A Risk Warning Model for Anemia Based on Facial Visible Light Reflectance Spectroscopy: Cross-Sectional Study.

Area of Science:

• Medical Diagnostics
• Biomedical Engineering
• Machine Learning in Healthcare

Background:

• Anemia is a global health concern with significant morbidity and mortality risks.
• Current diagnostic methods, like blood tests, are invasive and uncomfortable for patients.
• There is a need for non-invasive, accurate anemia detection methods.

Purpose of the Study:

• To investigate facial spectral characteristics in anemic patients.
• To develop and validate a machine learning model for predicting anemia risk using facial spectral data.
• To explore the potential of non-invasive spectral analysis for anemia screening.

Main Methods:

• Collected facial images from 78 anemic patients and 78 healthy controls.
• Analyzed spectral differences in facial regions between groups.
• Developed and compared 10 machine learning algorithms for anemia prediction.
• Utilized Shapley additive explanations (SHAP) for model interpretability.

Main Results:

• Significant differences in facial spectral features were observed between anemic and healthy individuals.
• A support vector machine model achieved 87.5% accuracy in distinguishing anemia.
• Key spectral features identified include forehead-570 nm, right cheek-520 nm, and others.

Conclusions:

• Facial spectral data holds clinical significance for anemia diagnosis.
• An early warning model based on spectral information demonstrates high predictive accuracy.
• This approach offers a promising non-invasive tool for anemia risk assessment.