# Menopausal Women With Diabetes and Comorbid Anxiety: Integrated Management Involving Cognitive Behavioral Therapy, Diabetes Education, and Pharmacotherapy

## TL;DR

Menopausal women with diabetes and anxiety may benefit from combined therapy involving CBT, education, and medication, with machine learning improving outcome predictions.

## Contribution

This study evaluates integrated management for menopausal women with diabetes and anxiety, using machine learning to enhance treatment prediction.

## Key findings

- Integrated management showed potential for improving both metabolic and psychological outcomes.
- Machine learning models like Random Forest and XGBoost outperformed logistic regression in predicting treatment outcomes.
- Key predictors of outcomes included depression scores, HbA1c levels, and treatment history.

## Abstract

Background: Menopausal women with diabetes frequently experience comorbid anxiety, which complicates both metabolic and psychological management. Integrated interventions combining cognitive behavioral therapy (CBT), diabetes education, and pharmacotherapy may offer comprehensive benefits, yet evidence remains limited.

Methodology: A retrospective study was conducted among menopausal women with diabetes and anxiety. Participants were categorized into three groups: CBT + education + pharmacotherapy, education only, and medications only. Clinical, demographic, and psychological data were analyzed using descriptive and inferential statistics. Machine learning (ML) models, including Logistic Regression, Random Forest, and XGBoost, were applied to identify predictors of treatment outcomes, with SHAP analysis used for model interpretability.

Results: The study included 300 women with a mean age of 54.13 years and a body mass index (BMI) of 28.07 kg/m². HbA1c averaged 7.47% and fasting glucose 140.08 mg/dL. Socioeconomic distribution was high (n = 106, 35.3%), middle (n = 106, 35.3%), and low (n = 88, 29.3%). Blood pressure categories differed significantly across groups (χ² = 0.037, P < 0.05). Anxiety scores trended toward improvement under integrated management (F = 3.707, P = 0.055). Logistic regression highlighted age (Exp(B) = 1.045, P = 0.064) and cognitive function (Exp(B) = 1.087, P = 0.090) as borderline predictors. ML outperformed logistic regression, with Random Forest (accuracy 81.5%, Receiver Operating Characteristic-Area Under the Curve (ROC-AUC) 0.85) and XGBoost (accuracy 83.4%, ROC-AUC 0.82). Depression score (mean 11.80, standard deviation (SD) = 4.54), HbA1c, and treatment history emerged as key predictors.

Conclusions: Integrated management strategies appear to be effective in addressing both metabolic and psychological outcomes. Ensemble ML offers superior predictive insight, supporting personalized care.