Screening for Osteoporosis from Blood Test Data in Elderly Women Using a Machine Learning Approach

Area of Science:

• Gerontology
• Medical Informatics
• Biostatistics

Background:

• Osteoporosis diagnosis relies on bone mineral density (BMD) measured by dual-energy X-ray absorptiometry (DXA).
• Predicting low BMD without DXA could offer a more accessible screening method for elderly women.
• Machine learning (ML) presents a potential approach for developing such predictive models.

Purpose of the Study:

• To develop and evaluate machine learning models for predicting low BMD in elderly women.
• To compare the performance of various ML algorithms in identifying individuals at risk for osteoporosis.
• To identify key clinical and demographic factors associated with low BMD.

Main Methods:

• Utilized medical records from 2541 females visiting an osteoporosis clinic.
• Employed machine learning models including logistic regression, decision tree, random forest, gradient boosting trees, and lightGBM.
• Input features included age, body mass index (BMI), and blood test results; model performance was assessed using accuracy and area under the curve (AUC).

Main Results:

• The lightGBM model achieved the highest accuracy (0.834) and AUC (0.961).
• Gradient boosting also showed strong performance with an accuracy of 0.800 and AUC of 0.840.
• Key predictors for low BMD identified were BMI, age, and platelet count, with BMI, age, and ALT being significant in the lightGBM model.

Conclusions:

• Machine learning, particularly the lightGBM model, can effectively predict low BMD in elderly women without DXA.
• Clinical factors such as BMI and age are crucial indicators for low BMD.
• This approach holds promise for non-invasive osteoporosis screening and risk stratification.