A Microcosmic Syndrome Differentiation Model for Metabolic Syndrome with Multilabel Learning

Area of Science:

• Computational Medicine
• Bioinformatics
• Traditional Chinese Medicine

Background:

• Metabolic syndrome (MS) is a complex multisystem disease with challenges in Traditional Chinese Medicine (TCM) syndrome differentiation.
• Existing methods struggle with concurrent TCM syndromes and a holistic view of microscopic indexes.
• This study addresses these limitations by proposing a novel approach.

Purpose of the Study:

• To develop a machine learning model for objective and comprehensive syndrome differentiation in MS.
• To address the challenges of concurrent TCM syndromes and the relationship between microscopic indexes and TCM syndromes.
• To establish a foundation for multidimensional syndrome differentiation in MS.

Main Methods:

• A standardization scale for TCM four diagnostic information for MS was designed.
• A microcosmic syndrome differentiation model, ML-kNN, was constructed using multilabel learning (MLL) techniques based on 39 physicochemical indexes.
• MLL was compared with single learning algorithms, and ML-kNN results were evaluated against TCM information.

Main Results:

• The ML-kNN model demonstrated superior performance compared to other algorithms, achieving an average diagnostic precision of 71.4%.
• Results derived from physicochemical indexes were comparable to those from TCM information.
• The parameter 'k' showed minimal influence on the prediction accuracy of the ML-kNN model.

Conclusions:

• The developed microcosmic syndrome differentiation model using MLL effectively predicts syndrome elements and handles multiple concurrent TCM syndromes in MS.
• A complex correlation exists between TCM syndrome elements and physicochemical indexes, warranting further investigation.
• This approach promotes the development of objective syndrome differentiation for MS.