Classification of self-care patterns in Korean adults with prediabetes using unsupervised machine learning: a...

Area of Science:

• Nursing
• Public Health
• Data Science

Background:

• Prediabetes affects a significant portion of the adult population, necessitating effective self-care strategies.
• Understanding diverse self-care patterns is crucial for developing targeted interventions.
• Orem's Self-Care Theory provides a framework for analyzing self-care deficits and promoting health.

Purpose of the Study:

• To classify self-care patterns among Korean adults with prediabetes using unsupervised machine learning.
• To apply Orem's Self-Care Theory components (demands, agencies, behaviors) in classifying self-care patterns.
• To identify distinct subgroups based on their self-care profiles for tailored interventions.

Main Methods:

• Secondary data analysis of the 2023 Korea National Health and Nutrition Examination Survey.
• Application of Principal Component Analysis for dimensionality reduction.
• K-means clustering to identify self-care pattern groups, with variable standardization using min-max normalization.

Main Results:

• Three distinct self-care pattern groups were identified: high self-care performance, latent self-care risk, and self-care vulnerable.
• Significant differences were observed between groups regarding education, income, health literacy, fasting blood glucose, and HbA1c levels.
• Each group demonstrated unique profiles across self-care demands, agencies, and behaviors.

Conclusions:

• Unsupervised machine learning effectively classifies self-care patterns in prediabetic adults.
• Findings underscore the need for multidimensional self-care profiles in designing nursing interventions.
• Orem's Self-Care Theory is applicable, and machine learning can identify at-risk subgroups for early intervention.