Systematic Mapping Study of AI/Machine Learning in Healthcare and Future Directions

Area of Science:

• Healthcare Informatics
• Machine Learning Applications
• Medical Research

Background:

• The integration of machine learning (ML) into healthcare is rapidly expanding.
• A structured overview of current ML research in healthcare is needed to guide future endeavors.
• Systematic mapping studies are crucial for understanding research trends and identifying knowledge gaps.

Purpose of the Study:

• To systematically categorize and map the research landscape of machine learning applications in healthcare.
• To identify prevalent research themes, methodologies, and disease focuses within this domain.
• To provide insights into emerging trends and future research directions in healthcare ML.

Main Methods:

• Conducted a systematic mapping study of literature on the use of machine learning in healthcare.
• Utilized keywords 'use of machine learning in healthcare' across major academic databases, including Google Scholar.
• Categorized 1400 retrieved papers based on study objective, methodology, problem type, and disease studied.

Main Results:

• Identified five primary research categories: interpretable ML, medical image evaluation, EHR processing, security/privacy frameworks, and transfer learning.
• Cancer emerged as the most frequently studied disease, while epilepsy was among the least studied.
• Medical image evaluation is a highly researched area, and Interpretable ML/Explainable AI is a rapidly growing field.

Conclusions:

• The field of machine learning in healthcare is diverse, with distinct areas of focus and varying levels of research attention.
• Future research should consider understudied diseases and further explore the potential of Interpretable ML/Explainable AI.
• This mapping provides a foundational understanding for researchers entering or advancing within the domain of machine learning in healthcare.