You are viewing the site in preview mode

Skip to main content
Download PDF

Machine learning for temporary stoma after intestinal resection in surgical decision-making of Crohn’s disease

BMC Gastroenterology volume 25, Article number: 117 (2025) Cite this article

Abstract

Background

Crohn’s disease (CD) often necessitates surgical intervention, with temporary stoma creation after intestinal resection (IR) being a crucial decision. This study aimed to construct novel models based on machine learning (ML) to predict temporary stoma formation after IR for CD.

Methods

Patient data who underwent IR for CD at our center between July 2017 and March 2023 were collected for inclusion in this retrospective study. Eligible CD patients were randomly divided into training and validation cohorts. Feature selection was executed using the least absolute shrinkage and selection operator. We employed three ML algorithms including traditional logistic regression, novel random forest and XG-Boost to create prediction models. The area under the curve (AUC), accuracy, sensitivity, specificity, precision, recall, and F1 score were used to evaluate these models. SHapley Additive exPlanation (SHAP) approach was used to assess feature importance.

Results

A total of 252 patients with CD were included in the study, 150 of whom underwent temporary stoma creation after IR. Eight independent predictors emerged as the most valuable features. An AUC between 0.886 and 0.998 was noted among the three ML algorithms. The random forest (RF) algorithms demonstrated the most optimal performance (0.998 in the training cohort and 0.780 in the validation cohort). By employing the SHAP method, we identified the variables that contributed to the model and their correlation with temporary stoma formation after IR for CD.

Conclusions

The proposed RF model showed a good predictive ability for identifying patients at high risk for temporary stoma formation after IR for CD, which can assist in surgical decision-making in CD management, provide personalized guidance for temporary stoma formation, and improve patient outcomes.

Peer Review reports

Background

Crohn’s disease (CD) is a chronic inflammatory bowel disorder characterized by transmural inflammation, which can lead to various complications, such as strictures, fistulas, and abscesses [1]. The incidence of CD is much higher in developed countries in North America and Europe than in Asia [2, 3]. However, over the past two decades, the incidence and prevalence of CD has increased dramatically in Asia, especially in developed regions and newly industrialized countries, including China [4,5,6]. Despite advancements in medical therapies, a significant proportion of patients with CD eventually require surgical intervention to manage disease-related complications or to improve their quality of life [7].

One crucial aspect of surgical management is the decision-making process regarding temporary stoma creation after intestinal resection (IR), which involves a second surgery for the restoration of intestinal continuity. The choice between proceeding with a primary anastomosis or adopting a two-stage approach remains a challenging clinical decision. The decision is currently usually based on the clinician’s experience, taking into account individual patient characteristics, disease severity, presence of complications, and overall surgical risk. While some patients may benefit from immediate one-stage surgery, others may require staged procedures to minimize postoperative complications and optimize long-term outcomes, although the enterostomy can adversely impact on the patient’s body image as well as lead to negative emotions and stress [8].

In this context, predictive models that aid surgical decision-making play a pivotal role in guiding clinicians towards the most appropriate treatment strategy for individual patients. Machine learning (ML), inspired by the human brain’s pattern recognition strategies, offers an artificial intelligence-based technique to tackle classification problems. This involves adjusting the weights of hidden layers during training to minimize an error function [9]. ML has shown promise in predicting clinical outcomes across various medical conditions [10,11,12]. However, few studies have attempted to identify predictors of the need for an enterostomy after IR in patients with CD [13, 14] or have only established scoring systems [15], but none have developed a risk preoperative model utilizing advanced machine algorithms for this purpose based on preoperative patient disease characteristics as well as surgical strategies. In the present study, the factors influencing temporary stoma formation for CD patients who underwent IR were explored and novel ML model predicting temporary stoma formation in surgical management of CD was established and validated.

Patients and methods

Study Population

Patients with CD who underwent IR with either primary anastomosis or temporary stoma formation at our center from July 2017 to March 2023 were included in this retrospective analysis. Patient demographic information, preoperative clinical factors, surgical outcomes, and follow-up data were extracted from electronic medical records. The inclusion criteria were: (1) 10–80 years old; (2) postoperative histopathology confirmed the diagnosis of CD; (3) there were clear surgical indications during the operation, such as stenosis, fistula, abscess, perforation, bleeding, etc., and intestinal resection was performed. Exclusion criteria were: (1) patients undergoing permanent ostomy; (2) more than 20% of the data items were missing; (3) patients with stoma status upon admission. A total of 260 patients were screened, 8 patients were excluded, 1 patient received permanent ostomy surgery, 3 patients had ileostomy status at admission, and 4 patients had no postoperative follow-up data.

Data collection

Basic information (gender, age), disease data (duration of CD, Montreal classification), preoperative general status (body mass index (BMI), weight loss in 6 months prior to surgery, history of IR for CD, preoperative medication, nutritional and inflammatory indicators in hematology) were collected. In addition, surgical data were extracted from the database, including surgical priority, surgical approach, and type of resection. Postoperative outcomes such as postoperative complications within 30 days and BMI at the time of restoration of intestinal continuity and during postoperative follow-up were also collected.

Statistical analysis

Temporary stoma is defined as a protective stoma to avoid complications such as postoperative anastomotic fistula in patients, who would receive a secondary intestinal continuity surgery after at least 3 months of postoperative recuperation time. The paired T-test was utilized to compare the difference of BMI before and after surgery in the same patient. For this study, patients were randomly allocated into a training cohort (70%) and a validation cohort (30%). Feature selection was conducted employing the least absolute shrinkage and selection operator (LASSO), a prevalent technique. LASSO limits the sum of the absolute values of the coefficients to be below a fixed threshold, setting several regression coefficients to zero, thereby producing a more concise model. LASSO skillfully manages data with intricate covariance structures, preserving only the most crucial variables.

We assessed multicollinearity among potential predictive variables, selected via a LASSO regression feature selection procedure, using the variance inflation factor (VIF). Subsequently, the final chosen variables were incorporated into the prediction models. We constructed and validated three ML algorithms, each using a different algorithm (logistic regression (LR), random forest (RF), extreme gradient boosting (XG-Boost)), with the aim of predicting temporary stoma creation in surgical management of CD. The entire process of model establishment is analyzed and processed using R software version 3.6.1. We compared three distinct ML algorithms mentioned above based on these selected predictor variables, from which the best model was chosen. We evaluated predictive accuracy using area under the curve (AUC). The validation cohort served to assess and compare the performance of each model. To ensure the reliability of the models, we utilized seven common evaluation criteria including AUC, accuracy, sensitivity, specificity, precision, recall, and F1 score. Further, Shapley Additive Explanations (SHAP) values were applied to ascertain the variable importance of each predictor and to visualize their association with temporary stoma creation in surgical management of CD.

Continuous variables were shown either as means (standard deviation), and analyzed with the two-tailed Student’s t-test or the Mann-Whitney U test respectively. Represented as counts (percentage), categorical variables were examined using either the chi-square test or Fischer’s exact test, depending on suitability. All statistical analyses were deemed significant at a P value of < 0.05. For variables with less than 5% missing data, we employed multiple imputation techniques to fill in the gaps. However, for variables with 5% or more missing data, we chose not to include them in our final analysis to ensure the reliability and validity of our results.

Results

Patient characteristics and surgical outcomes

The process of screening and analysis is shown in Fig. 1. The study cohort consisted of 252 CD patients, which were randomly assigned to the training cohort (n = 176, Tables 1 and 2) and the validation cohort (n = 76, Supplementary Tables 1, 2). Further, the patients were divided into temporary stoma formation group (Fig. 2A, B) and primary anastomosis group (Fig. 2D, E) according to whether they underwent primary anastomosis after IR. Overall, 59.5% (150/252) of patients did temporary stoma creation after IR, and the mean BMI of patients who underwent staged surgery was 18.7 ± 3.1 kg/m2, and their mean BMI raised to 20.5 ± 3.3 kg/m2 at the time of restoration of intestinal continuity after an average of 7.2 ± 4.3 months (Fig. 2C). Similarly, the mean BMI of patients who underwent primary anastomosis increased from 20.5 ± 3.7 kg/m2 to 21.4 ± 3.5 kg/m2 over an average follow-up period of 8.1 ± 6.1 months (Fig. 2F).

Fig. 1
figure 1

Flowchart of data screening and analysis. LASSO: the least absolute shrinkage and selection operator; LR: logistic regression; RF: random forest; AUC: area under the curve

Full size image
Table 1 Demographic and Crohn disease data in patients from the training cohort undergoing intestinal resection
Full size table
Table 2 Surgical data and postoperative outcome from the training cohort
Full size table
Fig. 2
figure 2

Patients underwent IR with primary anastomosis, or temporary stoma creation followed by a second surgery for the restoration of intestinal continuity. (A) The stoma position was marked before surgery for patients who plan to undergo temporary stoma after IR, and the location of the ileal stoma is usually chosen in the right lower abdomen. (B) A stoma was made at the preoperative marked position. (C) Patients with staged surgery had significantly increased BMI when restoring intestinal continuity after an average of 7.2 ± 4.3 months. (D) Ileocolic side-to-side anastomosis was performed with hand-driven stapler, followed by manual suture, after IR for CD. (E) Abdominal incision of laparoscopic IR with primary anastomosis for CD. (F) The mean BMI of patients who underwent primary anastomosis increased over an average follow-up period of 8.1 ± 6.1 months. IR: intestinal resection; CD: Crohn’s disease; BMI: body mass index

Full size image

Clinical data from the training cohort of patients in the temporary stoma formation group were compared to that of patients in the primary anastomosis group (Tables 1 and 2). The results revealed that compared with the primary anastomosis group, most patients in the temporary stoma formation group experienced weight loss > 10% in 6 months prior to surgery, and had lower BMI, preoperative albumin (ALB) and hemoglobin, while higher preoperative C-reactive protein, erythrocyte sedimentation rate (ESR), and platelets. In addition, patients in the temporary stoma formation group had higher proportions of L2 and L3 in disease location, and B2P, B3, as well as B3P in disease behavior, compared with the primary anastomosis group. Moreover, the proportion of laparotomy (including conversion) and subtotal colectomy were higher in the staged surgery group (all P < 0.05).

Postoperative complications of Clavien-Dindo ≥ III occurred in 11 cases (4.4%) in all cohorts, including fistula in 2 cases, intra-abdominal septic complications (IASC) in 4 cases and hemorrhage in 5 cases. And there was no significant correlation between the major postoperative complications and the primary anastomosis or staged surgery. Fistulas and IASCs were treated with irrigation, drainage and antibiotics. Two cases of intra-abdominal hemorrhage were treated by re-operation, two cases of intestinal hemorrhage were treated by endoscopic intervention, and one case of intestinal hemorrhage was treated with hemostatic drugs. All patients recovered and discharged.

Variable selection

In this study, LASSO regression was employed for variable screening. The optimal λ value was determined through 10-fold cross-validation. In Fig. 3A, the left and right dotted lines correspond to λ min and λ 1se, respectively. A λ 1se of 0.04259802 was selected, resulting in a model with eight non-zero coefficient variables. Figure 3B illustrates the filtering of model variables as λ values change. Compared to the full model, the final model reduces redundant variables and improves prediction accuracy. Details are provided in Table 3.

Fig. 3
figure 3

The potential risk factors were selected using the LASSO regression. (A) Trend graph of variance filter coefficients. (B) Graph of cross-validation results

Full size image
Table 3 Analysis of LASSO regression variable screening results
Full size table

Model construction and evaluation

After applying LASSO regression and accounting for multicollinearity, a total of eight indicators were identified (history of intestinal resection, weight loss > 10% in 6 months prior to surgery, disease location, disease behavior, type of resection, surgical priority, ESR, ALB), and then utilized in model construction.

In the training cohort, the AUCs of ML algorithm from high to low were: 0.998 (RF), 0.939 (XG-boost), and 0.886 (LR). In the validation cohort, the AUCs of ML algorithm from high to low were: 0.780 (RF), 0.764 (XG-boost), and 0.758 (LR). The RF algorithm was chosen as the best algorithm due to the highest AUC in both the training and validation cohorts (Fig. 4A, B).

Fig. 4
figure 4

Construction and assessment of ML model. (A-B) ROC curves for 3 ML algorithms in the training cohort and validation cohort. (C) Summary plot of SHAP values for the model constructed by RF. The vertical coordinates show the importance of the features, sorted by the importance of the variables in descending order, with the upper variables being more important to the model. For the horizontal position ‘SHAP value’ shows whether the impact of the value is associated with a higher or lower prediction. The colour of each SHAP value point indicates whether the observed value is higher (yellow) or lower (purple)

Full size image

Among the three algorithms, the algorithms ranked from highest to lowest sensitivity were 0.991 (RF), 0.849 (XG-boost), and 0.802 (LR). In terms of specificity and accuracy, the performance of the model was consistent with the sensitivity ranking, that is, the specificity from high to low was 1.000 (RF), 0.900 (XG-boost), and 0.828 (LR), and the accuracy was 0.971 (RF), 0.835 (XG-boost), and 0.789 (LR). The algorithms for F1 scoring from high to low were 0.963 (RF), 0.814 (LR), and 0.775 (XG-boost). In summary, RF algorithm exhibits superior performance in predicting temporary stoma formation in surgical management of CD (Table 4).

Table 4 Performance of the three machine learning algorithms in the training and validation cohorts for CD patients
Full size table

To better elucidate the associations depicting our prediction model, we further evaluated the RF model using SHAP. The results indicated that disease behavior made the most significant contribution to the model, followed by ESR, ALB, type of resection, weight loss > 10% in 6 months prior to surgery, disease location, history of intestinal resection, surgical priority (Fig. 4C).

Discussion

For surgery in patients with CD, a concern for both patients and doctors is whether a temporary stoma is needed. According to statistics, the 5-year incidence of ostomy in patients diagnosed with CD is about 3.5% [16]. Temporary stoma formation can help patients with CD effectively induce disease relief to improve quality of life [17], but at the same time, the presence of temporary stoma can also adversely affect the patient’s life, including body image concerns, decreased social function and low self-esteem, resulting in negative emotions such as anxiety and depression [18, 19]. The decision-making process regarding temporary stoma creation in CD presents challenges due to the lack of standardized guidelines or predictive models. Marking the location of the abdominal stoma before operation is conducive to postoperative recovery [20, 21]. Adequate preoperative information acquisition could enable informed decision-making [22]. Therefore, it is crucial to develop evidence-based tools that can aid clinicians in making informed decisions regarding temporary stoma creation, thereby improving treatment efficiency and quality of life for patients with CD.

Our findings suggested that disease behavior, primarily of B2P, B3, and B3P were associated with an increased chance of temporary stoma formation after IR for CD compared to B2. Intestinal inflammation caused by CD can invade the entire wall of the tube and, over time, can even penetrate the intestinal wall to form internal fistulas and abscesses [7]. Penetrating lesions are an independent risk factor for intra-abdominal septic complications (IASC) after CD surgery [23,24,25]. The incidence of IASC in patients with penetrating CD undergoing primary anastomosis was significantly higher than in patients undergoing staged surgery, and temporary ostomy was an independent protective factor [14]. The frequency of perianal complications in CD ranges from 17 to 43% of patients [26,27,28]. Temporary ileostomy has been well described as a treatment in perianal CD, which diverts the fecal stream from the colon, and may decrease downstream inflammation while medical therapy is optimized. However, fecal diversion without IR did not improve the outcome of perianal CD [29]. Therefore, for patients with penetrating CD or perianal CD, not only the target bowel segment should be excised, but also the possibility of a temporary stoma should be fully considered.

Our findings indicated that preoperative hypoalbuminemia and a weight loss exceeding 10% in the 6 months preceding surgery were associated with a significantly higher risk of requiring a temporary stoma. Hypoalbuminemia, a marker of malnutrition and inflammation, has been linked to increased postoperative complications [25, 30] and shown to be associated with an increased likelihood of requiring a stoma [15]. Although BMI is the most frequently measured anthropometric nutritional parameter, there are different BMI values reported for IBD patients in the literature, indicating that BMI is not a specific or sensitive marker of malnutrition in IBD patients [31]. Studies have shown that significant weight loss prior to surgery, not BMI, is a risk factor for postoperative complications in CD [13, 32, 33]. And our results further supported the likelihood of temporary stoma formation after IR in patients with a significant weight loss before surgery. Considering that the recent severe weight loss is more indicative of the patient’s preoperative nutritional status and disease severity, by combining preoperative ALB levels with a history of weight loss, a personalized treatment plan can be developed for patients with high nutritional risk, including intensive nutritional support and a two-stage surgical procedure to reduce postoperative complications, promote postoperative recovery, and improve the quality of life of patients with CD [34, 35].

CD can affect the entire gastrointestinal site, which requires different surgical approaches. In particular, the surgical management of colonic CD can be challenging due to the high risk of postoperative IASCs [36]. Notably, the effects of both disease location and type of resection on temporary stoma formation were included in the variables predicted by our model, especially colonic, and ileocolic CD, as well as the operation for colonic CD, were associated with a higher risk of stoma formation, especially after a subtotal colectomy with extensive colon lesions involvement. In the scoring system constructed by Dakshitha W et al. [15], simultaneous colon resection was also identified as a risk factor for staged surgery in CD patients. The high incidence of postoperative IASCs for colonic CD is significantly associated with primary anastomosis [37], which may be partly related to the poor nutritional status of patients with extensive diseased intestines [38]. In our study, patients who underwent subtotal colectomy had a significantly lower BMI than patients with other types of resections (17.2 vs. 19.8, P < 0.001), in which case anastomotic reconstruction of the colon or ileum with rectum may be a risk factor. Especially when colon lesions are extensive, the possibility of proctitis should also be considered. Therefore, preoperative nutritional support and antibiotics should be optimized [39], to reduce postoperative complications. Meanwhile, planning for temporary stoma should be discussed in detail with patients undergoing colectomy.

Some studies have shown that repeated enterectomy elevates the risk of postoperative anastomotic fistula in patients with Crohn’s disease (CD) [40, 41]. Similarly, emergency surgeries often heighten surgical risks due to abbreviated preoperative preparations, thereby increasing the likelihood of significant postoperative complications [25]. Our findings corroborate these observations, indicating that CD patients with a history of intestinal resection or those who have undergone emergency procedures are also at an increased risk of requiring a temporary stoma. This outcome is logically consistent with efforts to minimize the occurrence of postoperative anastomotic fistulas.

Notably, beyond the established risk factors of above mentioned, our findings also suggested that ESR, a marker of systemic inflammation, was also associated with an increased risk of requiring a temporary stoma after IR. Studies have shown that ESR is associated with fistula and stenosis complications in CD [42], as well as postoperative infectious complications [43]. Our study further showed that ESR was also significantly associated with temporary stoma formation after IR for CD. The ML model’s inclusion of ESR addresses this need by providing clinicians with a quantitative tool to assess the risk of temporary stoma creation based on the patient’s inflammatory status, and ESR is a good choice.

In terms of model construction, this study is the first paper to construct a novel ML model based on comprehensive clinical characteristics to predict the need for temporary stoma in patients undergoing IR for CD. All variables inserted in the prediction model can be used for clinicians’ reference. The main limitations of this study lied in its retrospective nature and it was single-centered, potentially introducing selection bias and limiting generalizability. Additionally, the sample size of this study was not large, and only internal verification was used to evaluate the accuracy and effectiveness of the model. Future studies should employ larger sample sizes, and external validation across diverse cohorts is warranted to confirm our findings.

Conclusions

In this study, we present a novel ML-based predictive model designed to estimate the likelihood of temporary stoma creation in CD surgery. By synthesizing preoperative disease characteristics, patient status, and surgical strategy, our tool aims to bolster clinical decision-making and optimize patient care. Despite demonstrating potential benefits, the predictive accuracy of the model must be considered in light of the study’s limitations. Specifically, its retrospective and single-center nature, along with the absence of external validation, necessitate cautious extrapolation of our findings. To ascertain the ML model’s broader applicability and refine its predictive power, prospective validation across various populations and clinical settings is imperative. We remain dedicated to advancing this research, seeking to translate our preliminary findings into actionable insights for enhanced clinical practice.

Data availability

The datasets used and analyzed during the current study are available from the corresponding author on reasonable request.

Abbreviations

CD:

Crohn’s disease

IR:

Intestinal resection

LASSO:

The least absolute shrinkage and selection operator

LR:

Logistic regression

RF:

Random forest

XG-boost:

Extreme gradient boosting

BMI:

Body mass index

ROC:

Receiver operating characteristic

AUC:

Area under the curve

ALB:

Albumin

ESR:

Erythrocyte sedimentation rate

IASC:

Intra-abdominal septic complications

References

  1. Peyrin-Biroulet L, Loftus EV, Colombel JF, et al. Long-term complications, Extraintestinal manifestations, and Mortality in Adult Crohn’s Disease in Population-based cohorts. Inflamm Bowel Dis. 2011;17(1):471–8.

    Article  PubMed  Google Scholar 

  2. Gunesh S, Thomas GAO, Williams GT, et al. The incidence of Crohn’s disease in Cardiff over the last 75 years: an update for 1996–2005. Aliment Pharm Ther. 2008;27(3):211–9.

    Article  CAS  Google Scholar 

  3. Shivashankar R, Tremaine WJ, Harmsen WS, et al. Incidence and prevalence of Crohn’s Disease and Ulcerative Colitis in Olmsted County, Minnesota from 1970 through 2010. Clin Gastroenterol H. 2017;15(6):857–63.

    Article  Google Scholar 

  4. Molodecky NA, Soon IS, Rabi DM, et al. Increasing incidence and prevalence of the inflammatory Bowel diseases with Time, based on systematic review. Gastroenterology. 2012;142(1):46–54.

    Article  PubMed  Google Scholar 

  5. Li Y, Chen BL, Gao X, et al. Current diagnosis and management of Crohn’s disease in China: results from a multicenter prospective disease registry. Bmc Gastroenterol. 2019;19(1):145.

    Article  PubMed  PubMed Central  Google Scholar 

  6. Wu EH, Duan M, Han JQ, et al. Patients with Crohn’s Disease undergoing abdominal surgery: clinical and prognostic evaluation based on a single-center cohort in China. World J Surg. 2022;46(2):450–60.

    Article  PubMed  Google Scholar 

  7. Rispo A, Imperatore N, Testa A, et al. Combined Endoscopic/Sonographic-based risk Matrix Model for Predicting one-year risk of surgery: a prospective observational study of a tertiary centre Severe/Refractory Crohn’s Disease Cohort. J Crohns Colitis. 2018;12(7):784–93.

    Article  PubMed  Google Scholar 

  8. Kuruvilla K, Osler T, Hyman NH. A comparison of the quality of life of Ulcerative Colitis patients after IPAA vs Ileostomy. Dis Colon Rectum. 2012;55(11):1131–7.

    Article  PubMed  Google Scholar 

  9. Ong ME, Lee Ng CH, Goh K, et al. Prediction of cardiac arrest in critically ill patients presenting to the emergency department using a machine learning score incorporating heart rate variability compared with the modified early warning score. Crit Care. 2012;16(3):R108.

    Article  PubMed  PubMed Central  Google Scholar 

  10. Hu X, Dong H, Qin W, et al. Machine learning-based identification of a consensus immune-derived gene signature to improve head and neck squamous cell carcinoma therapy and outcome. Front Pharmacol. 2024;15:1341346.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  11. Pruthi N, Yap T, Moore C, et al. Applying machine learning for enhanced MicroRNA analysis: a Companion Risk Tool for oral squamous cell carcinoma in Standard Care Incisional Biopsy. Biomolecules. 2024;14(4):458.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  12. Shi J, Chen Y, Wang Y. Deep learning and machine learning approaches to classify stomach distant metastatic tumors using DNA methylation profiles. Comput Biol Med. 2024;175:108496.

    Article  CAS  PubMed  Google Scholar 

  13. Neary PM, Aiello AC, Stocchi L, et al. High-risk Ileocolic Anastomoses for Crohn’s Disease: when is diversion indicated? J Crohns Colitis. 2019;13(7):856–63.

    Article  PubMed  Google Scholar 

  14. Wang KH, Huang LY, Huang W, et al. Predictive value of CT Enterography Index for postoperative intra-abdominal septic complications in patients with Crohn’s Disease: implications for Surgical decision-making. Dis Colon Rectum. 2021;64(8):964–76.

    Article  PubMed  Google Scholar 

  15. Wickramasinghe D, Carvello M, Di Candido F, et al. Factors associated with stoma formation in ileocolic resection for Crohn’s disease and the development of a predictive scoring system. Langenbeck Arch Surg. 2022;407(7):2997–3003.

    Article  Google Scholar 

  16. Everhov AH, Kalman TD, Soderling J, et al. Probability of Stoma in Incident patients with Crohn’s Disease in Sweden 2003–2019: a Population-based study. Inflamm Bowel Dis. 2022;28(8):1160–8.

    Article  PubMed  Google Scholar 

  17. Abdalla MI, Sandler RS, Kappelman MD, et al. The impact of ostomy on Quality of Life and Functional Status of Crohn’s Disease patients. Inflamm Bowel Dis. 2016;22(11):2658–64.

    Article  PubMed  Google Scholar 

  18. Polidano K, Chew-Graham CA, Farmer AD, et al. Access to Psychological Support for Young people following Stoma surgery: exploring patients’ and clinicians’ perspectives. Qual Health Res. 2021;31(3):535–49.

    Article  PubMed  Google Scholar 

  19. Bianchi R, Mamadou-Pathe B, von Kanel R, et al. Effect of closed and permanent stoma on disease course, psychological well-being and working capacity in Swiss IBD cohort study patients. PLoS ONE. 2022;17(9):e0274665.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  20. Person B, Ifargan R, Lachter J, et al. The impact of Preoperative Stoma Site marking on the incidence of complications, Quality of Life, and Patient’s independence. Dis Colon Rectum. 2012;55(7):783–7.

    Article  PubMed  Google Scholar 

  21. Cakir SK, Ozbayir T. The effect of preoperative stoma site marking on quality of life. Pak J Med Sci. 2018;34(1):149–53.

    PubMed  PubMed Central  Google Scholar 

  22. Dibley L, Czuber-Dochan W, Wade T, et al. Patient decision-making about Emergency and Planned Stoma surgery for IBD: a qualitative exploration of patient and clinician perspectives. Inflamm Bowel Dis. 2018;24(2):235–46.

    Article  PubMed  Google Scholar 

  23. Huang WP, Tang YB, Nong LG, et al. Risk factors for postoperative intra-abdominal septic complications after surgery in Crohn’s disease: a meta-analysis of observational studies. J Crohns Colitis. 2015;9(3):293–301.

    Article  PubMed  Google Scholar 

  24. Morar PS, Hodgkinson JD, Thalayasingam S, et al. Determining predictors for intra-abdominal septic complications following Ileocolonic Resection for Crohn’s Disease-considerations in pre-operative and peri-operative optimisation techniques to improve outcome. J Crohns Colitis. 2015;9(6):483–91.

    Article  PubMed  Google Scholar 

  25. Galata C, Weiss C, Hardt J, et al. Risk factors for early postoperative complications and length of hospital stay in ileocecal resection and right hemicolectomy for Crohn’s disease: a single-center experience. Int J Colorectal Dis. 2018;33(7):937–45.

    Article  PubMed  Google Scholar 

  26. Farmer RG, Hawk WA, Turnbull RB. Clinical patterns in Crohn’s disease: a statistical study of 615 cases. Gastroenterology. 1975;68:627–35.

    Article  CAS  PubMed  Google Scholar 

  27. Williams DR, Coller JA, Corman ML, et al. Anal complications in Crohn’s disease. Dis Colon Rectum. 1981;24(1):22–4.

    Article  CAS  PubMed  Google Scholar 

  28. Hancock L, Windsor AC, Mortensen NJ. Inflammatory bowel disease: the view of the surgeon. Colorectal Dis. 2006;8:10–4.

    Article  PubMed  Google Scholar 

  29. Bafford AC, Latushko A, Hansraj N, et al. The Use of Temporary Fecal Diversion in Colonic and Perianal Crohn’s Disease does not improve outcomes. Dig Dis Sci. 2017;62(8):2079–86.

    Article  PubMed  Google Scholar 

  30. Zhu F, Li Y, Guo Z, et al. Nomogram to predict postoperative intra-abdominal septic complications after Bowel Resection and primary anastomosis for Crohn’s Disease. Dis Colon Rectum. 2020;63(5):629–38.

    Article  PubMed  Google Scholar 

  31. Jablonska B, Mrowiec S. Nutritional status and its detection in patients with inflammatory Bowel diseases. Nutrients. 2023;15(8):1991.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  32. Rozich JJ, Zhao B, Luo J, et al. Conventional Frailty Index does not predict risk of postoperative complications in patients with IBD: a Multicenter Cohort Study. Dis Colon Rectum. 2023;66(8):1085–94.

    Article  PubMed  Google Scholar 

  33. Aydinli HH, Aytac E, Remzi FH, et al. Factors Associated with short-term morbidity in patients undergoing Colon resection for Crohn’s Disease. J Gastrointest Surg. 2018;22(8):1434–41.

    Article  PubMed  Google Scholar 

  34. Meade S, Patel KV, Luber RP, et al. A retrospective cohort study: pre-operative oral enteral nutritional optimisation for Crohn & apos;s disease in a UK tertiary IBD centre. Aliment Pharm Ther. 2022;56(4):646–63.

    Article  Google Scholar 

  35. Ferrandis C, Souche R, Bardol T, et al. Personalized pre-habilitation reduces anastomotic complications compared to up front surgery before ileocolic resection in high-risk patients with Crohn’s disease: a single center retrospective study. Int J Surg. 2022;105:106815.

    Article  PubMed  Google Scholar 

  36. Celentano V, Dis SCSC. Surgical treatment of colonic Crohn’s disease: a national snapshot study. Langenbeck Arch Surg. 2021;406(4):1165–72.

    Article  Google Scholar 

  37. Iesalnieks I, Spinelli A, Frasson M, et al. Risk of postoperative morbidity in patients having bowel resection for colonic Crohn’s disease. Tech Coloproctol. 2018;22(12):947–53.

    Article  PubMed  Google Scholar 

  38. Galata C, Kienle P, Weiss C, et al. Risk factors for early postoperative complications in patients with Crohn’s disease after colorectal surgery other than ileocecal resection or right hemicolectomy. Int J Colorectal Dis. 2019;34(2):293–300.

    Article  PubMed  Google Scholar 

  39. Spinelli A, Allocca M, Jovani M, et al. Review article: optimal preparation for surgery in Crohn’s disease. Aliment Pharm Ther. 2014;40(9):1009–22.

    Article  CAS  Google Scholar 

  40. Johnston WF, Stafford C, Francone TD, et al. What is the risk of anastomotic leak after repeat intestinal resection in patients with Crohn’s Disease? Dis Colon Rectum. 2017;60(12):1299–306.

    Article  PubMed  Google Scholar 

  41. Duan Y, Liu Y, Li Y. Previous intestinal resection is Associated with Postoperative complications in Crohn’s Disease: a Cohort Study. Gastroenterol Res Pract. 2020;2020:2194382.

    Article  PubMed  PubMed Central  Google Scholar 

  42. Li T, Qian YT, Bai TT, et al. Prediction of complications in inflammatory bowel disease using routine blood parameters at diagnosis. Ann Transl Med. 2022;10(4):185.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  43. Tang SS, Dong X, Liu W, et al. Compare risk factors associated with postoperative infectious complication in Crohn’s disease with and without preoperative infliximab therapy: a cohort study. Int J Colorectal Dis. 2020;35(4):727–37.

    Article  PubMed  Google Scholar 

Download references

Acknowledgements

Not applicable.

Funding

The present study was supported by the horizontal project of Shanghai tenth people’s hospital (Grant No. 05.06.22.016, 05.06.22.017).

Author information

Author notes

  1. Fang-Tao Wang and Yin Lin contributed equally to this work.

Authors and Affiliations

  1. Department of Abdominal Surgery, Shanghai Tenth People’s Hospital, Tongji University School of Medicine, Shanghai, 200072, China

    Fang-Tao Wang, Ren-Yuan Gao, Xiao-Cai Wu, Tian-Qi Wu, Yi-Ran Jiao, Ji-Yuan Li, Lu Yin & Chun-Qiu Chen

  2. Department of General Surgery, Yangpu Hospital, Tongji University School of Medicine, Shanghai, 200090, China

    Yin Lin

Authors
  1. Fang-Tao Wang
    View author publications

    You can also search for this author inPubMed Google Scholar

  2. Yin Lin
    View author publications

    You can also search for this author inPubMed Google Scholar

  3. Ren-Yuan Gao
    View author publications

    You can also search for this author inPubMed Google Scholar

  4. Xiao-Cai Wu
    View author publications

    You can also search for this author inPubMed Google Scholar

  5. Tian-Qi Wu
    View author publications

    You can also search for this author inPubMed Google Scholar

  6. Yi-Ran Jiao
    View author publications

    You can also search for this author inPubMed Google Scholar

  7. Ji-Yuan Li
    View author publications

    You can also search for this author inPubMed Google Scholar

  8. Lu Yin
    View author publications

    You can also search for this author inPubMed Google Scholar

  9. Chun-Qiu Chen
    View author publications

    You can also search for this author inPubMed Google Scholar

Contributions

Wang FT contributed to manuscript writing and data collection, and data analysis; Gao RY, Wu XC and Wu TQ contributed to manuscript editing; Lin Y, Jiao YR and Li JY contributed to data collection; Yin L and Chen CQ contributed to conceptualization and supervision; all authors have read and approved the final manuscript.

Corresponding author

Correspondence to Chun-Qiu Chen.

Ethics declarations

Ethics approval and consent to participate

This study was approved by the Shanghai Tenth People’s Hospital Ethics Committee (SHSY-IEC-5.0/24K9/P01), and the requirement for informed consent was waived because of the retrospective nature of this study.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.

Additional information

Publisher’s note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Supplementary Material 1

Supplementary Material 2

Supplementary Material 3

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Wang, FT., Lin, Y., Gao, RY. et al. Machine learning for temporary stoma after intestinal resection in surgical decision-making of Crohn’s disease. BMC Gastroenterol 25, 117 (2025). https://doiorg.publicaciones.saludcastillayleon.es/10.1186/s12876-025-03668-7

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doiorg.publicaciones.saludcastillayleon.es/10.1186/s12876-025-03668-7

Keywords

BMC Gastroenterology

ISSN: 1471-230X

Contact us