A clinical prediction model for the probability of developing an anal fistula in patients with crohn's disease and a design method

Abstract

This invention proposes a clinical prediction model and design method for the probability of developing anal fistula in Crohn's disease patients, belonging to the field of clinical medical technology. The risk prediction model includes: an input module receiving patient parameters consisting of age at diagnosis, biological agent mode switching, and Montreal manifestation classification at initial diagnosis; a processing module connected to the input module, calculating the assessment and prediction probability of anal fistula in Crohn's disease patients based on the age at diagnosis, biological agent mode switching, and Montreal manifestation classification data at initial diagnosis, according to the following formula; and an output module connected to the processing module, outputting the risk probability of anal fistula in Crohn's disease patients obtained from the risk prediction model. The prediction model is validated on the training and validation sets using the area under the receiver operating characteristic (ROC) curve and a calibration curve to determine its discriminative power and calibration. It can accurately predict the risk of anal fistula in Crohn's disease patients before treatment.

Description

Technical Field

[0001] This invention belongs to the field of clinical medical technology, specifically relating to a clinical prediction model and design method for the probability of anal fistula in Crohn's disease patients. Background Technology

[0002] Crohn's disease (CD) is a chronic, nonspecific inflammatory bowel disease of unknown etiology, associated with autoimmunity. Perianal fistulizing Crohn's disease (pfCD) is one of the most common complications of CD and is clinically refractory. pfCD commonly affects young adults aged 15-35 and is characterized by its complexity, refractory nature, and high recurrence rate. The development of pfCD may be influenced by multiple factors, including genetics, environment, gut microbiota dysbiosis, and immune disorders, and is directly related to the activity and severity of Crohn's disease. 70-80% of pfCD cases present as complex anal fistulas, which may involve multiple fistulas, abscesses, and branches, accompanied by symptoms such as perianal pain, bowel discomfort, bloody or purulent stools. Currently, the cure rate of traditional surgical treatment for pfCD is only about 50%, and one-third of patients experience recurrence. Many patients face a lifelong risk of fistula rupture and anal fistula malignancy.

[0003] In clinical practice, Crohn's disease-related anal fistulas are easily misdiagnosed as ordinary glandular infection-related anal fistulas, leading to increased complexity of the fistula and damage to the structure and function of the anus. Therefore, it is crucial to identify the pathogenesis-related factors of pfCD through correlation analysis to provide early warning of pfCD development and promote early diagnosis and treatment of Crohn's disease. Based on this, this invention develops a predictive model that can accurately predict the risk of anal fistula development in Crohn's disease patients before treatment, which has significant clinical implications. Summary of the Invention

[0004] The purpose of this invention is to propose a clinical prediction model and design method for the probability of anal fistula in Crohn's disease patients. The established risk prediction model can be conveniently used to predict the risk of anal fistula in Crohn's disease patients.

[0005] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0006] A clinical predictive model for the probability of developing anal fistula in patients with Crohn's disease includes:

[0007] The input module receives patient parameters consisting of age at diagnosis, switching of biological agent mode, and Montreal presentation classification at the time of initial diagnosis.

[0008] The processing module, connected to the input module, calculates the Logit(P) score and predicted probability P for Crohn's disease patients developing anal fistulas based on the patient's age at diagnosis, biological agent mode switching, and Montreal manifestation classification data at initial diagnosis, according to the following formula:

[0009] Logit(P)=-1.0956+1.6158*(A)-1.6019*(B)+1.4384*(C)

[0010] Logit(P) = Ln

[0011] in,

[0012] A: Age at diagnosis: 0 for ages ≤ 16; 2 for ages 17-40; 1 for ages > 40.

[0013] B: Replacing the biological agents mode switch, the value is 0 when it is YES and 1 when it is NO;

[0014] C: Montreal classification (Behavior) at initial diagnosis. The value is 1 when the classification is non-stenotic and non-penetrating; the value is 0 when the classification is stenotic; and the value is 0 when the classification is penetrating.

[0015] The output module, connected to the processing module, outputs the probability of anal fistula in Rohn's disease patients obtained from the risk prediction model.

[0016] In addition, this invention also proposes a design method for the risk prediction model, the steps of which are as follows:

[0017] Step 1: Record at least the following patient parameters as predictors:

[0018] Gender, age at diagnosis, initial symptoms, history of perianal surgery, history of abdominal surgery, duration of use of biologics, type of biologic, switching of biologic modality, Montreal location classification at initial diagnosis, and Montreal presentation classification at initial diagnosis;

[0019] Step 2: The above predictive factors were analyzed using multivariate logistic regression to obtain independent risk factors for anal fistula in patients with Crohn's disease, including age at diagnosis, switching of biological agent modality, and Montreal presentation type at initial diagnosis.

[0020] Step 3: Construct a formula for calculating the probability of anal fistula in Crohn's disease patients, composed of the aforementioned independent risk factors:

[0021] Logit(P)=-1.0956+1.6158*(A)-1.6019*(B)+1.4384*(C)

[0022] Logit(P) = Ln

[0023] in,

[0024] A: Age at diagnosis: 0 for ages ≤ 16; 2 for ages 17-40; 1 for ages > 40.

[0025] B: Replacing the biological agents mode switch, the value is 0 when it is YES and 1 when it is NO;

[0026] C: Montreal classification (Behavior) at initial diagnosis. The value is 1 when the classification is non-stenotic and non-penetrating; the value is 0 when the classification is stenotic; and the value is 0 when the classification is penetrating.

[0027] Step 4: Construct a risk prediction model consisting of an input module, a processing module, and an output module. The input module receives patient parameters based on age at diagnosis, switching of biological agent mode, and Montreal presentation classification at initial diagnosis. The processing module is connected to the input module and calculates the Logit(P) score and predicted probability P of Crohn's disease patients developing anal fistula according to the formula obtained in Step 3. The output module is connected to the processing module and inputs the prediction results.

[0028] Step 5: After obtaining the risk prediction model, the prediction model is validated in the modeling population and the validation population using the area under the receiver operating characteristic (ROC) curve (AUC), as well as the calibration curve, decision curve, and clinical impact curve, respectively, to evaluate its performance.

[0029] Compared with the prior art, the beneficial effects of the present invention are mainly reflected in:

[0030] 1. Retrospectively analyze the relevant clinical data of patients diagnosed with Crohn's disease, screen for independent risk factors using multivariate logistic regression, and establish a clinical prediction model for them.

[0031] 2. The predictive model was validated on the training and validation sets using the area under the receiver operating characteristic curve and the calibration curve, respectively, to determine its discrimination and calibration. Attached Figure Description

[0032] Figure 1 This is a nomogram predicting the risk of developing anal fistula in patients with Crohn's disease.

[0033] Figure 2 These are the ROC curves (A) and calibration curves (B) for the training set.

[0034] Figure 3 Here are the ROC curves (A) and calibration curves (B) for the validation set. Detailed Implementation

[0035] Example 1

[0036] Step 1: Retrospective Data Collection

[0037] 1. Research Subjects

[0038] A retrospective analysis was conducted on the clinical data of patients diagnosed with Crohn's disease in the inpatient department between January 2014 and October 2024.

[0039] 2. Diagnostic Criteria: The diagnosis of Crohn's disease anal fistula is based on the "Expert Consensus on the Diagnosis and Treatment of Crohn's Disease Anal Fistula" published by the China Association of Traditional Chinese Medicine in 2019.

[0040] 3. Inclusion criteria

[0041] ① Meets the Western medical diagnostic criteria for Crohn's disease and anal fistula;

[0042] ②This belongs to the Crohn's disease anal fistula of the damp-heat type;

[0043] ③ Age 18 to 65 years old;

[0044] ④ No obvious abnormalities were found in the anal structure and function before the operation;

[0045] ⑤ Sign the informed consent form and cooperate with follow-up visits.

[0046] 4. Exclusion Criteria

[0047] ① Suffering from serious heart, brain, lung, or mental illnesses, etc.;

[0048] ②Having obvious contraindications to surgery;

[0049] ③ History of tuberculosis, diabetes, and perianal malignant tumors;

[0050] ④ During pregnancy or lactation;

[0051] ⑤ Allergic to any of the drugs involved in this study;

[0052] ⑥ Those who are unwilling to sign the informed consent form.

[0053] 5. Standards for shedding and rejection

[0054] (1) Those with poor compliance and who do not cooperate with the treatment plan;

[0055] (2) Those who voluntarily withdraw from the trial;

[0056] (3) Those who were lost to follow-up during the follow-up period.

[0057] 6. Criteria for Termination

[0058] Patients who experience severe adverse reactions during the trial, or whose condition cannot be controlled or worsens and requires treatment with other drugs.

[0059] Step 2: Record at least the following patient parameters as predictors:

[0060] Gender, age at diagnosis, initial symptoms, history of perianal surgery, history of abdominal surgery, duration of use of biologics, type of biologic, switching of biologic modality, Montreal location classification at initial diagnosis, and Montreal presentation classification at initial diagnosis.

[0061] The 169 patients were randomly divided into a training set (124 patients) and a validation set (45 patients) at a ratio of 74:26. ​​The training set was used for model development, and the validation set was used for internal validation. Multivariate logistic regression was used to screen for independent risk factors. The comparison of clinical data between the training and validation sets is shown in Table 1.

[0062] Table 1 Comparison of clinical data between training and validation sets

[0063] variable training set Validation set P value n 124 45 gender 0.022 male 93 (55%) 41 (24.3%) female 31 (18.3%) 4 (2.4%) Age at diagnosis 0.591 A1 (≤16 years old) 35 (20.7%) 12 (7.1%) A2 (17-40 years old) 78 (46.2%) 31 (18.3%) A3 (>40 years old) 11 (6.5%) 2 (1.2%) Current age 0.047 <30 73 (43.2%) 34 (20.1%) ≥30 51 (30.2%) 11 (6.5%) Initial symptoms 0.173 Abdominal pain or bloody stool 41 (24.3%) 20 (11.8%) perianal pain 83 (49.1%) 25 (14.8%) History of perianal surgery < 0.001 No 103 (60.9%) 15 (8.9%) Yes 21 (12.4%) 30 (17.8%) History of abdominal surgery < 0.001 No 70 (41.4%) 39 (23.1%) Yes 54 (32%) 6 (3.6%) Biological reagent administration time 0.854 <1 46 (27.2%) 16 (9.5%) ≥1 78 (46.2%) 29 (17.2%) Biologics 0.103 Infliximab 74 (43.8%) 33 (19.5%) ustekinumab or other 50 (29.6%) 12 (7.1%) Biologics mode switching 0.042 Yes 42 (24.9%) 23 (13.6%) No 82 (48.5%) 22 (13%) Montreal fractal (location) 0.007 L1 (terminal ileum) or L2 (colon) 68 (40.2%) 35 (20.7%) L3 (ileocolon) 56 (33.1%) 10 (5.9%) Montreal Fractal 0.275 B1 (Non-narrow, non-penetrating) 49 (29%) 22 (13%) B2 (narrow type) or B3 (penetrating type) 75 (44.4%) 23 (13.6%) Patient status 0.795 Remission period 80 (47.3%) 30 (17.8%) Activity period 44 (26%) 15 (8.9%)

[0064] Multivariate logistic regression analysis was used to analyze the above predictive factors and found that age at diagnosis, switching of biological agent modality, and Montreal presentation type at initial diagnosis were independent risk factors for anal fistula in Crohn's disease patients.

[0065] Step 3: Construct a formula for calculating the probability of anal fistula in Crohn's disease patients, composed of the aforementioned independent risk factors:

[0066] Logit(P)=-1.0956+1.6158*(A)-1.6019*(B)+1.4384*(C)

[0067] Logit(P) = Ln

[0068] in,

[0069] A: Age at diagnosis: 0 for ages ≤ 16; 2 for ages 17-40; 1 for ages > 40.

[0070] B: Replacing the biological agents mode switch, the value is 0 when it is YES and 1 when it is NO;

[0071] C: Montreal classification (Behavior) at initial diagnosis. The value is 1 when the classification is non-stenotic and non-penetrating; the value is 0 when the classification is stenotic; and the value is 0 when the classification is penetrating.

[0072] Step 4: Construct a risk prediction model consisting of an input module, a processing module, and an output module. The input module receives patient parameters based on age at diagnosis, switching of biological agent mode, and Montreal presentation classification at initial diagnosis. The processing module is connected to the input module and calculates the Logit(P) score and predicted probability P of Crohn's disease patients developing anal fistula according to the formula obtained in Step 3. The output module is connected to the processing module and inputs the prediction results.

[0073] Step 5: After obtaining the risk prediction model, the prediction model is validated in the modeling population and the validation population using the area under the receiver operating characteristic (ROC) curve (AUC), as well as the calibration curve, decision curve, and clinical impact curve, respectively, to evaluate its performance.

[0074] Example 2

[0075] Establishment and validation of nomogram prediction model

[0076] like Figure 1 As shown, three independent risk factors were included in the diagnosis: age at diagnosis, switching of biologic modality, and Montreal presentation type at initial diagnosis. A clinical prediction model was successfully established. The scale at the top of the model provides individual scores for each of the three independent factors. The sum of these individual scores yields the total score, and the predicted probability corresponding to the total score is the probability of Crohn's disease patients developing anal fistula.

[0077] Example 3

[0078] Validation of clinical prediction models

[0079] 1. Discrimination

[0080] After building the model using the training set, its predictive performance was evaluated on the validation set. The model achieved an AUC of 0.790 (95% CI: 0.707–0.874) on the training set and an AUC of 0.816 (95% CI: 0.690–0.941) on the validation set, demonstrating strong discriminative ability (see [link to training set]). Figure 2 , 3 ).

[0081] 2. Calibration

[0082] Calibration analysis showed that the predicted results were in excellent agreement with the observed results (see...). Figure 2 , 3 ).

[0083] Experiments have shown that the predictive model constructed in this invention has good clinical applicability.

Claims

1. A clinical prediction model of the probability of developing an anal fistula in a Crohn's disease patient, characterized in that, include: The input module receives patient parameters consisting of age at diagnosis, switching of biological agent mode, and Montreal presentation classification at the time of initial diagnosis. The processing module, connected to the input module, calculates the Logit(P) score and predicted probability P for Crohn's disease patients developing anal fistulas based on the patient's age at diagnosis, biological agent mode switching, and Montreal manifestation classification data at initial diagnosis, according to the following formula: Logit(P)=-1.0956+1.6158*(A)-1.6019*(B)+1.4384*(C) Logit(P) = Ln in, A: Age at diagnosis: 0 for ages ≤ 16; 2 for ages 17-40; 1 for ages > 40. B: Replacing the biological agents mode switch, the value is 0 when it is YES and 1 when it is NO; C: Montreal classification (Behavior) at initial diagnosis. The value is 1 when the classification is non-stenotic and non-penetrating; the value is 0 when the classification is stenotic; and the value is 0 when the classification is penetrating. The output module, connected to the processing module, outputs the probability of anal fistula in Rohn's disease patients obtained from the risk prediction model.

2. The method of claim 1, wherein the risk prediction model is designed to: The steps are as follows: Step 1: Record at least the following patient parameters as predictors: Gender, age at diagnosis, initial symptoms, history of perianal surgery, history of abdominal surgery, duration of use of biologics, type of biologic, switching of biologic modality, Montreal location classification at initial diagnosis, and Montreal presentation classification at initial diagnosis; Step 2: The above predictive factors were analyzed using multivariate logistic regression to obtain independent risk factors for anal fistula in patients with Crohn's disease, including age at diagnosis, switching of biological agent modality, and Montreal presentation type at initial diagnosis. Step 3: Construct a formula for calculating the probability of anal fistula in Crohn's disease patients, composed of the aforementioned independent risk factors: Logit(P)=-1.0956+1.6158*(A)-1.6019*(B)+1.4384*(C) Logit(P) = Ln in, A: Age at diagnosis: 0 for ages ≤ 16; 2 for ages 17-40; 1 for ages > 40. B: Replacing the biological agents mode switch, the value is 0 when it is YES and 1 when it is NO; C: Montreal classification (Behavior) at initial diagnosis. The value is 1 when the classification is non-stenotic and non-penetrating; the value is 0 when the classification is stenotic; and the value is 0 when the classification is penetrating. Step 4: Construct a risk prediction model consisting of an input module, a processing module, and an output module. The input module receives patient parameters based on age at diagnosis, switching of biological agent mode, and Montreal presentation classification at initial diagnosis. The processing module is connected to the input module and calculates the Logit(P) score and predicted probability P of Crohn's disease patients developing anal fistula according to the formula obtained in Step 3. The output module is connected to the processing module and inputs the prediction results. Step 5: After obtaining the risk prediction model, the prediction model is validated in the modeling population and the validation population using the area under the receiver operating characteristic (ROC) curve (AUC), as well as the calibration curve, decision curve, and clinical impact curve, respectively, to evaluate its performance.

Images