Multimodal machine learning models using body composition analysis show good performance predicting colorectal cancer complicationsMachine learning models help predict complications after colorectal cancer surgery

Frontiers in Medicine Published October 7, 2026 DOI ↗ Editorial oversight: Dr. Amelia Tan, PhD · Internal Medicine & Chronic Disease

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

Key Takeaway

Note that multimodal machine learning models using body composition analysis show good performance in predicting complications.

This meta-analysis evaluated the predictive performance of multimodal machine learning (ML) models in patients with colorectal cancer. These models integrate clinical variables with CT-based body composition analysis (BCA) to identify risks for postoperative complications. The study included a total of 7,835 patients across various settings.

The analysis found that multimodal ML models demonstrated good discriminative performance with a pooled AUC of 0.84 (95% CI: 0.76-0.89). Nomogram-based models showed a pooled AUC of 0.87 (95% CI: 0.78-0.92). For the specific complication of anastomotic leakage, the pooled AUC was 0.83, though this estimate was noted as imprecise due to the inclusion of only two contributing studies.

Several limitations were identified, including substantial heterogeneity (I2 = 79.7%) and the small number of studies contributing to the anastomotic leakage data. These findings suggest that multimodal ML models incorporating BCA biomarkers may provide a robust framework for preoperative risk stratification in colorectal cancer patients, though the high heterogeneity necessitates cautious interpretation of the results.

How this fits prior evidence

This meta-analysis addresses a gap in preoperative risk stratification for colorectal cancer patients. While prior evidence has identified specific risk factors such as an ASA score of 3 or higher for UTI in rectal cancer patients, this study focuses on using machine learning and body composition analysis to predict postoperative complications. The findings provide a different technological approach to risk assessment than the previously noted clinical markers or imaging parameters like IVIM MRI parameter D.

Predicting complications after surgery is vital for patients facing colorectal cancer. Doctors want to know who might face risks like anastomotic leakage, which is a common complication where the surgical connection leaks. This study looked at how well machine learning models could predict these issues by combining body composition data from CT scans with clinical information.

The analysis of 7,835 patients showed that these multimodal models performed well at identifying risks. Specifically, the models showed good discriminative performance with a score of 0.84. Another type of model, called a nomogram based model, also showed strong performance. These tools aim to give doctors a clearer picture of a patient's risk before they ever enter the operating room.

While the results are promising, there are some notes of caution. The data for predicting anastomotic leakage specifically came from only two studies, making that specific estimate less precise. There was also a lot of variation in the data across the different studies included. These tools are meant to help doctors understand risk, not to replace clinical judgment.

What this means for you:

Machine learning models using body composition data can help doctors predict complications after colorectal cancer surgery.

Common questions

How accurate are these models at predicting complications?

The study found that machine learning models had good discriminative performance, with a score of 0.84 for predicting postoperative complications. Another type of model, a nomogram based model, also showed strong performance with a score of 0.87. These scores suggest the models are effective at identifying which patients might face issues after surgery.

Can these models predict specific issues like anastomotic leakage?

The models were tested for anastomotic leakage, which is a specific type of surgical complication. While the results showed a score of 0.83, the estimate was considered imprecise because the data came from only two studies. You should talk to your doctor about how these tools might apply to your specific case.

What information do these models use to predict risk?

These models are multimodal, meaning they combine different types of information. They specifically integrate body composition analysis from CT scans with various clinical variables to create a more complete picture of a patient's risk before surgery.

Study Details

Study typeMeta analysis

EvidenceLevel 1

PublishedOct 2026

View Original Abstract ↓

BackgroundMalnutrition and sarcopenic obesity are prevalent in colorectal cancer (CRC) and linked to adverse surgical outcomes. Conventional screening often overlooks this occult malnutrition. Preoperative computed tomography (CT)-based body composition analysis (BCA) provides objective biomarkers for muscle depletion and fat distribution. This systematic review evaluates multimodal machine learning (ML) models integrating BCA and clinical variables to predict CRC postoperative complications and inform precision nutrition strategies.MethodsWe systematically searched PubMed, Embase, Web of Science, and Cochrane Library (inception to April 13, 2026) for ML models predicting postoperative complications in CRC using BCA. Quality was assessed using PROBAST. Pooled area under the curve (AUC) was estimated via a random-effects model, alongside subgroup analyses.ResultsEight studies (n = 7,835) were included. The pooled AUC was 0.84 (95% CI: 0.76-0.89), indicating good discriminative performance, although substantial heterogeneity was present (I2 = 79.7%). Nomogram-based models showed a pooled AUC of 0.87 (95% CI: 0.78-0.92). For anastomotic leakage, the pooled AUC was 0.83 (95% CI: 0.35-0.98), although this estimate was imprecise because only two studies contributed to this subgroup.ConclusionMultimodal ML models integrating BCA biomarkers offer a robust, data-driven framework for preoperative risk stratification in CRC. Identifying high-risk body composition phenotypes may facilitate tailored prehabilitation and targeted immunonutrition strategies. Future efforts must prioritize standardized BCA and multi-center external validation.Systematic review registrationCRD420261364363.

Multimodal machine learning models using body composition analysis show good performance predicting colorectal cancer complicationsMachine learning models help predict complications after colorectal cancer surgery

How this fits prior evidence

Common questions

Study Details

Sitagliptin and liraglutide superior for CGM targets in type 2 diabetes

Two diabetes drugs outperform others in long-term glucose control

Clinical research that matters. Delivered to your inbox.

Multimodal machine learning models using body composition analysis show good performance predicting colorectal cancer complicationsMachine learning models help predict complications after colorectal cancer surgery

How this fits prior evidence

Common questions

More on Colorectal Cancer

Study Details

Sitagliptin and liraglutide superior for CGM targets in type 2 diabetes

Two diabetes drugs outperform others in long-term glucose control

Clinical research that matters. Delivered to your inbox.

Related in Nutrition & Obesity Medicine

From Other Specialties