Related Experiment Video
Updated: Jan 10, 2026
05:47
Evidence-based Knowledge Synthesis and Hypothesis Validation: Navigating Biomedical Knowledge Bases via Explainable AI and Agentic Systems
Published on: June 13, 2025
1.3K
Automated Machine Learning in medical research: A systematic literature mapping study
Giovanna A Castro1, Luiza G Barioto1, Yu H Cao1
1Department of Computer Science (DComp), Federal University of São Carlos (UFSCar), Sorocaba, 18052-780, São Paulo, Brazil.
Artificial Intelligence in Medicine
|November 22, 2025
Summary
Automated Machine Learning (AutoML) enhances healthcare efficiency by automating ML pipelines. Challenges remain in model selection and data preprocessing, with Explainable AI (XAI) integration growing to address interpretability.
Area of Science:
- Medical Informatics
- Artificial Intelligence in Healthcare
- Machine Learning Applications
Background:
- Machine Learning (ML) is crucial in healthcare but demands specialized expertise and time.
- Automated Machine Learning (AutoML) streamlines ML integration into clinical workflows by automating pipeline steps.
- This study maps the literature on AutoML applications in medicine.
Purpose of the Study:
- To systematically map and analyze the application of AutoML in healthcare.
- To identify common AutoML tasks, data types, and challenges.
- To assess the integration of Explainable AI (XAI) with AutoML in medical research.
Main Methods:
- Systematic literature mapping across multiple academic databases.
- Analysis of 244 studies published between 2016 and 2025.
- Categorization of studies by application (diagnosis, prognosis), data type (tabular, image), and ML task (classification, regression).
Main Results:
- AutoML is predominantly used for diagnosis (52.8%) and prognosis (31.9%) prediction, primarily with tabular (43.4%) and image (31.5%) data.
- Classification tasks (81.1%) are most common. Model selection (25.3%) and data preprocessing (13.7%) are key challenges.
- Explainable AI (XAI) adoption is increasing (30.7%), with a notable rise in 2024, indicating a trend towards addressing AutoML's "black-box" nature.
Conclusions:
- AutoML adoption in medicine is expanding, but its lack of interpretability hinders critical applications.
- Integrating Explainable AI (XAI) with AutoML is a growing trend to enhance trust and clinical utility.
- Further research is needed to evaluate the clinical impact of combined AutoML and XAI approaches for decision support systems.
