Integrating machine learning into acupuncture research: a scoping review.

Area of Science:

• Integrative Medicine
• Computational Biology
• Biomedical Informatics

Background:

• Acupuncture research faces challenges with variability and subjectivity.
• Machine learning (ML) offers novel approaches to enhance acupuncture studies.
• This scoping study maps the current landscape of ML applications in acupuncture.

Purpose of the Study:

• To identify disease conditions targeted by ML in acupuncture research.
• To examine ML methods and data inputs used in acupuncture studies.
• To assess limitations and knowledge gaps in ML-driven acupuncture research.

Main Methods:

• Comprehensive literature search in PubMed (2011-2024) using "acupuncture" and "machine learning".
• Identification and analysis of 36 relevant articles.
• Categorization of studies by application, ML algorithm, and data features.

Main Results:

• Seventeen studies focused on predicting acupuncture efficacy, primarily for pain conditions.
• Support Vector Machines (SVM) were the most common algorithm for efficacy prediction.
• Magnetic Resonance Imaging (MRI) was frequently used as a data input (7/11 studies).

Conclusions:

• ML methods, particularly SVM, show potential for improving acupuncture efficacy prediction and treatment personalization.
• Methodological heterogeneity, small sample sizes, and data inconsistencies limit current ML model generalizability.
• Future research requires larger, standardized, multi-center studies to validate clinical utility and advance ML integration in acupuncture.