Height and Weight Estimation From Anthropometric Measurements Using Machine Learning Regressions.

Area of Science:

• Biometrics
• Machine Learning
• Anthropometry

Background:

• Accurate height and weight are vital for tracking health conditions, energy expenditure, and medical treatments.
• Estimating these measurements can be challenging for non-ambulatory or non-communicative patients.
• Current methods often rely on linear regressions, which have limitations.

Purpose of the Study:

• To evaluate the efficacy of advanced machine learning models for estimating height and weight from anthropometric measurements.
• To compare the predictive accuracy of these models against conventional linear regression techniques.

Main Methods:

• Application of Support Vector Regression (SVR).
• Utilizing Gaussian Process (GP) regression.
• Employing Artificial Neural Networks (ANNs).
• Analysis of anthropometric data for model training and validation.

Main Results:

• Machine learning models demonstrated significantly higher accuracy in predicting height and weight compared to linear regressions.
• Predictions were found to be non-sensitive to ethnicity and gender when using more than two anthropometric parameters.
• The models provide robust estimations even when direct measurements are not feasible.

Conclusions:

• Advanced learning models offer superior accuracy for anthropometric estimations.
• These methods enhance the reliability of health monitoring and clinical assessments.
• The study opens new avenues for anthropometric applications in industry, healthcare, and technology.