AIMS: An Automatic Semantic Machine Learning Microservice Framework to Support Biomedical and Bioengineering...

Area of Science:

• Biomedical Informatics
• Machine Learning in Healthcare
• Computational Biology

Background:

• Biomedical research faces challenges integrating complex data with machine learning (ML) development.
• Automating ML pipelines is crucial for efficient biomedical data analysis and application.

Purpose of the Study:

• To introduce the Automatic Semantic Machine Learning Microservice (AIMS) framework.
• To address challenges in ML for biomedical research through automation and domain-specific ontologies.
• To enable self-supervised learning and continuous adaptation of ML models in the biomedical domain.

Main Methods:

• Developed the AIMS framework with an ontology for ML services tailored to biomedical data.
• Integrated domain knowledge, prioritized model interpretability, and ensured efficient data handling.
• Utilized reinforcement learning and an ontology-based policy schema for self-supervised knowledge learning.

Main Results:

• Demonstrated automation of ML processes in the biomedical domain.
• Successfully integrated a rich domain knowledge base with ML workflows.
• Showcased the ability of ML models to self-learn and adapt to new tasks and data through case studies.

Conclusions:

• AIMS effectively automates ML pipelines in biomedical research, integrating domain knowledge.
• The framework enables self-learning capabilities in machines, enhancing adaptability to novel biomedical challenges.
• AIMS simplifies research routines and elevates the quality of scientific exploration in healthcare.