Abstract / Summary
Abstract Diabetes mellitus is often accompanied by psychological distress and sexual dysfunction, yet the complex relationships between these outcomes and biochemical characteristics in women remain insufficiently understood. This cross-sectional study, using a data-mining approach, examined the associations between sexual and mental health indices and metabolic and biochemical markers in 124 reproductive-aged women with type 2 diabetes. Sexual function was assessed using the Female Sexual Function Index (FSFI), psychological status was evaluated with the Depression, Anxiety, and Stress Scale-21 (DASS-21), and biochemical measures included fasting blood glucose, glycated hemoglobin (HbA1c), two-hour postprandial glucose (2HPP), and lipid profile components. The findings showed that most participants had impaired sexual function, particularly in the domains of desire, arousal, lubrication, and pain, while considerable levels of depression, anxiety, and stress were also observed. Data-mining analyses and machine learning models identified 2HPP as the most important glycemic predictor of overall sexual function and several FSFI domains, whereas HbA1c showed the strongest associations with sexual desire and pain. Triglycerides emerged as the most important lipid predictor of both sexual and psychological health indices, while HDL demonstrated relatively protective associations. Mutual information analysis further revealed the greatest degree of shared information between triglycerides, VLDL, and the domains of sexual pain, stress, and depression. These findings underscore the central role of postprandial glucose dysregulation and dyslipidemia in the co-occurrence of sexual dysfunction and psychological distress in women with diabetes and suggest that data mining and machine learning may provide clinically meaningful insights into the complex metabolic, psychological, and sexual interactions underlying diabetes-related health outcomes.