Uncovering Dental Caries Heterogeneity in NHANES Using Machine Learning

Area of Science:

• Oral Health Research
• Data Science
• Machine Learning

Background:

• The National Health and Nutrition Examination Survey (NHANES) is a vital resource for population health indicators but poses challenges for machine learning due to data complexity.
• Dental caries, a prevalent global disease, requires advanced analytical methods to understand its multifactorial nature and clinical variability.

Purpose of the Study:

• To develop an integrated data-cleaning and unsupervised machine learning pipeline for analyzing the NHANES database.
• To identify previously unrecognized subtypes and associated variables contributing to dental caries heterogeneity.
• To reveal novel dietary signatures and risk factors linked to dental caries.

Main Methods:

• Developed a multidimensional pipeline to clean and optimize the NHANES dataset, addressing missingness and outliers.
• Applied unsupervised machine learning for comprehensive data analysis and visualization.
• Identified distinct patient clusters, particularly in pediatric and senior populations.

Main Results:

• Discovered previously unrecognized subtypes of dental caries and associated variables within the NHANES data.
• Observed distinct data patterns across different age groups, with specific clustering in children and older adults.
• Identified novel associations between dental caries, lead exposure, laboratory markers, and specific dietary patterns.

Conclusions:

• The developed pipeline effectively processes complex health data, revealing significant heterogeneity in dental caries.
• The findings support the development of more precise machine learning models for dental caries and other health conditions.
• This approach highlights the potential for uncovering complex disease patterns in large-scale health surveys.