Machine learning for epigenetics and future medical applications.

Area of Science:

• Epigenetics and Computational Biology
• Genomic Medicine
• Machine Learning in Healthcare

Background:

• Epigenetic processes are key to medical advancements, but require sophisticated analytical tools.
• Machine learning (ML) has been applied to predict epimutations and disease associations using genomic data.
• Previous ML approaches faced challenges with feature selection and imbalanced genomic datasets.

Purpose of the Study:

• To review the application of molecular epigenetic data in advanced machine learning analysis for medicine.
• To highlight the potential of novel ML techniques, including Deep Learning (DL), for analyzing complex biological data.
• To discuss the integration of ML with epigenetics for improved disease prediction and understanding.

Main Methods:

• Utilizing a combination of Active Learning (ACL) and Imbalanced Class Learning (ICL) for efficient feature selection and handling imbalanced genomic datasets.
• Exploring Deep Learning (DL) for generating and computing novel genomic features tailored to specific classification tasks.
• Applying ML algorithms to epigenetic data associated with transgenerational inheritance and disease.

Main Results:

• ACL and ICL demonstrated improved feature selection and addressed data imbalance issues in genomic datasets.
• Deep Learning (DL) offers a promising new approach for simultaneous feature generation and computation, reducing computational burden.
• The reviewed ML approaches show potential for predicting epigenetic phenomena and associated diseases.

Conclusions:

• Advanced ML techniques, particularly DL, are vital for unlocking the medical potential of epigenetic data.
• The integration of ML with epigenetics can lead to more accurate disease prediction and a deeper understanding of biological mechanisms.
• These computational approaches are broadly applicable to various genomic and biological datasets in medicine.