously evaluated for final inclusion. A generic architecture of computer-based decision support systems using ML is provided. However, the study does not include the comparison of these CDSS in terms of their performance because of heterogeneity in the disease type, modality used for diagnosis, and the ML approach used for detection in CDSS.