Deep Learning-Based Body Shape Clustering Analysis Using 3D Body Scanner: Application of Transformer Algorithm.

Area of Science:

• Anthropometry
• Biomedical Engineering
• Data Science

Background:

• Accurate body shape classification is crucial for personalized health insights.
• Traditional methods may lack the granularity needed for precise analysis.
• 3D Body Scanner technology offers advanced measurement capabilities.

Purpose of the Study:

• To apply deep learning, specifically transformer learning, for body shape cluster analysis.
• To utilize 3D Body Scanner data for detailed human body shape classification.
• To explore the efficacy of transformer algorithms in body type categorization.

Main Methods:

• Collected 54 variables from 366 adults using a 3D Body Scanner.
• Employed transformer learning and dimensionality reduction models for cluster analysis.
• Utilized Mann-Whitney and Kruskal-Wallis tests for statistical significance.

Main Results:

• Transformer algorithms demonstrated superior performance in body type classification compared to other methods.
• The analysis successfully divided body types into six distinct clusters.
• These clusters included variations within endomorphic and ectomorphic body types.

Conclusions:

• The developed six-cluster model provides significantly more granular body type information.
• This detailed classification can serve as foundational data for predicting health and disease risks.
• Deep learning approaches offer powerful tools for advancing anthropometric analysis.