21 Novembre 2026

Psicologia Punto One | Psicologia e Psicologi nel Web

Menu

Menu

SCHEDA SCIENTIFICA2026Studio osservazionale

Development and internal validation of a machine learning-based disease burden index for irritable bowel syndrome: a multicentre cross-sectional study.

Yanbin Wei, Heyang Zhang, Nan Wang, Haifeng Jin, Zitan Feng, Jiapeng Guo, Jiafei Peng, Jia Feng, Jia Zhi, Shutian Zhang, Shengtao Zhu, Xin Yao

Annals of medicine · 2026

Abstract originale

BACKGROUND: Irritable bowel syndrome (IBS) is one of the most common disorders of gut-brain interaction in gastroenterology clinics worldwide, becoming a significant public health burden. This study aims to provide an exploratory approach for comprehensive assessment of IBS burden and may serve as a basis for future external validation and further evaluation of its potential clinical application. PATIENTS AND METHODS: This multicentre cross-sectional study included IBS patients diagnosed according to the Rome III and/or Rome IV criteria. Rome criteria-based comparisons and latent class analysis (LCA) were performed to characterize clinical heterogeneity, followed by the development of the IBS burden index (IBS-BI) using machine learning. RESULTS: Compared to the Rome IV-ineligible group, the Rome IV-eligible group exhibited significantly more severe IBS symptoms, somatization symptoms, anxiety symptoms, depression symptoms, and poorer health-related quality of life (all p <0.05). Linear regression analysis showed that the severity of these symptoms was inversely related to health-related quality of life (all p <0.05). LCA identified four heterogeneous IBS burden phenotypes: high-symptom burden, low-symptom burden, somatization-dominant, and psychological comorbid phenotypes. Using machine learning, we developed the IBS-BI and an interactive clinical calculator. The validation was limited to an internal training-validation split, and no external validation was performed. CONCLUSIONS: Patients with IBS who fulfilled the Rome IV criteria showed a higher disease burden, which may be related to the more stringent diagnostic requirements of these criteria. The IBS-BI may have potential value for evaluating disease burden and patient stratification. Its clinical utility and broader applicability require further validation in independent external cohorts and prospective studies.

Parole chiave

HumansIrritable Bowel SyndromeCross-Sectional StudiesMachine LearningFemaleQuality of LifeAdultSeverity of Illness IndexSymptom BurdenMiddle AgedMaleDepressionAnxietyCost of IllnessDisease BurdenLatent Class AnalysisRome Criteria
Leggi il testo integrale originale

Testo estratto dalla pubblicazione: sono esclusi immagini, media, supplementi e parti soggette a diritti distinti.

Development and internal validation of a machine learning-based disease burden index for irritable bowel syndrome: a multicentre cross-sectional study

Abstract

Background

Irritable bowel syndrome (IBS) is one of the most common disorders of gut-brain interaction in gastroenterology clinics worldwide, becoming a significant public health burden. This study aims to provide an exploratory approach for comprehensive assessment of IBS burden and may serve as a basis for future external validation and further evaluation of its potential clinical application.

Patients and Methods

This multicentre cross-sectional study included IBS patients diagnosed according to the Rome III and/or Rome IV criteria. Rome criteria-based comparisons and latent class analysis (LCA) were performed to characterize clinical heterogeneity, followed by the development of the IBS burden index (IBS-BI) using machine learning.

Results

Compared to the Rome IV-ineligible group, the Rome IV-eligible group exhibited significantly more severe IBS symptoms, somatization symptoms, anxiety symptoms, depression symptoms, and poorer health-related quality of life (all p <0.05). Linear regression analysis showed that the severity of these symptoms was inversely related to health-related quality of life (all p <0.05). LCA identified four heterogeneous IBS burden phenotypes: high-symptom burden, low-symptom burden, somatization-dominant, and psychological comorbid phenotypes. Using machine learning, we developed the IBS-BI and an interactive clinical calculator. The validation was limited to an internal training-validation split, and no external validation was performed.

Conclusions

Patients with IBS who fulfilled the Rome IV criteria showed a higher disease burden, which may be related to the more stringent diagnostic requirements of these criteria. The IBS-BI may have potential value for evaluating disease burden and patient stratification. Its clinical utility and broader applicability require further validation in independent external cohorts and prospective studies.

Keywords: Irritable Bowel Syndrome, Rome Criteria, Latent Class Analysis, Machine Learning, Disease Burden

Introduction

Irritable bowel syndrome (IBS) is a functional gastrointestinal disorder characterized by recurrent abdominal pain, accompanied by changes in bowel habits and stool characteristics. Its development involves complex interactions among impaired gut-brain communication, visceral hypersensitivity, gut microbiota dysbiosis, and psychological factors, making it one of the most common disorders of gut-brain interaction worldwide [1]. A systematic review and meta-analysis of epidemiological studies estimated the global prevalence of IBS to be approximately 15.0%, with a gradually increasing trend in recent years [2]. However, considerable heterogeneity remains in the reported prevalence of IBS due to variations in diagnostic criteria, study designs, and geographical regions [3,4]. Although IBS is not a life-threatening condition, its chronic and recurrent nature, together with its multifactorial pathogenesis, imposes a substantial physical and psychological burden on patients. IBS significantly impairs quality of life and daily functioning, while also increasing healthcare utilization and socioeconomic costs, representing an important global public health challenge [5,6].

IBS subtypes were initially proposed within the framework of the Rome III criteria based on the Bristol Stool Form Scale. Patients were classified into four major subtypes: IBS with predominant constipation (IBS-C), IBS with predominant diarrhoea (IBS-D), IBS with mixed bowel habits (IBS-M), and IBS unclassified (IBS-U) [7]. Although the Rome IV criteria, introduced in 2016, incorporated updated insights into biopsychosocial factors and gut-brain interactions, the classification system remains primarily based on gastrointestinal symptoms and does not include important dimensions such as somatization, anxiety, depression, or quality of life [8]. Therefore, some studies have attempted to move beyond traditional IBS subtypes by incorporating psychological burden and quality-of-life measures to identify more clinically meaningful patient phenotypes. Several studies have classified IBS patients into 4–7 distinct clusters, providing new perspectives for personalized management [9,10]. However, differences in study populations, data sources, and assessment approaches have resulted in inconsistent findings, and these classification systems have not yet been integrated into a quantitative framework for assessing overall disease burden, limiting their clinical utility.

In addition, commonly used IBS assessment tools mainly focus on individual domains, such as gastrointestinal symptoms, psychological status, or quality of life [11–13]. Although multiple assessment tools can provide information across different clinical domains, interpreting these measures separately may increase complexity and may not adequately reflect the overall disease burden of IBS. Therefore, an integrated tool combining gastrointestinal symptoms, psychological status, and quality of life is needed. Such a tool can consolidate multidimensional clinical information into a single measure, facilitating IBS burden quantification, patient stratification, individualized management, treatment response evaluation, and long-term monitoring.

This study conducted a cross-sectional analysis based on the demographic characteristics of IBS patients, incorporating standardized scale assessments of IBS symptoms, somatization symptoms, anxiety symptoms, depression symptoms, and health-related quality of life. We systematically compared the multidimensional clinical characteristics of IBS patients who met and did not meet the Rome IV criteria. Subsequently, latent class analysis (LCA) was performed to identify homogeneous clinical phenotypes of IBS patients. Based on these findings, we developed the IBS Burden Index (IBS-BI) to integrate multidimensional clinical information and provide a quantitative measure of overall disease burden. A simplified clinical tool was also developed to facilitate IBS-BI calculation. This study aims to provide an exploratory approach for comprehensive assessment of IBS burden, and may serve as a basis for future external validation and further evaluation of its potential clinical application.

Patients and methods

Study design and Ethics

This multicentre cross-sectional study was conducted in the Bethune International Peace Hospital (July 2017 to September 2018) and Beijing Friendship Hospital, Capital Medical University (September 2021 to March 2026). The study protocol was approved by the Ethics Committee of Bethune International Peace Hospital (2016-KY-020) and Beijing Friendship Hospital, Capital Medical University (2021-P2-239-01). Identical standardized diagnostic procedures, questionnaires, interviewer training, and data collection protocols were applied at both centers to ensure consistency in study procedures and data comparability. All participants underwent simultaneous assessment based on Rome III and Rome IV diagnostic criteria on the day of enrollment. The study was conducted in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology statement (Supplementary File 1). All participants were rigorously selected according to inclusion and exclusion criteria.

Study population

A total of 468 patients aged 18–75 years who met the Rome III and/or Rome IV criteria for IBS were enrolled [8,14]. Organic diseases were ruled out through detailed clinical history, physical examination, and negative auxiliary examination. Further details are provided in the Supplementary Methods (Supplementary File 2). 49 patients were excluded due to severe missing questionnaire data that could not be reasonably imputed to meet the requirements of statistical analysis. Ultimately, 419 patients were included in the analysis. Based on fulfilment of the Rome III and Rome IV diagnostic criteria, all included patients were classified into three diagnostic categories: Rome III only, Rome IV only, and both Rome III and Rome IV. This classification was used to describe the overlap between the two diagnostic criteria and the distribution of patients across different diagnostic categories. In subsequent analyses, patients were further categorized according to Rome IV diagnostic status into Rome IV criteria-eligible and Rome IV criteria-ineligible groups to evaluate the association between Rome IV classification and disease burden. The enrolment, exclusion, and grouping process is shown in Figure 1.

Data collection and assessment tools

All enrolled IBS patients underwent face-to-face interviews conducted by experienced clinicians with expertise in the diagnosis and management of IBS. Detailed information was collected, including demographic characteristics, overlapping upper gastrointestinal symptoms, IBS symptoms, somatization symptoms, anxiety and depressive symptoms, and quality of life. Demographic characteristics included gender, age, body mass index (BMI), marital status, education level, occupation, smoking status, and alcohol status.

Overlapping upper gastrointestinal symptoms were defined as the presence of one or more upper gastrointestinal symptoms occurring at least once per week during the past three months in patients with IBS. These symptoms included acid regurgitation, heartburn, belching, globus sensation, dysphagia, retrosternal pain, anorexia, food regurgitation, rumination, nausea, vomiting, early satiety, postprandial fullness, and epigastric pain [15]. Abdominal pain, bloating, and discomfort were considered IBS-related abdominal symptoms, and were not included in the definition of overlapping upper gastrointestinal symptoms. The proportion of days with IBS-related abdominal symptoms was calculated as the number of days with relevant symptoms during the past 100 days divided by the total number of days (100 days). IBS-related defecatory symptoms included urgent, difficult, and incomplete defecation.

Stool consistency was evaluated using the Bristol Stool Form Scale (BSFS) (Figure S1) [16]. The severity of IBS-related abdominal and defecatory symptoms was assessed using a patient-reported scoring system. IBS symptom severity was evaluated using the Irritable Bowel Syndrome Severity Scoring System (IBS-SSS) [11]. Somatization symptoms were assessed using the 12-item Patient Health Questionnaire (PHQ-12), while anxiety and depressive symptoms were evaluated using the Self-Rating Anxiety Scale (SAS) and Self-Rating Depression Scale (SDS), respectively [12,17,18]. Health-related quality of life was assessed using the 36-Item Short Form Health Survey (SF-36) [19–21]. Detailed criteria can be found in the Supplementary File 2.

LCA of IBS burden phenotypes

LCA is a clustering approach based on finite mixture models and probability theory [22]. It identifies unobserved heterogeneity among individuals by examining patterns of observed variables and classifies individuals with similar characteristics into distinct latent classes. This approach enables detailed phenotypic stratification of study populations. Compared with conventional clustering methods, LCA does not require predefined assumptions about data distributions or subjective distance measures. It can simultaneously handle different types of variables and determine the optimal number of classes using model fit indices, including the log-likelihood, Akaike information criterion (AIC), and Bayesian information criterion (BIC). In addition, LCA estimates the posterior probability of class membership for each individual, improving the statistical robustness and clinical interpretability of the identified phenotypes. LCA has been widely applied in gastrointestinal disorders and psychiatric conditions to identify disease heterogeneity, define clinical phenotypes, and explore personalized treatment strategies [10,23].

The poLCA package in R software was used to construct latent class models with 2–6 classes, using the severity of IBS symptoms, somatization symptoms, and anxiety and depression symptoms as observed variables in this study. Health-related quality of life was used as an independent disease burden indicator to validate differences between latent classes. The optimal number of classes was determined based on model fit indices, clinical interpretability, and model parsimony.

Machine learning-based construction of the IBS-BI

Least absolute shrinkage and selection operator (LASSO) regression and random forest regression were jointly applied to develop the IBS-BI. LASSO regression is a linear regression model based on L1 regularization [24]. By introducing a penalty term, LASSO shrinks some coefficients to zero, enabling feature selection and reducing the risk of overfitting. Random forest regression is an ensemble machine learning method based on multiple decision trees [25]. It can capture potential nonlinear relationships and complex interactions among variables. As IBS disease burden involves multiple dimensions, including gastrointestinal symptoms, psychological status, and quality of life, complex nonlinear associations may exist among these factors. Therefore, random forest regression was selected over traditional linear regression to construct a composite disease burden index.

The IBS-BI was developed based on multidimensional assessments from the IBS-SSS, PHQ-12, SAS, SDS, and SF-36. SF-36 scores were reverse-coded. A higher positivized SF-36 value indicates a worse outcome. All candidate variables were then standardized using Z-score transformation to eliminate differences in measurement scales. Subsequently, LASSO regression was performed using the glmnet package in R to select key features for model development. The study population was randomly divided into a training set and an internal validation set at a 7:3 ratio. A random forest regression model was developed based on the features selected by LASSO regression using the randomForest package in R (ntree = 500). The variable importance was calculated using the built-in importance() function from the randomForest package. The feature importance score was based on the increase in node purity (IncNodePurity), which reflects the contribution of each variable to improving node purity across all trees in the random forest model. The relative weight proportion of each feature was subsequently calculated by normalizing the IncNodePurity values according to the following formula: Relative contribution (%) = IncNodePurityi/∑IncNodePurityi×100%. The model-predicted values were used as the initial IBS-BI scores. These scores were then normalized using min-max scaling to generate standardized IBS-BI scores ranging from 0 to 1. Model performance was evaluated using the coefficient of determination (R2), mean squared error (MSE), and mean absolute error (MAE). Only internal validation was performed in this study, and no independent external cohort validation was conducted.

Based on the distribution of standardized IBS-BI scores in the study population, the 33rd (P33) and 66th (P66) percentiles were used as cut-off values to classify patients into mild, moderate, and severe IBS-BI severity groups. Since no validation was performed using independent cohorts, these severity subgroups are regarded as preliminary statistical classifications. In addition, an interactive clinical calculator was developed using the Shiny package in R to enable individualized calculation of IBS-BI scores.

Statistical analysis

All statistical analyses were conducted using R software (Version 4.5.3, R Core Team, R Foundation for Statistical Computing, Vienna, Austria), with results validated using IBM SPSS Statistics 26.0 (IBM Corp., Armonk, NY, USA).

Continuous variables were assessed for normality and homogeneity of variance. Normally distributed variables with equal variances were presented as mean ± standard deviation (SD). Comparisons between two groups were performed using the independent-samples Student’s t-test, while comparisons among multiple groups were performed using one-way analysis of variance (ANOVA) followed by Tukey’s honestly significant difference (HSD) test for post-hoc pairwise comparisons. For normally distributed variables with unequal variances, Welch’s correction was applied. Non-normally distributed continuous variables were presented as median (P25, P75). Comparisons between two groups were performed using the Mann-Whitney U test, and comparisons among multiple groups were performed using the Kruskal-Wallis H test followed by Dunn (Bonferroni) test for post-hoc pairwise comparisons. Categorical variables were expressed as numbers and percentages (n, %). Chi-square tests were used for comparisons of nominal categorical variables. Ordinal categorical variables were compared using the Mann-Whitney U test for two groups and the Kruskal-Wallis H test for multiple groups. All statistical tests were two-sided, and p <0.05 was considered statistically significant.

Results

Baseline characteristics and diagnostic classification of IBS patients

Among the 468 patients initially diagnosed with IBS according to the Rome III and/or Rome IV criteria, 49 patients were excluded due to substantial missing questionnaire data. Finally, 419 patients were included in the analysis. Based on the original diagnostic classifications, patients were categorized into three diagnostic categories: Rome III only (n = 199), Rome IV only (n = 2), and both Rome III and Rome IV (n = 218). Only two patients met the Rome IV criteria but not the Rome III criteria, whereas most patients meeting the Rome IV criteria also met the Rome III criteria. The demographic and clinical characteristics of these diagnostic categories are presented in Table 1. Statistical comparisons were not performed because the Rome IV-only category included only two patients.

Table 1.Distribution and baseline characteristics of IBS patients according to Rome III and Rome IV diagnostic classifications.

Characteristics Rome III only(n = 199) Rome IV only(n = 2) Both Rome III and IV(n = 218)

Gender, n (%)      

Male 104 (52.26) 1 (50.00) 135 (61.93)

Female 95 (47.74) 1 (50.00) 83 (38.07)

Average age, years 44.73 58.00 42.49

Average BMI, kg/m² 23.79 23.40 23.68

Marital status, n (%)      

Unmarried 28 (14.07) 0 (0.00) 44 (20.18)

Married 159 (79.90) 2 (100.00) 161 (73.85)

Divorced/Widowed 12 (6.03) 0 (0.00) 13 (5.96)

Education level, n (%)      

Primary school or below 30 (15.08) 0 (0.00) 19 (8.72)

Junior high school 40 (20.10) 1 (50.00) 39 (17.89)

Senior high school 37 (18.59) 0 (0.00) 32 (14.68)

Undergraduate 80 (40.20) 1 (50.00) 102 (46.79)

Postgraduate 12 (6.03) 0 (0.00) 26 (11.93)

Occupation, n (%)      

Manual 38 (19.10) 1 (50.00) 38 (17.43)

Non-manual 133 (66.83) 0 (0.00) 156 (71.56)

Not working 28 (14.07) 1 (50.00) 24 (11.01)

Smoking status, n (%) 67 (33.67) 0 (0.00) 54 (24.77)

Alcohol status, n (%) 49 (24.62) 0 (0.00) 42 (19.27)

IBS subtype, n (%)      

IBS-C 31 (15.58) 0 (0.00) 17 (7.80)

IBS-D 135 (67.84) 2 (100.00) 174 (79.82)

IBS-M 8 (4.02) 0 (0.00) 8 (3.67)

IBS-U 25 (12.56) 0 (0.00) 19 (8.72)

Abbreviations: BMI, body mass index; IBS, irritable bowel syndrome; IBS-C, IBS with predominant constipation; IBS-D, IBS with predominant diarrhoea; IBS-M, IBS with mixed bowel habits; IBS-U, IBS unclassified.

For subsequent analyses, patients were further categorized according to Rome IV diagnostic status into Rome IV criteria-eligible (n = 220) and Rome IV criteria-ineligible (n = 199) groups. The corresponding demographic and disease characteristics are shown in Table S1.

Comparison of overlapping upper gastrointestinal symptoms according to Rome IV diagnostic status

Regarding overlapping upper gastrointestinal symptoms, anorexia was less common in the Rome IV criteria-eligible group than in the Rome IV criteria-ineligible group (10.00% vs 34.67%, p <0.001) (Table S2). No significant differences were observed between the two groups for the remaining individual upper gastrointestinal symptoms and the total number of overlapping upper gastrointestinal symptoms (all p > 0.05).

Comparison of IBS symptoms according to Rome IV diagnostic status

The IBS-SSS score was significantly higher in the Rome IV criteria-eligible group than in the Rome IV criteria-ineligible group [222.03 (177.38, 283.34) vs 116.66 (75.00, 150.00), p <0.001] (Figure 2a; Table 2). According to IBS-SSS severity categories, only one patient (0.45%) in the Rome IV criteria-eligible group had very mild or almost normal symptoms, whereas 50 (22.73%), 125 (56.82%), and 44 (20.00%) patients had mild, moderate, and severe symptoms, respectively. In contrast, 60 (30.15%), 113 (56.78%), 24 (12.06%), and 2 (1.01%) patients in the Rome IV criteria-ineligible group had very mild or almost normal symptoms, mild symptoms, moderate symptoms, and severe symptoms, respectively. The distribution of IBS symptom severity differed significantly between the two groups (p <0.001).

Table 2.Comparison of IBS symptoms, somatization symptoms, anxiety symptoms, depression symptoms according to Rome IV diagnostic status among IBS patients.

Characteristics Rome IV criteria-eligible group(n = 220) Rome IV criteria-ineligible group(n = 199) P value

IBS-SSS score[Median (P25, P75)] 222.03 (177.38, 283.34) 116.66 (75.00, 150.00) <0.001a

IBS symptoms, n (%)      

Very mild/Almost normal 1 (0.45) 60 (30.15) <0.001a

Mild 50 (22.73) 113 (56.78)

Moderate 125 (56.82) 24 (12.06)

Severe 44 (20.00) 2 (1.01)

PHQ-12 score[Median (P25, P75)] 5.00 (2.00, 9.00) 3.00 (1.00, 6.00) <0.001a

Somatization symptoms, n (%)      

None 84 (38.18) 101 (50.75) 0.001a

Mild 65 (29.55) 66 (33.17)

Moderate 51 (23.18) 20 (10.05)

Severe 20 (9.09) 12 (6.03)

SAS score[Median (P25, P75)] 45.50 (40.00, 51.00) 43.00 (37.00, 48.00) 0.035a

Anxiety symptoms, n (%)      

None 146 (66.36) 150 (75.38) 0.061a

Mild 59 (26.82) 36 (18.09)

Moderate 13 (5.91) 10 (5.03)

Severe 2 (0.91) 3 (1.51)

SDS score[Median (P25, P75)] 50.00 (40.00, 58.00) 46.00 (35.00, 57.00) 0.013a

Depression symptoms, n (%)      

None 124 (56.36) 136 (68.34) 0.002a

Mild 63 (28.64) 42 (21.11)

Moderate 27 (12.27) 17 (8.54)

Severe 6 (2.73) 4 (2.01)

aMann-Whitney U test.

Abbreviations: IBS, irritable bowel syndrome; IBS-SSS, IBS-Severity Scoring System; PHQ-12, Patient Health Questionnaire-12; SAS, Self-Rating Anxiety Scale; SDS, Self-Rating Depression Scale.

Specifically, the comparison of IBS-related abdominal symptoms is presented in Table S3. Significant differences were observed between the Rome IV criteria-eligible and Rome IV criteria-ineligible groups in abdominal pain (p <0.001) and abdominal discomfort (p <0.001). All patients in the Rome IV criteria-eligible group reported abdominal pain, whereas only 22 patients (11.06%) in the Rome IV criteria-ineligible group reported this symptom. In contrast, abdominal discomfort was more common in the Rome IV criteria-ineligible group than in the Rome IV criteria-eligible group (71.86% vs 10.00%). No significant difference was found in bloating between the two groups (p = 0.139). In addition, the proportion of days with IBS-related abdominal symptoms was significantly higher in the Rome IV criteria-eligible group than in the Rome IV criteria-ineligible group (60.00% vs 43.00%, p <0.001). IBS-related defecatory symptoms are summarized in Table S4. Compared with the Rome IV criteria-ineligible group, patients in the Rome IV criteria-eligible group more frequently reported severe urgent defecation (p <0.001) and incomplete defecation (p = 0.008). No significant difference was observed in difficult defecation between the two groups (p = 0.668).

Comparison of somatization, anxiety, and depression symptoms according to Rome IV diagnostic status

As for somatization symptoms, the median PHQ-12 score was significantly higher in the Rome IV criteria-eligible group than in the Rome IV criteria-ineligible group [5.00 (2.00, 9.00) vs 3.00 (1.00, 6.00), p <0.001] (Figure 2b; Table 2). Based on PHQ-12 severity categories, 84 patients (38.18%) in the Rome IV criteria-eligible group had no somatization symptoms, while 65 (29.55%), 51 (23.18%), and 20 (9.09%) patients had mild, moderate, and severe symptoms, respectively. In comparison, the Rome IV criteria-ineligible group had a higher proportion of patients without somatization symptoms (50.75%, n = 101), with a significant difference in symptom distribution between the two groups (p = 0.001).

The median SAS and SDS scores were also significantly higher in the Rome IV criteria-eligible group than in the Rome IV criteria-ineligible group [SAS: 45.50 (40.00, 51.00) vs 43.00 (37.00, 48.00), p = 0.035; SDS: 50.00 (40.00, 58.00) vs 46.00 (35.00, 57.00), p = 0.013] (Figure 2c, 2d; Table 2). Stratified analysis based on SDS severity showed a significant difference in depression symptom distribution between the two groups (p = 0.002). Although anxiety severity showed a similar trend, no statistically significant difference was observed based on SAS classification (p = 0.061).

Comparison of health-related quality of life according to Rome IV diagnostic status

The median SF-36 score was significantly lower in the Rome IV criteria-eligible group than in the Rome IV criteria-ineligible group [68.05 (56.90, 78.10) vs 73.50 (62.90, 82.90), p = 0.033] (Figure 2e; Table 3). The median physical component summary score was significantly lower in the Rome IV criteria-eligible group than in the Rome IV criteria-ineligible group (p = 0.001), whereas no significant difference was observed in the median mental component summary score (70.84 vs 72.92, p = 0.525). Among the eight SF-36 domains, the Rome IV criteria-eligible group had significantly lower scores for physical functioning (p <0.001) and bodily pain (p <0.001). No significant differences were found in role physical, general health, vitality, social functioning, role emotional, or mental health between the two groups (all p > 0.05) (Figure S2).

Table 3.Comparison of health-related quality of life according to Rome IV diagnostic status among IBS patients.

Characteristics Rome IV criteria-eligible group(n = 220) Rome IV criteria-ineligible group(n = 199) P value

SF-36 score[Median (P25, P75)] 68.05 (53.82, 83.63) 73.50 (60.42, 84.25) 0.033a

Physical component summary [Median (P25, P75)] 69.50 (51.56, 81.25) 75.50 (64.75, 84.75) 0.001a

Physical functioning[Median (P25, P75)] 95.00 (85.00, 100.00) 100.00 (90.00, 100.00) <0.001a

Role physical[Median (P25, P75)] 75.00 (25.00, 100.00) 100.00 (50.00, 100.00) 0.105a

Bodily pain[Median (P25, P75)] 66.00 (52.50, 84.00) 84.00 (64.00, 100.00) <0.001a

General health[Median (P25, P75)] 45.00 (30.00, 59.25) 47.00 (35.00, 62.00) 0.105a

Mental component summary[Median (P25, P75)] 70.84 (53.55, 86.00) 72.92 (52.96, 86.75) 0.525a

Vitality[Median (P25, P75)] 65.00 (50.00, 78.75) 65.00 (55.00, 80.00) 0.449a

Social functioning[Median (P25, P75)] 87.50 (62.50, 100.00) 75.00 (62.50, 100.00) 0.865a

Role emotional[Median (P25, P75)] 100.00 (33.33, 100.00) 100.00 (33.33, 100.00) 0.403a

Mental health[Median (P25, P75)] 64.00 (52.00, 80.00) 64.00 (48.00, 76.00) 0.445a

aMann-Whitney U test.

Abbreviations: IBS, irritable bowel syndrome; SF-36, 36-Item Short Form Health Survey.

Independent factors associated with health-related quality of life in IBS patients

As shown in Table 4, the multivariable linear regression model was statistically significant (F = 19.342, p <0.001), with an adjusted R2 of 0.401, explaining 40.1% of the variance in SF-36 scores. The Durbin-Watson statistic was 1.979, and all variance inflation factor values were below 10, indicating no significant autocorrelation or severe multicollinearity. Higher IBS-SSS (p <0.001), PHQ-12 (p <0.001), SAS (p = 0.002), and SDS (p <0.001) scores were independently associated with lower SF-36 scores. Alcohol consumption was independently associated with higher SF-36 scores (p = 0.014). No other evaluated factors showed significant associations with health-related quality of life (all p > 0.05).

Table 4.Multivariable linear regression analysis of factors associated with health-related quality of life in IBS patients.

Characteristics Unstandardized β 95% CI for β Standardized β t value P value VIF

(Constant) 114.996 100.245 to 129.746 – 15.326 <0.001 –

Gender            

Male Reference          

Female −0.278 −3.458 to 2.902 −0.008 −0.172 0.864 1.322

Age −0.062 −0.194 to 0.069 −0.047 −0.931 0.352 1.736

BMI −0.063 −0.456 to 0.331 −0.013 −0.313 0.755 1.095

Marital status            

Unmarried Reference          

Married 3.107 −1.180 to 7.394 0.072 1.425 0.155 1.747

Divorced/Widowed 6.321 −1.211 to 13.853 0.083 1.650 0.100 1.701

Education level 0.471 −0.981 to 1.922 0.031 0.638 0.524 1.599

Occupation            

Not working Reference          

Manual −3.638 −9.066 to 1.791 −0.078 −1.317 0.188 2.362

Non-manual −4.483 −9.000 to 0.035 −0.115 −1.951 0.052 2.333

Smoking status −2.634 −6.207 to 0.938 −0.066 −1.449 0.148 1.401

Alcohol status 4.749 0.974 to 8.524 0.108 2.473 0.014 1.294

IBS-SSS score −0.033 −0.050 to −0.016 −0.163 −3.783 <0.001 1.259

PHQ-12 score −0.998 −1.404 to −0.591 −0.240 −4.826 <0.001 1.671

SAS score −0.339 −0.549 to −0.130 −0.178 −3.182 0.002 2.115

SDS score −0.339 −0.488 to −0.189 −0.237 −4.449 <0.001 1.913

Abbreviations: BMI, body mass index; CI, confidence interval; IBS, irritable bowel syndrome; IBS-SSS, IBS-Severity Scoring System; PHQ-12, Patient Health Questionnaire-12; SAS, Self-Rating Anxiety Scale; SDS, Self-Rating Depression Scale; VIF, variance inflation factor.

Distinct IBS burden phenotypes identified by LCA

Based on the differences observed between the Rome IV criteria-eligible and Rome IV criteria-ineligible groups and the results of multivariable linear regression, LCA was performed using IBS symptoms, somatization symptoms, anxiety symptoms, and depression symptoms as indicator variables. Health-related quality of life was used as a validation measure to assess differences among latent classes. Models containing 2–6 latent classes were constructed, and model fit was evaluated using the log-likelihood, AIC, and BIC (Table S5). Considering model fit, clinical interpretability, and parsimony, the four-class model was selected as the optimal solution.

The characteristics of the four latent classes are presented in Figure 3a and Table 5. Class 1 was identified as the high-symptom burden phenotype, characterized by the highest levels of IBS, somatization, anxiety, and depression symptoms, together with the poorest health-related quality of life. Traditional IBS subtypes were widely distributed in this phenotype, with higher proportions of IBS-C, IBS-D, IBS-M, and IBS-U compared with other classes (Figure 3b). Class 2 represented the low-symptom burden phenotype, characterized by relatively low symptom severity across all domains and the highest health-related quality of life among the four classes. IBS-D was the predominant traditional subtype in this class, whereas other subtypes were uncommon. Class 3 was characterized as the somatization-dominant phenotype, with moderate IBS symptoms, prominent somatization symptoms, relatively low anxiety and depression symptoms, and intermediate health-related quality of life. IBS-D was also the most common subtype in this phenotype. Class 4 was identified as the psychological comorbidity phenotype, characterized by mild IBS symptoms, moderate somatization symptoms, and marked anxiety and depression symptoms, accompanied by reduced health-related quality of life. Most patients in this phenotype had IBS-D, while IBS-C was not observed.

Table 5.Characteristics of the four latent class-derived IBS burden phenotypes.

Characteristics aClass 1(n = 42) bClass 2(n = 216) cClass 3(n = 83) dClass 4(n = 78)

Class proportions, % 10.02 51.55 19.81 18.62

Severity of IBS symptoms,Mean ± SD 3.24 ± 0.53 2.07 ± 0.74 2.92 ± 0.67 2.47 ± 0.99

Severity of somatization symptoms,Mean ± SD 3.17 ± 1.06 1.23 ± 0.48 2.46 ± 0.59 2.37 ± 0.77

Severity of anxiety symptoms,Mean ± SD 2.52 ± 0.67 1.02 ± 0.19 1.04 ± 0.33 2.09 ± 0.29

Severity of depression symptoms,Mean ± SD 2.57 ± 0.91 1.23 ± 0.55 1.41 ± 0.63 1.94 ± 0.74

Health-related quality of life,Mean ± SD 49.01 ± 19.80 76.73 ± 13.57 65.28 ± 15.45 58.84 ± 17.29

IBS subtype, n (%)        

IBS-C 14 (33.33) 22 (10.19) 12 (14.46) 0 (0.00)

IBS-D 19 (45.24) 160 (74.07) 64 (77.11) 68 (87.18)

IBS-M 4 (9.52) 7 (3.24) 4 (4.82) 1 (1.28)

IBS-U 5 (11.90) 27 (12.50) 3 (3.61) 9 (11.54)

aClass 1, high-symptom burden phenotype.

bClass 2, low-symptom burden phenotype.

cClass 3, somatization-dominant phenotype.

dClass 4, psychological comorbidity phenotype.

Abbreviations: IBS, irritable bowel syndrome; IBS-C, IBS with predominant constipation; IBS-D, IBS with predominant diarrhoea; IBS-M, IBS with mixed bowel habits; IBS-U, IBS unclassified; SD, standard deviation.

Significant differences in health-related quality of life were observed among the four phenotypes (p <0.05), with post-hoc comparisons showing significant differences between all class pairs (all p <0.05) (Figure 3c; Table S6). In addition, the four phenotypes differed significantly in IBS symptoms, somatization symptoms, anxiety symptoms, and depression symptoms (all p <0.001).

Development and evaluation of the IBS-BI and its clinical calculator

The IBS-BI was developed based on IBS symptoms, somatization symptoms, anxiety symptoms, depression symptoms, and health-related quality of life. The optimal penalty coefficient for the LASSO model was λ = 0.07946 after 10-fold cross-validation (Figure S3). IBS-SSS, PHQ-12, SAS, SDS, and positivized SF-36 standardized scores were retained as the five core features for IBS-BI construction (Table S7).

The study population was randomly divided into a training set (n = 295, 70.41%) and an internal validation set (n = 124, 29.59%) at a 7:3 ratio. No significant differences were observed in core scale scores or baseline characteristics between the two sets (all p > 0.05), indicating good baseline comparability (Table S8).

A random forest regression model was subsequently developed using the five selected features. Feature importance analysis demonstrated different contributions of individual variables to model performance. Based on IncNodePurity, the positivized SF-36 score showed the highest importance, whereas PHQ-12 contributed most to predictive accuracy according to percentage increase in MSE (%IncMSE) (Figure S3). The relative contribution weights calculated from IncNodePurity ranked the five features as follows: positivized SF-36 (22.5%), SAS (20.8%), PHQ-12 (20.6%), SDS (20.1%), and IBS-SSS (16.0%) (Table 6). Model performance evaluation showed that the random forest model achieved R2, MSE, and MAE values of 0.995, 0.145, and 0.246 in the training set, and 0.984, 0.470, and 0.442 in the internal validation set, respectively, indicating good model fit and internal stability (Figure 4a). Pearson correlation analysis revealed significant positive correlations between IBS-BI and all core features (all p <0.001), with the strongest correlations observed for SAS (r = 0.796) and PHQ-12 (r = 0.781), followed by positivized SF-36 (r = 0.767), SDS (r = 0.756), and IBS-SSS (r = 0.616) (Figure 4b). Further analysis demonstrated significant variation in IBS-BI standardized scores across the four LCA-derived phenotypes (p <0.001) (Figure 4c; Table S9). Post-hoc testing showed that most phenotype pairs differed significantly, except for Class 1 versus Class 4 (p = 0.113). Specifically, significant differences were observed between Class 1 and Class 2, Class 1 and Class 3, Class 2 and Class 3, Class 2 and Class 4 (all p <0.001), and Class 3 and Class 4 (p = 0.009).

Table 6.Feature importance and relative contribution of variables in the random forest regression model.

Characteristics Feature importance score Weight proportion (%)

IBS-SSS standardized score 650.799 16.0

PHQ-12 standardized score 837.704 20.6

SAS standardized score 847.755 20.8

SDS standardized score 820.636 20.1

Positivized SF-36 standardized score 918.840 22.5

Abbreviations: IBS, irritable bowel syndrome; IBS-SSS, IBS-Severity Scoring System; PHQ-12, Patient Health Questionnaire-12; SAS, Self-Rating Anxiety Scale; SDS, Self-Rating Depression Scale; 36-Item Short Form Health Survey.

Based on the distribution of IBS-BI standardized scores, the P33 and P66 cut-off values were 0.233 and 0.422, respectively, classifying the 419 patients into mild (< 0.233) (n = 138, 32.94%), moderate (≥ 0.233 and < 0.422) (n = 138, 32.94%), and severe (≥ 0.422) (n = 143, 34.13%) severity groups, which were interpreted as preliminary statistical classifications until validated in independent cohorts. The distribution of patients across the three groups was comparable. Significant differences were observed among the three groups in IBS-SSS, PHQ-12, SAS, SDS, and positivized SF-36 standardized scores (all p <0.001), with increasing IBS-BI levels associated with greater IBS symptoms, somatization symptoms, anxiety symptoms and depression symptoms, and poorer health-related quality of life (Figure 4d–h). Post-hoc analyses confirmed significant differences between all pairwise comparisons (all p <0.001) (Table S10).

Furthermore, an interactive clinical calculator was developed using the R Shiny package (https://wyb000926.shinyapps.io/ibs_bi_web/), strictly adhering to the programming environment standards used in the model construction, ensuring that the development environment is fully consistent with the modeling environment. All core model parameters and indicator transformation rules were pre-locked, ensuring the stability, consistency, and reproducibility of the output results from a technical perspective, thus meeting the reproducibility requirements for clinical research. This calculator incorporates the final random forest regression model established in this study and can completely reproduce the process from inputting raw scale scores to outputting the IBS-BI standardized score, automating and standardizing both data processing and result output. The interface design adopts a user-friendly dual-column layout (Figure S4). The left input module clearly labels the valid scoring ranges for each scale (IBS-SSS, PHQ-12, SAS, SDS, SF-36), effectively preventing invalid inputs and improving operational convenience. The right output module provides real-time feedback on the IBS-BI standardized score and the corresponding disease burden stratification results after clicking the Calculate button. This calculator requires no installation of specialized statistical software and can be accessed directly via mainstream web browsers, compatible with multiple devices such as computers and tablets.

Discussion

In this multicentre cross-sectional study, we systematically compared IBS patients who met and did not meet the Rome IV diagnostic criteria in terms of demographic characteristics, overlapping upper gastrointestinal symptoms, IBS symptoms, somatization symptoms, anxiety symptoms, depression symptoms, and health-related quality of life. Based on multidimensional clinical features, we identified four distinct IBS phenotypes with different clinical characteristics. Furthermore, we developed and internally validated the IBS-BI as a quantitative tool for assessing overall disease burden in IBS patients and established an interactive web-based calculator to facilitate its potential application in clinical assessment and future research. However, as IBS-BI has currently undergone only internal validation, its generalizability and clinical utility require further confirmation in independent external cohorts.

This study found that patients fulfilling the Rome IV criteria had greater symptom burden, more prominent psychological symptoms, and poorer health-related quality of life, which is consistent with previous studies [26–29]. However, these findings should be interpreted in the context of the diagnostic characteristics of the Rome IV criteria. The Rome IV criteria require the presence of recurrent abdominal pain associated with defecation, occurring at least once per week during the previous 3 months [8,14,30]. The observed differences in disease burden may also largely reflect the higher symptom threshold of the Rome IV criteria rather than indicating an inherent superiority of Rome IV criteria in identifying patients with greater disease burden. In addition, patients meeting the Rome IV criteria showed several demographic differences, including a higher proportion of males, higher educational attainment, and lower smoking rates. These findings are partly consistent with previous studies [31,32]. Such differences may be related to the stricter symptom requirements of the Rome IV criteria. The higher diagnostic threshold may preferentially identify individuals with greater symptom awareness, higher health literacy, and a stronger tendency to seek medical care. The proportion of IBS-D was higher among patients fulfilling the Rome IV criteria than among those who did not (80.00% vs 67.84%). This finding is consistent with a previous multinational study conducted by Oka et al. involving 34 countries [33]. One possible explanation is that patients with IBS-D may have a stronger association between abdominal pain and bowel movements, making them more likely to meet the core diagnostic requirement of the Rome IV criteria.

Meanwhile, multivariable linear regression analysis further demonstrated that greater severity of IBS symptoms, somatization symptoms, anxiety, and depression were independently associated with poorer health-related quality of life. This finding is in line with the study by Trindade et al. using structural equation modeling, further supporting the concept that IBS is a multidimensional disorder involving gastrointestinal symptoms, somatic symptoms, and psychological factors [34]. In clinical practice, management strategies should not only focus on improving gastrointestinal symptoms but also address somatization symptoms and psychological comorbidities. Notably, alcohol-consuming patients showed better health-related quality of life in this study. However, this finding should be interpreted with caution. In this study, alcohol use was assessed only as a binary variable (yes/no), without further evaluation of consumption amount, frequency, or drinking patterns. Therefore, the observed association does not indicate a causal relationship or a beneficial effect of alcohol consumption. In addition, alcohol use may be influenced by potential confounding factors, including social participation, occupational status, age, and disease severity. For example, patients with milder symptoms and better social functioning may be more likely to participate in social activities involving alcohol consumption, which could partially explain this association. Further studies with detailed assessment of alcohol exposure and comprehensive adjustment for potential confounders are needed to clarify this relationship.

Based on the previous study by Han et al. which identified IBS subtypes using LCA, we further incorporated somatization symptoms and anxiety/depression-related features to capture a broader spectrum of clinical characteristics [9]. Four clinically distinct IBS phenotypes were identified: high-symptom burden phenotype, low-symptom burden phenotype, somatization-dominant phenotype, and psychological comorbidity phenotype. Compared with the seven-category classification proposed by Black et al. in 2023, the major contribution of our study was the inclusion of somatization features [10]. Cluster analysis has become one of the most widely used approaches for identifying novel IBS classifications and phenotypes. Both k-means clustering and LCA have been frequently applied in previous studies. To date, more than 10 studies have used clustering methods to define IBS subtypes [35]. Despite differences in classification strategies and selected variables, most studies have consistently shown that a large proportion of IBS patients have relatively mild symptom severity and better quality of life, which is also consistent with our findings.

The recently proposed Rome V criteria further emphasize that IBS is a biopsychosocial disorder influenced by gastrointestinal symptoms, somatization symptoms, psychological factors, and health-related quality of life [36]. Therefore, comprehensive management strategies may be required for IBS patients. However, individual indicators usually reflect only specific aspects of the disease and cannot fully capture the overall disease burden. For example, some patients may have high IBS-SSS scores but mild psychological symptoms and limited impairment in quality of life. In contrast, other patients may present with moderate gastrointestinal symptoms accompanied by significant anxiety, depression, and reduced quality of life. Relying solely on symptom severity may underestimate the overall burden and management needs of the latter group. In clinical practice, IBS patients often undergo multiple assessments, including IBS-SSS, PHQ-12, SAS, SDS, and SF-36. However, these scales provide fragmented information, and the overall assessment of disease status may largely depend on clinicians’ experience, introducing potential subjectivity. By integrating multiple clinical dimensions, IBS-BI may provide a unified measure of disease burden and may facilitate the identification of patients with different levels of overall disease impact. Furthermore, IBS-BI was developed using a random forest algorithm, which allows the model to capture potential complex interactions among clinical features rather than simply combining individual scores. By providing an integrated measure, IBS-BI may help evaluate changes in overall disease burden and provide a reference for clinical research, outcome evaluation, and comparisons of different intervention strategies. For instance, a decrease in IBS-BI score from 0.72 to 0.35 after treatment may indicate an improvement in the overall disease burden. Importantly, IBS-BI was not developed to replace existing individual assessment tools, but rather to complement them. The composite index provides an overall evaluation of disease burden, while specific scales can further identify the major contributing factors. To facilitate the accessibility and practical application of IBS-BI, we further developed an interactive clinical calculator based on the R Shiny platform. This tool enables rapid calculation of IBS-BI scores and corresponding burden categories using commonly available clinical scale scores. It may provide a convenient approach for comprehensive assessment of IBS burden in clinical and research settings.

This study has several limitations. Although identical standardized diagnostic procedures, questionnaires, interviewer training, and data collection protocols were applied at both centres, the long interval between patient recruitment periods may have introduced temporal heterogeneity, including changes in clinical practice, implementation of the Rome IV criteria, treatment strategies, and the potential impact of the COVID-19 pandemic. Another limitation is that the IBS-BI model was developed and internally validated within the current cohort, without external validation. Although the model demonstrated strong internal performance, its generalizability to other populations remains uncertain. Future studies should validate the IBS-BI model in independent external cohorts from different regions and clinical settings, and further assess its calibration and clinical applicability before widespread implementation. Furthermore, the lack of longitudinal follow-up prevented evaluation of the analyses of treatment outcomes, therapeutic algorithms, or clinical utility. Finally, this study only collected information on whether patients consumed alcohol and did not assess the consumption amount, frequency, or drinking patterns. Therefore, the observed association between alcohol consumption and quality of life should be interpreted cautiously, and further studies with more detailed assessments of alcohol exposure are needed.

Conclusions

Patients with IBS who fulfilled the Rome IV criteria showed a higher disease burden, which may be related to the more stringent diagnostic requirements of these criteria. Based on multidimensional clinical characteristics, this study further identified distinct IBS phenotypes and developed the IBS-BI together with an accompanying online calculator, providing a novel methodological approach for comprehensive assessment of IBS. The IBS-BI may have potential value for evaluating disease burden and patient stratification. However, its clinical utility and broader applicability require further validation in independent external cohorts and prospective studies.

Supplementary Material

Acknowledgements

The authors thank the support of the Beijing Hospitals Authority’s Ascent Plan.

Funding Statement

This work was supported by the Beijing Hospitals Authority’s Ascent Plan (Code: DF20230102).

Ethics statement

All participants consented to participate in this study, and data collection was conducted only on the premise of the participants’ voluntary participation. The forms of informed consent included written informed consent and verbal informed consent, which were reasonably selected according to the actual situation of each participant. This was a cross-sectional study that only collected questionnaire data and routine clinical information, without any invasive procedures or additional medical interventions, thus posing minimal risks to the participants. The entire informed consent process of this study was strictly abiding by the relevant requirements of medical ethical review. For participants who were eligible to sign, written informed consent forms were signed, and the signed forms were uniformly sorted out and properly archived by the researchers for verification and traceability. For a small number of participants who were unable to independently complete the written informed consent signing process due to practical constraints such as poor eyesight, limited physical mobility, low literacy level, or inability to write their own signatures, verbal informed consent was adopted. For these participants, researchers explicitly and in detail explained the study purpose, procedures, potential impacts, and privacy protection principles face-to-face. Verbal consent was confirmed only after the participants fully understood and explicitly agreed to participate verbally, and all relevant information regarding the verbal informed consent was truthfully recorded and uniformly archived.

Disclosure statement

All authors declare no financial or non-financial competing interests.

Data availability statement

The datasets generated and/or analysed during the current study are not publicly available due to but are available from the corresponding author on reasonable request.

References

1.Ford AC, Sperber AD, Corsetti M, et al.
Irritable bowel syndrome. Lancet. 2020;396(10263):1675–1688. doi: 10.1016/S0140-6736(20)31548-8.

2.Li C, Ying Y, Zheng Y, et al.
Epidemiology of irritable bowel syndrome: a systematic review and meta-analysis. Epidemiol Public Health. 2024;2(1):1034.

3.Sperber AD, Dumitrascu D, Fukudo S, et al.
The global prevalence of IBS in adults remains elusive due to the heterogeneity of studies: a Rome Foundation working team literature review. Gut. 2017;66(6):1075–1082. doi: 10.1136/gutjnl-2015-311240.

4.Sperber AD.
Epidemiology and burden of irritable bowel syndrome: an international perspective. Gastroenterol Clin North Am. 2021;50(3):489–503. doi: 10.1016/j.gtc.2021.04.001.

5.Canavan C, West J, Card T..
Review article: the economic impact of the irritable bowel syndrome. Aliment Pharmacol Ther. 2014;40(9):1023–1034. doi: 10.1111/apt.12938.

6.Black CJ, C Ford A..
Global burden of irritable bowel syndrome: trends, predictions and risk factors. Nat Rev Gastroenterol Hepatol. 2020;17(8):473–486. doi: 10.1038/s41575-020-0286-8.

7.Blake MR, Raker JM, Whelan K..
Validity and reliability of the Bristol Stool Form Scale in healthy adults and patients with diarrhoea-predominant irritable bowel syndrome. Aliment Pharmacol Ther. 2016;44(7):693–703. doi: 10.1111/apt.13746.

8.Drossman DA.
Functional gastrointestinal disorders: history, pathophysiology, clinical features and Rome IV. Gastroenterology. 2016;150(6):1262–1279.e2. doi: 10.1053/j.gastro.2016.02.032.

9.Han CJ, Pike K, Jarrett ME, et al.
Symptom-based latent classes of persons with irritable bowel syndrome. Res Nurs Health. 2019;42(5):382–391. doi: 10.1002/nur.21974.

10.Black CJ, Ng C, Goodoory VC, et al.
Novel symptom subgroups in individuals with irritable bowel syndrome predict disease impact and burden. Clin Gastroenterol Hepatol. 2024;22(2):386–396.e10. doi: 10.1016/j.cgh.2023.02.016.

11.Francis CY, Morris J, Whorwell PJ..
The irritable bowel severity scoring system: a simple method of monitoring irritable bowel syndrome and its progress. Aliment Pharmacol Ther. 1997;11(2):395–402. doi: 10.1046/j.1365-2036.1997.142318000.x.

12.Spiller RC, Humes DJ, Campbell E, et al.
The Patient Health Questionnaire 12 Somatic Symptom scale as a predictor of symptom severity and consulting behaviour in patients with irritable bowel syndrome and symptomatic diverticular disease. Aliment Pharmacol Ther. 2010;32(6):811–820. doi: 10.1111/j.1365-2036.2010.04402.x.

13.Bijkerk CJ, de Wit NJ, Muris JW, et al.
Outcome measures in irritable bowel syndrome: comparison of psychometric and methodological characteristics. Am J Gastroenterol. 2003;98(1):122–127. doi: 10.1111/j.1572-0241.2003.07158.x.

14.Drossman DA.
The functional gastrointestinal disorders and the Rome III process. Gastroenterology. 2006;130(5):1377–1390. doi: 10.1053/j.gastro.2006.03.008.

15.Yao X, Yang YS, Cui LH, et al.
The overlap of upper functional gastrointestinal disorders with irritable bowel syndrome in Chinese outpatients: a multicenter study. J Gastroenterol Hepatol. 2016;31(9):1584–1593. doi: 10.1111/jgh.13317.

16.Lewis SJ, Heaton KW..
Stool form scale as a useful guide to intestinal transit time. Scand J Gastroenterol. 1997;32(9):920–924. doi: 10.3109/00365529709011203.

17.Zung WW.
A rating instrument for anxiety disorders. Psychosomatics. 1971;12(6):371–379. doi: 10.1016/S0033-3182(71)71479-0.

18.Zung WW.
A Self-Rating Depression Scale. Arch Gen Psychiatry. 1965;12(1):63–70. doi: 10.1001/archpsyc.1965.01720310065008.

19.Ware JE, Sherbourne CD..
The MOS 36-item short-form health survey (SF-36). I. Conceptual framework and item selection. Med Care. 1992;30(6):473–483.

20.McHorney CA, Ware JE, Raczek AE..
The MOS 36-Item Short-Form Health Survey (SF-36): II. Psychometric and clinical tests of validity in measuring physical and mental health constructs. Med Care. 1993;31(3):247–263. doi: 10.1097/00005650-199303000-00006.

21.McHorney CA, Ware JE, Lu JF, et al.
The MOS 36-item Short-Form Health Survey (SF-36): III. Tests of data quality, scaling assumptions, and reliability across diverse patient groups. Med Care. 1994;32(1):40–66. doi: 10.1097/00005650-199401000-00004.

22.Mori M, Krumholz HM, Allore HG..
Using latent class analysis to identify hidden clinical phenotypes. JAMA. 2020;324(7):700–701. doi: 10.1001/jama.2020.2278.

23.Ulbricht CM, Chrysanthopoulou SA, Levin L, et al.
The use of latent class analysis for identifying subtypes of depression: a systematic review. Psychiatry Res. 2018;266:228–246. doi: 10.1016/j.psychres.2018.03.003.

24.Friedman J, Hastie T, Tibshirani R..
Regularization paths for generalized linear models via coordinate descent. J Stat Softw. 2010;33(1):1–22. doi: 10.18637/jss.v033.i01.

25.Rigatti SJ.
Random forest. J Insur Med. 2017;47(1):31–39. doi: 10.17849/insm-47-01-31-39.1.

26.Bai T, Xia J, Jiang Y, et al.
Comparison of the Rome IV and Rome III criteria for IBS diagnosis: a cross-sectional survey. J Gastroenterol Hepatol. 2017;32(5):1018–1025. doi: 10.1111/jgh.13642.

27.Black CJ, Yiannakou Y, Houghton LA, et al.
Epidemiological, clinical, and psychological characteristics of individuals with self-reported irritable bowel syndrome based on the Rome IV vs Rome III Criteria. Clin Gastroenterol Hepatol. 2020;18(2):392–398.e2. doi: 10.1016/j.cgh.2019.05.037.

28.Black CJ, Ford AC..
Assessing the impact of changes to the Rome IV criteria for clinical practice in irritable bowel syndrome. Gastroenterology. 2022;162(6):1752–1754.e1. doi: 10.1053/j.gastro.2022.01.021.

29.Aziz I, Törnblom H, Palsson OS, et al.
How the change in IBS criteria from Rome III to Rome IV impacts on clinical characteristics and key pathophysiological factors. Am J Gastroenterol. 2018;113(7):1017–1025. doi: 10.1038/s41395-018-0074-z.

30.Corsetti M, Shin A, Lacy BE, et al.
Bowel disorders. Gastroenterology. 2026;170(6):1261–1282. doi: 10.1053/j.gastro.2026.02.003.

31.Palsson OS, Whitehead W, Törnblom H, et al.
Prevalence of Rome IV functional bowel disorders among adults in the United States, Canada, and the United Kingdom. Gastroenterology. 2020;158(5):1262–1273.e3. doi: 10.1053/j.gastro.2019.12.021.

32.AlButaysh OF, AlQuraini AA, Almukhaitah AA, et al.
Epidemiology of irritable bowel syndrome and its associated factors in Saudi undergraduate students. Saudi J Gastroenterol. 2020;26(2):89–93. doi: 10.4103/sjg.SJG_459_19.

33.Oka P, Parr H, Barberio B, et al.
Global prevalence of irritable bowel syndrome according to Rome III or IV criteria: a systematic review and meta-analysis. Lancet Gastroenterol Hepatol. 2020;5(10):908–917. doi: 10.1016/S2468-1253(20)30217-X.

34.Trindade IA, Melchior C, Törnblom H, et al.
Quality of life in irritable bowel syndrome: exploring mediating factors through structural equation modelling. J Psychosom Res. 2022;159:110809. doi: 10.1016/j.jpsychores.2022.110809.

35.Zarei D, Saghazadeh A, Rezaei N..
Subtyping irritable bowel syndrome using cluster analysis: a systematic review. BMC Bioinformatics. 2023;24(1):478. doi: 10.1186/s12859-023-05567-8.

36.Drossman DA, Chang L, Tack J..
Disorders of gut-brain interaction and the Rome V process. Gastroenterology. 2026;170(6):1083–1098. doi: 10.1053/j.gastro.2026.02.014.

Associated Data

Supplementary Materials

Data Availability Statement

The datasets generated and/or analysed during the current study are not publicly available due to but are available from the corresponding author on reasonable request.

Come citare questo studio

Yanbin Wei, Heyang Zhang, Nan Wang, Haifeng Jin, Zitan Feng, Jiapeng Guo, Jiafei Peng, Jia Feng, Jia Zhi, Shutian Zhang, Shengtao Zhu, Xin Yao. (2026). Development and internal validation of a machine learning-based disease burden index for irritable bowel syndrome: a multicentre cross-sectional study.. Annals of medicine. DOI: 10.1080/07853890.2026.2721087

Consulta la pubblicazione originale ↗