Machine Learning Prediction Model for Inflammatory Bowel Disease Based on Laboratory Markers. Working Model in a Discovery Cohort Study

Artificial intelligence and machine learning technologies in ulcerative colitis

Interest in artificial intelligence (AI) applications for ulcerative colitis (UC) has grown tremendously in recent years. In the past 5 years, there have been over 80 studies focused on machine learning (ML) tools to address a wide range of clinical problems in UC, including diagnosis, prognosis, identification of new UC biomarkers, monitoring of disease activity, and prediction of complications. AI classifiers such as random forest, support vector machines, neural networks, and logistic regression models have been used to model UC clinical outcomes using molecular (transcriptomic) and clinical (electronic health record and laboratory) datasets with relatively high performance (accuracy, sensitivity, and specificity). Application of ML algorithms such as computer vision, guided image filtering, and convolutional neural networks have also been utilized to analyze large and high-dimensional imaging datasets such as endoscopic, histologic, and radiological images for UC diagnosis and prediction of complications (post-surgical complications, colorectal cancer). Incorporation of these ML tools to guide and optimize UC clinical practice is promising but will require large, high-quality validation studies that overcome the risk of bias as well as consider cost-effectiveness compared to standard of care.

Explainable AI-prioritized plasma and fecal metabolites in inflammatory bowel disease and their dietary associations

Fecal metabolites effectively discriminate inflammatory bowel disease (IBD) and show differential associations with diet. Metabolomics and AI-based models, including explainable AI (XAI), play crucial roles in understanding IBD. Using datasets from the UK Biobank and the Human Microbiome Project Phase II IBD Multi'omics Database (HMP2 IBDMDB), this study uses multiple machine learning (ML) classifiers and Shapley additive explanations (SHAP)-based XAI to prioritize plasma and fecal metabolites and analyze their diet correlations. Key findings include the identification of discriminative metabolites like glycoprotein acetyl and albumin in plasma, as well as nicotinic acid metabolites andurobilin in feces. Fecal metabolites provided a more robust disease predictor model (AUC [95%]: 0.93 [0.87-0.99]) compared to plasma metabolites (AUC [95%]: 0.74 [0.69-0.79]), with stronger and more group-differential diet-metabolite associations in feces. The study validates known metabolite associations and highlights the impact of IBD on the interplay between gut microbial metabolites and diet.

Risk Factors and Prediction of 28-Day-All Cause Mortality Among Critically Ill Patients with Acute Pancreatitis Using Machine Learning Techniques: A Retrospective Analysis of Multi-Institutions

Objective

This study aimed to identify the risk factors and construct a reliable prediction model of 28-day all-cause mortality in critically ill patients with acute pancreatitis (AP) using machine learning techniques.

Methods

A total of 534 patients from three different institutions were included. Thirty-eight possible variables were collected from the Intensive care unit (ICU) admission for investigation. Patients were split into a training cohort (n = 400) and test cohort (n = 134) according to their source of hospital. The synthetic minority oversampling technique (SMOTE) was introduced to handle the inherent class imbalance. Six machine learning algorithms were applied in this study. The optimal machine learning model was chosen after patients in the test cohort were selected to validate the models. SHapley Additive exPlanation (SHAP) analysis was performed to rank the importance of variable. The predictive performance of the models was evaluated by the calibration curve, area under the receiver operating characteristics curves (AUROC), and decision clinical analysis.

Results

About 13.5% (72/534) of all patients eventually died of all-cause within 28 days of ICU admission. Eight important variables were screened out, including white blood cell count, platelets, body temperature, age, blood urea nitrogen, red blood cell distribution width, SpO2, and hemoglobin. The support vector machine (SVM) algorithm performed best in predicting 28-d all-cause death. Its AUROC reached 0.877 (95% CI: 0.809 to 0.927, p < 0.001), the Youden index was 0.634 (95% CI: 0.459 to 0.717). Based on the risk stratification system, the difference between the high-risk and low-risk groups was significantly different.

Conclusion

In conclusion, this study developed and validated SVM model, which better predicted 28-d all-cause mortality in critically ill patients with AP. In the future, we will continue to include patients from more institutions to conduct validation in different contexts and countries.

Challenges in IBD Research 2024: Precision Medicine

Precision medicine is part of 5 focus areas of the Challenges in IBD Research 2024 research document, which also includes preclinical human IBD mechanisms, environmental triggers, novel technologies, and pragmatic clinical research. Building on Challenges in IBD Research 2019, the current Challenges aims to provide a comprehensive overview of current gaps in inflammatory bowel diseases (IBDs) research and deliver actionable approaches to address them with a focus on how these gaps can lead to advancements in interception, remission, and restoration for these diseases. The document is the result of multidisciplinary input from scientists, clinicians, patients, and funders, and represents a valuable resource for patient-centric research prioritization. In particular, the precision medicine section is focused on the main research gaps in elucidating how to bring the best care to the individual patient in IBD. Research gaps were identified in biomarker discovery and validation for predicting disease progression and choosing the most appropriate treatment for each patient. Other gaps were identified in making the best use of existing patient biosamples and clinical data, developing new technologies to analyze large datasets, and overcoming regulatory and payer hurdles to enable clinical use of biomarkers. To address these gaps, the Workgroup suggests focusing on thoroughly validating existing candidate biomarkers, using best-in-class data generation and analysis tools, and establishing cross-disciplinary teams to tackle regulatory hurdles as early as possible. Altogether, the precision medicine group recognizes the importance of bringing basic scientific biomarker discovery and translating it into the clinic to help improve the lives of IBD patients.

Cross-scale multi-instance learning for pathological image diagnosis

Analyzing high resolution whole slide images (WSIs) with regard to information across multiple scales poses a significant challenge in digital pathology. Multi-instance learning (MIL) is a common solution for working with high resolution images by classifying bags of objects (i.e. sets of smaller image patches). However, such processing is typically performed at a single scale (e.g., 20× magnification) of WSIs, disregarding the vital inter-scale information that is key to diagnoses by human pathologists. In this study, we propose a novel cross-scale MIL algorithm to explicitly aggregate inter-scale relationships into a single MIL network for pathological image diagnosis. The contribution of this paper is three-fold: (1) A novel cross-scale MIL (CS-MIL) algorithm that integrates the multi-scale information and the inter-scale relationships is proposed; (2) A toy dataset with scale-specific morphological features is created and released to examine and visualize differential cross-scale attention; (3) Superior performance on both in-house and public datasets is demonstrated by our simple cross-scale MIL strategy. The official implementation is publicly available at https://github.com/hrlblab/CS-MIL.

Utility of four machine learning approaches for identifying ulcerative colitis and Crohn's disease

Objective

Peripheral blood routine parameters (PBRPs) are simple and easily acquired markers to identify ulcerative colitis (UC) and Crohn's disease (CD) and reveal the severity, whereas the diagnostic performance of individual PBRP is limited. We, therefore used four machine learning (ML) models to evaluate the diagnostic and predictive values of PBRPs for UC and CD.

Methods

A retrospective study was conducted by collecting the PBRPs of 414 inflammatory bowel disease (IBD) patients, 423 healthy controls (HCs), and 344 non-IBD intestinal diseases (non-IBD) patients. We used approximately 70 % of the PBRPs data from both patients and HCs for training, 30 % for testing, and another group for external verification. The area under the receiver operating characteristic curve (AUC) was used to evaluate the diagnosis and prediction performance of these four ML models.

Results

Multi-layer perceptron artificial neural network model (MLP-ANN) yielded the highest diagnostic performance than the other three models in six subgroups in the training set, which is helpful for discriminating IBD and HCs, UC and CD, active CD and remissive CD, active UC and remissive UC, non-IBD and HCs, and IBD and non-IBD with the AUC of 1.00, 0.988, 0.942, 1.00, 0.986, and 0.97 in the testing set, as well as the AUC of 1.00, 1.00, 0.773, 0.904, 1.00 and 0.992 in the external validation set.

Conclusion

PBRPs-based MLP-ANN model exhibited good performance in discriminating between UC and CD and revealing the disease activity; however, a larger sample size and more models need to be considered for further research.

Impact of artificial intelligence on prognosis, shared decision-making, and precision medicine for patients with inflammatory bowel disease: a perspective and expert opinion

INTRODUCTION

Artificial intelligence (AI) is expected to impact all facets of inflammatory bowel disease (IBD) management, including disease assessment, treatment decisions, discovery and development of new biomarkers and therapeutics, as well as clinician-patient communication.

AREAS COVERED

This perspective paper provides an overview of the application of AI in the clinical management of IBD through a review of the currently available AI models that could be potential tools for prognosis, shared decision-making, and precision medicine. This overview covers models that measure treatment response based on statistical or machine-learning methods, or a combination of the two. We briefly discuss a computational model that allows integration of immune/biological system knowledge with mathematical modeling and also involves a 'digital twin', which allows measurement of temporal trends in mucosal inflammatory activity for predicting treatment response. A viewpoint on AI-enabled wearables and nearables and their use to improve IBD management is also included.

EXPERT OPINION

Although challenges regarding data quality, privacy, and security; ethical concerns; technical limitations; and regulatory barriers remain to be fully addressed, a growing body of evidence suggests a tremendous potential for integration of AI into daily clinical practice to enable precision medicine and shared decision-making.

Profiling the inflammatory bowel diseases using genetics, serum biomarkers, and smoking information

Crohn's disease (CD) and ulcerative colitis (UC) are two etiologically related yet distinctive subtypes of the inflammatory bowel diseases (IBD). Differentiating CD from UC can be challenging using conventional clinical approaches in a subset of patients. We designed and evaluated a novel molecular-based prediction model aggregating genetics, serum biomarkers, and tobacco smoking information to assist the diagnosis of CD and UC in over 30,000 samples. A joint model combining genetics, serum biomarkers and smoking explains 46% (42-50%, 95% CI) of phenotypic variation. Despite modest overlaps with serum biomarkers, genetics makes unique contributions to distinguishing IBD subtypes. Smoking status only explains 1% (0-6%, 95% CI) of the phenotypic variance suggesting it may not be an effective biomarker. This study reveals that molecular-based models combining genetics, serum biomarkers, and smoking information could complement current diagnostic strategies and help classify patients based on biologic state rather than imperfect clinical parameters.

Serological Biomarker-Based Machine Learning Models for Predicting the Relapse of Ulcerative Colitis

Purpose

To explore whether machine learning models using serological markers can predict the relapse of Ulcerative colitis (UC).

Patients and Methods

This clinical cohort study included 292 UC patients, and serological markers were obtained when patients were discharged from the hospital. Subsequently, four machine learning models including the random forest (RF) model, the logistic regression model, the decision tree, and the neural network were compared to predict the relapse of UC. A nomogram was constructed, and the performance of these models was evaluated by accuracy, sensitivity, specificity, and the area under the receiver operating characteristic curve (AUC).

Results

Based on the patients' characteristics and serological markers, we selected the relevant variables associated with relapse and developed a LR model. The novel model including gender, white blood cell count, percentage of leukomonocyte, percentage of monocyte, absolute value of neutrophilic granulocyte, and erythrocyte sedimentation rate was established for predicting the relapse. In addition, the average AUC of the four machine learning models was 0.828, of which the RF model was the best. The AUC of the test group was 0.889, the accuracy was 76.4%, the sensitivity was 78.5%, and the specificity was 76.4%. There were 45 variables in the RF models, and the relative weight coefficients of these variables were determined. Age has the greatest impact on classification results, followed by hemoglobin concentration, white blood cell count, and platelet distribution width.

Conclusion

Machine learning models based on serological markers had high accuracy in predicting the relapse of UC. The model can be used to noninvasively predict patient outcomes and can be an effective tool for determining personalized treatment plans.

Performance of Machine Learning Algorithms for Predicting Disease Activity in Inflammatory Bowel Disease

This study aimed to explore the effectiveness of predicting disease activity in patients with inflammatory bowel disease (IBD), using machine learning (ML) models. A retrospective research was undertaken on IBD patients who were admitted into the First Affiliated Hospital of Wenzhou Medical University between September 2011 and September 2019. At first, data were randomly split into a 3:1 ratio of training to test set. The least absolute shrinkage and selection operator (LASSO) algorithm was applied to reduce the dimension of variables. These variables were used to generate seven ML algorithms, namely random forests (RFs), adaptive boosting (AdaBoost), K-nearest neighbors (KNNs), support vector machines (SVMs), naïve Bayes (NB), ridge regression, and eXtreme gradient boosting (XGBoost) to train to predict disease activity in IBD patients. SHapley Additive exPlanation (SHAP) analysis was performed to rank variable importance. A total of 876 participants with IBD, consisting of 275 ulcerative colitis (UC) and 601 Crohn's disease (CD), were retrospectively enrolled in the study. Thirty-three variables were obtained from the clinical characteristics and laboratory tests of the participants. Finally, after LASSO analysis, 11 and 5 variables were screened out to construct ML models for CD and UC, respectively. All seven ML models performed well in predicting disease activity in the CD and UC test sets. Among these ML models, SVM was more effective in predicting disease activity in the CD group, whose AUC reached 0.975, sensitivity 0.947, specificity 0.920, and accuracy 0.933. AdaBoost performed best for the UC group, with an AUC of 0.911, sensitivity 0.844, specificity 0.875, and accuracy 0.855. ML algorithms were available and capable of predicting disease activity in IBD patients. Based on clinical and laboratory variables, ML algorithms demonstrate great promise in guiding physicians' decision-making.

Artificial-intelligence-based decision support tools for the differential diagnosis of colitis

BACKGROUND

Whereas Artificial Intelligence (AI) based tools have recently been introduced in the field of gastroenterology, application in inflammatory bowel disease (IBD) is in its infancies. We established AI-based algorithms to distinguish IBD from infectious and ischemic colitis using endoscopic images and clinical data.

METHODS

First, we trained and tested a Convolutional Neural Network (CNN) using 1796 real-world images from 494 patients, presenting with three diseases (IBD [n = 212], ischemic colitis [n = 157], and infectious colitis [n = 125]). Moreover, we evaluated a Gradient Boosted Decision Trees (GBDT) algorithm using five clinical parameters as well as a hybrid approach (CNN + GBDT). Patients and images were randomly split into two completely independent datasets. The proposed approaches were benchmarked against each other and three expert endoscopists on the test set.

RESULTS

For the image-based CNN, the GBDT algorithm and the hybrid approach global accuracies were .709, .792, and .766, respectively. Positive predictive values were .602, .702, and .657. Global areas under the receiver operating characteristics (ROC) and precision recall (PR) curves were .727/.585, .888/.823, and .838/.733, respectively. Global accuracy did not differ between CNN and endoscopists (.721), but the clinical parameter-based GBDT algorithm outperformed CNN and expert image classification.

CONCLUSIONS

Decision support systems exclusively based on endoscopic image analysis for the differential diagnosis of colitis, representing a complex clinical challenge, seem not yet to be ready for primetime and more diverse image datasets may be necessary to improve performance in future development. The clinical value of the proposed clinical parameters algorithm should be evaluated in prospective cohorts.

Identifying endotypes of individuals after an attack of pancreatitis based on unsupervised machine learning of multiplex cytokine profiles

After an attack of pancreatitis, individuals may develop metabolic sequelae (eg, new-onset diabetes) and/or pancreatic cancer. These new-onset morbidities are, at least in part, driven by low-grade inflammation. The aim was to study the profiles of cytokines/chemokines in individuals after an attack of pancreatitis. A commercially available panel including 31 cytokines/chemokines was investigated. Random forest classifier and unsupervised hierarchical clustering were applied to study participants (who had no persistent organ failure and did not require ICU admission) according to their cytokine/chemokine profiles. Pancreatitis-related characteristics, detailed body composition (determined using 3.0 T magnetic resonance imaging), markers of glucose, lipid, and iron metabolism, gut and pancreatic hormones, as well as liver and pancreatic enzymes, were compared between clusters. Bootstrap validation was employed. A total of 160 participants, including 107 postpancreatitis individuals (investigated at a median of 18 months after the last attack of pancreatitis) and 53 healthy volunteers, were studied. Twenty-two cytokines/chemokines were significantly different between postpancreatitis and health. Two distinct endotypes of individuals after an attack of pancreatitis were identified-‟inflammatory" and ‟noninflammatory." Sixteen cytokines/chemokines were significantly higher in the inflammatory endotype compared with the noninflammatory endotype. No cytokine/chemokine was significantly higher in the noninflammatory endotype. The inflammatory endotype was characterized by significantly elevated insulin (P= 0.001), glucose-dependent insulinotropic peptide (P = 0.001), peptide YY (P = 0.017), and ghrelin (P = 0.014). The noninflammatory endotype was characterized by significantly elevated hepcidin (P= 0.016). Pancreatitis-related factors, body composition, and other studied parameters did not differ significantly between the 2 endotypes. Individuals with a similar phenotype and clinical course of pancreatitis have differing cytokine/chemokine profiles after clinical resolution of the disease. People with the inflammatory endotype have distinct changes in the pancreatic and gut hormones known to be involved in the pathogenesis of new-onset morbidities after an attack of pancreatitis.

Canine chronic enteropathy—Current state-of-the-art and emerging concepts

Over the last decade, chronic inflammatory enteropathies (CIE) in dogs have received great attention in the basic and clinical research arena. The 2010 ACVIM Consensus Statement, including guidelines for the diagnostic criteria for canine and feline CIE, was an important milestone to a more standardized approach to patients suspected of a CIE diagnosis. Great strides have been made since understanding the pathogenesis and classification of CIE in dogs, and novel diagnostic and treatment options have evolved. New concepts in the microbiome-host-interaction, metabolic pathways, crosstalk within the mucosal immune system, and extension to the gut-brain axis have emerged. Novel diagnostics have been developed, the clinical utility of which remains to be critically evaluated in the next coming years. New directions are also expected to lead to a larger spectrum of treatment options tailored to the individual patient. This review offers insights into emerging concepts and future directions proposed for further CIE research in dogs for the next decade to come.

Cross-scale Attention Guided Multi-instance Learning for Crohn's Disease Diagnosis with Pathological Images

Multi-instance learning (MIL) is widely used in the computer-aided interpretation of pathological Whole Slide Images (WSIs) to solve the lack of pixel-wise or patch-wise annotations. Often, this approach directly applies "natural image driven" MIL algorithms which overlook the multi-scale (i.e. pyramidal) nature of WSIs. Off-the-shelf MIL algorithms are typically deployed on a single-scale of WSIs (e.g., 20× magnification), while human pathologists usually aggregate the global and local patterns in a multi-scale manner (e.g., by zooming in and out between different magnifications). In this study, we propose a novel cross-scale attention mechanism to explicitly aggregate inter-scale interactions into a single MIL network for Crohn's Disease (CD), which is a form of inflammatory bowel disease. The contribution of this paper is two-fold: (1) a cross-scale attention mechanism is proposed to aggregate features from different resolutions with multi-scale interaction; and (2) differential multi-scale attention visualizations are generated to localize explainable lesion patterns. By training ~250,000 H&E-stained Ascending Colon (AC) patches from 20 CD patient and 30 healthy control samples at different scales, our approach achieved a superior Area under the Curve (AUC) score of 0.8924 compared with baseline models. The official implementation is publicly available at https://github.com/hrlblab/CS-MIL.

Identification of useful genes from multiple microarrays for ulcerative colitis diagnosis based on machine learning methods

Ulcerative colitis (UC) is a chronic relapsing inflammatory bowel disease with an increasing incidence and prevalence worldwide. The diagnosis for UC mainly relies on clinical symptoms and laboratory examinations. As some previous studies have revealed that there is an association between gene expression signature and disease severity, we thereby aim to assess whether genes can help to diagnose UC and predict its correlation with immune regulation. A total of ten eligible microarrays (including 387 UC patients and 139 healthy subjects) were included in this study, specifically with six microarrays (GSE48634, GSE6731, GSE114527, GSE13367, GSE36807, and GSE3629) in the training group and four microarrays (GSE53306, GSE87473, GSE74265, and GSE96665) in the testing group. After the data processing, we found 87 differently expressed genes. Furthermore, a total of six machine learning methods, including support vector machine, least absolute shrinkage and selection operator, random forest, gradient boosting machine, principal component analysis, and neural network were adopted to identify potentially useful genes. The synthetic minority oversampling (SMOTE) was used to adjust the imbalanced sample size for two groups (if any). Consequently, six genes were selected for model establishment. According to the receiver operating characteristic, two genes of OLFM4 and C4BPB were finally identified. The average values of area under curve for these two genes are higher than 0.8, either in the original datasets or SMOTE-adjusted datasets. Besides, these two genes also significantly correlated to six immune cells, namely Macrophages M1, Macrophages M2, Mast cells activated, Mast cells resting, Monocytes, and NK cells activated (P  <  0.05). OLFM4 and C4BPB may be conducive to identifying patients with UC. Further verification studies could be conducted.