Machine learning models predict anti-TNF-alpha treatment response in Crohn's disease with C-indices up to 0.883Machine Learning Helps Predict Success of Crohn's Disease Treatment

Frontiers in Medicine Published October 8, 2026 Study authors: Yuchen Lin, Siyuan Sun, Siyi Yang, Jinhui Sun DOI ↗ Editorial oversight: Dr. Julia Lee, PhD · Oncology, Genomics & Drug Development

AI-generated summary of the cited source, checked by automated accuracy review. How we work

Key Takeaway

Note that machine learning models combining clinical and radiomic features may improve prediction of anti-TNF-alpha response.

This meta-analysis evaluates the performance of machine learning (ML) models in predicting treatment response to anti-TNF-alpha therapy in a population of 6,697 patients with Crohn's disease. The analysis focuses on the predictive accuracy of various data inputs, including clinical features, radiomic features, and endoscopic mucosal response.

Key findings indicate that radiomic features and combined clinical and radiomic features provide higher predictive accuracy than clinical features alone. In external validation, the combined model achieved a C-index of 0.883 (95% CI: 0.816-0.955), while radiomics alone achieved 0.865 (95% CI: 0.783-0.955). In internal validation, radiomic features yielded a C-index of 0.840 (95% CI: 0.781-0.904) and combined features yielded 0.830 (95% CI: 0.780-0.883). Models based on endoscopic mucosal response showed a C-index of 0.794.

The authors note that while ML shows promise for identifying non-responders early, the findings are subject to methodological limitations. Specifically, the authors highlight a need for multi-center, geographically diverse datasets to improve the generalizability of these models. Clinical application currently requires caution until model robustness is improved through broader data integration.

How this fits prior evidence

This meta-analysis addresses a gap in identifying patients unlikely to respond to anti-TNF-alpha therapy in Crohn's disease. While previous evidence has explored the role of Mycobacterium avium subsp. paratuberculosis in Crohn's disease and the potential of extracellular vesicle-based therapies for tissue repair, this study focuses on predictive analytics to optimize current treatment selection. It does not directly relate to the efficacy of subcutaneous infliximab or the role of traditional Chinese medicine in symptom management.

Researchers analyzed data from over 6,000 patients with Crohn's disease to see if machine learning could predict how well patients respond to anti-TNF-alpha therapy. This type of treatment is common for managing Crohn's disease, but not every patient responds the same way.

The study found that models using radiomic features (data from medical imaging) and combined clinical features performed well in predicting treatment success. These models showed higher accuracy scores compared to using clinical features alone. The results suggest that combining different types of data can help identify which patients might not respond to the medication early on.

Because this was a meta-analysis of existing data, there are some limitations. The researchers noted that the study had methodological limitations and needs more diverse data from different locations to be even more reliable. While these tools show promise for helping doctors choose the best treatment path, more research is needed to confirm how they can be used in everyday clinical practice.

What this means for you:

Machine learning models show promise in predicting how Crohn's patients respond to anti-TNF-alpha therapy.

Common questions

How does machine learning help with Crohn's disease?

Machine learning models can analyze clinical features and radiomic features from medical imaging. These models help predict how well a patient will respond to anti-TNF-alpha therapy. This could help doctors identify patients who might not respond well to the treatment early on.

What specific data was used to make these predictions?

The study looked at clinical features, radiomic features, and endoscopic mucosal response. The results showed that combining clinical features with radiomic features provided a high prediction score of 0.883 in some tests, outperforming clinical features alone.

Is this a proven way to treat Crohn's disease?

This study shows the potential of machine learning as a tool for prediction, not as a replacement for medical treatment. Because the study had some methodological limitations and needs more diverse data, you should talk to your doctor about your specific treatment plan.

Study Details

Study typeMeta analysis

EvidenceLevel 1

PublishedSep 2026

View Original Abstract ↓

This meta-analysis aimed to evaluate the performance of machine learning (ML) in predicting the response to anti-tumor necrosis factor-α (anti-TNF-α) therapy in individuals with Crohn’s disease (CD), providing an evidence-based basis for the construction or update of prediction tools. PubMed, Cochrane Library, Embase, and Web of Science were comprehensively searched for studies published up to November 18, 2025 on ML predictive models for the therapeutic response to anti-TNF treatment in individuals with CD. The Prediction model Risk Of Bias Assessment Tool (PROBAST) was used to assess the risk of bias in the eligible articles. Model performance was assessed using the concordance index (C-index), and a meta-analysis was conducted. A total of 29 studies involving 6,697 patients with CD were included. Of these studies, 22 focused on the response to anti–TNF-α therapy, 6 on mucosal healing, and 1 on other outcomes. The meta-analysis demonstrated that for predicting the treatment response to anti–TNF-α treatment, in the validation dataset, the C-indices were 0.762 for ML models based on clinical features (CFs), 0.852 for models based on radiomic features, and 0.854 for models based on CFs combined with radiomic features. In internal validation, the C-indices for models based on CFs, radiomics, and CFs combined with radiomics were 0.773 (95% CI: 0.711–0.841), 0.840 (95% CI: 0.781–0.904), and 0.830 (95% CI: 0.780–0.883), respectively. In the external validation set, models based on radiomics (0.865, 95% CI: 0.783–0.955) and models based on CFs combined with radiomics (0.883, 95% CI: 0.816–0.955) outperformed models based on CFs (0.733, 95% CI: 0.645–0.833). Furthermore, the C-index of models based on endoscopic mucosal response was 0.794. hile ML shows promise in predicting the therapeutic effects of anti-TNF-α treatment in individuals with CD, there are some methodological limitations. Future research should integrate multi-center, geographically diverse datasets to enhance the generalizability of models. Improving model robustness will help identify non-responders early, ultimately promoting the development of personalized therapeutic strategies and precision medicine. https://www.crd.york.ac.uk/PROSPERO/, identifier CRD420251276470.

Machine learning models predict anti-TNF-alpha treatment response in Crohn's disease with C-indices up to 0.883Machine Learning Helps Predict Success of Crohn's Disease Treatment

How this fits prior evidence

Common questions

Study Details

TLR4 Asp299Gly polymorphism associated with increased infection susceptibility (OR 2.05) and mortality (HR 1.78)

Genetic variations linked to higher risk of severe infections

Clinical research that matters. Delivered to your inbox.

Machine learning models predict anti-TNF-alpha treatment response in Crohn's disease with C-indices up to 0.883Machine Learning Helps Predict Success of Crohn's Disease Treatment

How this fits prior evidence

Common questions

More on Crohn's Disease

Study Details

TLR4 Asp299Gly polymorphism associated with increased infection susceptibility (OR 2.05) and mortality (HR 1.78)

Genetic variations linked to higher risk of severe infections

Clinical research that matters. Delivered to your inbox.

Related in Genetics & Precision Medicine

From Other Specialties