Understanding current states of machine learning approaches in medical informatics: a systematic literature review

• Published in: Health and technology Vol. 11; no. 3; pp. 471 - 482
• Main Authors: Hasan, Najmul, Bao, Yukun
• Format: Journal Article
• Language: English
• Published: Berlin/Heidelberg Springer Berlin Heidelberg 01.05.2021; Springer Nature B.V
• Subjects: Algorithms; Biological and Medical Physics; Biomedical Engineering and Bioengineering; Biomedicine; Biophysics; Computational Biology/Bioinformatics; Data mining; Decision making; Decision trees; Digital libraries; Engineering; Health informatics; Information systems; Literature reviews; Machine learning; Medicine/Public Health; Personal health; Prediction models; R & D/Technology Policy; Researchers; Review Paper; Support vector machines; Systematic review; Systematic review; Knowledge discovering; Machine learning; Medical informatics
• ISSN: 2190-7188; 2190-7196
• DOI: 10.1007/s12553-021-00538-6

Knowledge mining (KM) tends to deliver the tools and associated components to extract enormous amounts of data for strategic decision-making. Numerous machine learning (ML) techniques have been applied in medical information systems. These can significantly contribute to the decision-making process, such as diagnosis, prediction, and exploring the benefits of clinical care. This study aims to determine insights into the current state of data mining applications employed by ML in the field of medical informatics (MI). We believe that this exploration would lead to many unrevealed answers in predictive modelling in medical informatics. A systematic search was performed in the most influential scientific electronic databases and one specific another database between 2016 to 2020 (April). Research questions are outlined after the researcher has studied previous research done on the subject. We identified 51 related samples out of 1224 searched articles that satisfied our inclusion criteria. There is a significant increasing pattern of ML application in MI. In addition, the most popular algorithm for classification problem is Support Vector Machine (SVM), followed by random forest (RF). In contrast, "Accuracy" and "Specificity" are the most commonly used mechanisms for performance indicators calculation. This systematic literature review provides a new paradigm for the application of ML to MI. By this investigation, the unknown areas of ML on MI were explored.