Psychological determinants of irritable bowel syndrome and its impact on quality of life: a machine learning approaches.

Abstract

Aim: This study examined the associations between psychosocial factors, Irritable bowel syndrome (IBS) diagnosis, and quality of life (QOL) in both control and IBS groups. Additionally, we explored the potential influence of psychosocial factors on the onset of IBS and developed a machine-learning model for IBS prediction.

Background: IBS is a prevalent gastrointestinal disorder, with various factors predicting its severity and associated symptoms.

Methods: Through convenience sampling, a cross-sectional study recruited participants diagnosed with IBS (n=134) and healthy controls (n=150) from Arak Gastroenterology Clinics. Linear regression assessed the impact of psychosocial factors on IBS symptom severity and QOL. Logistic regression analyzed the association of these factors with IBS onset. Machine learning algorithms were used to predict IBS based on psychosocial features. Instruments include IBS-SSS, IBS-QOL, Toronto Alexithymia Scale (TAS-20), Visceral Sensitivity Index (VSI), and Pain Catastrophe Scale (PCS).

Results: A total of 284 participants (61.27% females) were recruited in the study, with a mean age of 36.48±10.75 years. Compared to controls, IBS patients exhibited significantly higher scores on measures of pain catastrophizing scale (PCS, 40.95 vs. 27.73), somatization (13.91 vs. 6.49), and alexithymia (60.23 vs. 54.71) as well as lower VSI (40.54 vs. 72.10). For those with IBS, only difficulty identifying feelings and somatization remained associated with worse symptoms, while VSI presented an inverse correlation. Psychological factors were inversely related to QOL. Elevated levels of alexithymia (OR 1.06; 95% CI 0.48, 1.63), somatization (OR 1.80; 95%CI 1.12, 2.48), and PCS (OR 1.70; 95% CI 1.30, 2.10) were associated with a higher chance of developing IBS, while higher VSI (OR -1.65; 95% CI -1.89, -1.42) was protective. Among machine learning models, logistic regression based on these factors (excluding alexithymia) and age achieved good performance (AUC: 0.86, 95% CI: 0.78-0.94; Accuracy: 0.83, 95% CI: 0.73-0.90) in predicting IBS onset.

Conclusion: Psychological factors were linked to worse IBS symptoms and quality of life. A machine learning model for IBS prediction presented promising results.