Radiomics and clinical data combined in AI models achieve 0.86 AUC for predicting ovarian cancer metastasisArtificial intelligence helps predict where ovarian cancer has spread

European journal of obstetrics, gynecology, and reproductive biology Published July 6, 2026 Study authors: Xiang Linmei, Li Yini, Li Mingxing, Qi XiaoYi PubMed ↗ DOI ↗ Editorial oversight: Dr. Lars van Dijk, PhD · Surgical, Procedural & Diagnostic

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

Key Takeaway

Note that combined radiomics and clinical data in AI models provide superior AUC (0.86) for predicting metastasis.

This systematic review and meta-analysis evaluated the diagnostic performance of medical image-based artificial intelligence (AI) models for the preoperative noninvasive prediction of metastasis in patients with ovarian cancer. The analysis included a sample size of 2343 patients to compare radiomics-only, clinical-only, and combined models.

The synthesis indicates that the optimal model, which combines radiomics and clinical parameters, achieved an AUC of 0.86 (95%CI: 0.83-0.89). This performance outperformed both the radiomics-only model (AUC 0.80) and the clinical-only model (AUC 0.79). The combined model also demonstrated a sensitivity of 0.81 (95%CI: 0.76-0.85) and a specificity of 0.79 (95%CI: 0.71-0.86).

Authors noted limitations including heterogeneity in study quality and variability in study quality, particularly regarding specificity. The findings suggest that while medical image-based AI models show strong performance for noninvasive prediction of metastasis, the evidence is currently limited by a lack of prospective, standardized, and externally validated studies. Clinical utility remains promising but requires further validation.

How this fits prior evidence

This meta-analysis addresses a gap in noninvasive diagnostic tools for ovarian cancer. While prior coverage has focused on biological markers like the NLRP3 inflammasome, follicular T cell subsets, and microbiome dysbiosis to understand pathogenesis or guide immunotherapy, this study provides evidence on the utility of AI and radiomics for predicting metastasis. It complements existing research by offering a technological approach to staging and risk assessment in ovarian cancer patients.

When a patient is diagnosed with ovarian cancer, doctors need to know quickly if the cancer has spread beyond the primary site. This information is vital for planning the right treatment. A large review of 2,343 patients looked at how well artificial intelligence (AI) can predict this spread using medical images.

The study compared different ways of using AI. It found that a combined model—which looks at both image data and clinical details—performed best. This combined approach showed high accuracy in identifying metastasis, which is the spread of cancer to other parts of the body. Specifically, it outperformed models that only looked at images or only at clinical notes.

While these results are promising for helping doctors make better decisions, there are some hurdles to clear first. The researchers noted that the quality of different studies varied, and more standardized tests are needed before this can be used routinely in every clinic. For now, it shows that AI has strong potential to help predict cancer spread noninvasively.

What this means for you:

AI models combining image data and clinical info show high accuracy in predicting ovarian cancer spread.

Common questions

How accurate is the AI at predicting cancer spread?

The best model, which combines image data with clinical information, showed a high accuracy score of 0.86. This combined approach performed better than models that used only images or only clinical data to predict if ovarian cancer had spread.

What kind of patients was this study based on?

The analysis included a large group of 2,343 patients who were diagnosed with ovarian cancer. This large sample size helped researchers evaluate how well AI models could predict metastasis before surgery.

Is this technology ready to be used in every hospital?

While the results are promising for clinical use, more work is needed. The study noted that because different studies had varying quality, we still need more standardized and validated tests before it can be used as a standard tool.

Study Details

Study typeMeta analysis

EvidenceLevel 1

PublishedJul 2026

PMID42107239

View Original Abstract ↓

OBJECTIVE: To conduct a systematic review and meta-analysis evaluating the diagnostic performance of medical image-based artificial intelligence (AI) models for the preoperative noninvasive prediction of metastasis in ovarian cancer (OC). METHODS: We conducted a systematic literature search in PubMed, Embase, Web of Science, IEEE Xplore, Cochrane Library, CNKI, Wanfang Data, and CQVIP for studies published before February 1 ,2026. Eligibility criteria included diagnostic studies developing AI models from medical images to predict OC metastasis, with sufficient data to construct 2 × 2 contingency tables. Two reviewers independently performed data extraction and quality assessment using the QUADAS-AI tool and Radiomics Quality Score (RQS). Heterogeneity was assessed using the I statistic. We used STATA 17.0 for meta-analysis to compute pooled sensitivity (PSen), specificity (PSpe), and area under the curve (PAUC). The protocol was registered prospectively (CRD42024619549). RESULTS: 9 studies involving 2343 OC patients were included. Study quality was moderate to high according to QUADAS-AI, with seven studies at low risk of bias in four or more domains. RQS ranged from 11 to 19 (median 18). The optimal model, integrating radiomics with clinical parameters, achieved a PSen of 0.81 (95%CI:0.76-0.85, I = 5.2%), PSpe of 0.79 (95%CI:0.71-0.86, I = 55.3%), and PAUC of 0.86 (95%CI:0.83-0.89), outperforming radiomics-only (PAUC 0.80) and clinical-only (PAUC 0.79) models. Heterogeneity for the combined model's sensitivity was low, while specificity showed moderate heterogeneity. Subgroup analyses suggested that the presence of independent validation and higher study quality (RQS ≥ 18) reduced heterogeneity. Subgroup analyses confirmed robustness across imaging modalities, algorithms, and validation methods. CONCLUSION: Medical image-based AI models demonstrate strong performance for noninvasive prediction of OC metastasis, supporting their potential clinical utility. However, heterogeneity and variability in study quality highlight the need for more prospective, standardized, and externally validated studies.

Radiomics and clinical data combined in AI models achieve 0.86 AUC for predicting ovarian cancer metastasisArtificial intelligence helps predict where ovarian cancer has spread

How this fits prior evidence

Common questions

Study Details

Machine Learning Model Performance for Predicting Contrast Induced Nephropathy in Coronary Interventions

Machine learning models help predict kidney damage during heart procedures

Clinical research that matters. Delivered to your inbox.

Radiomics and clinical data combined in AI models achieve 0.86 AUC for predicting ovarian cancer metastasisArtificial intelligence helps predict where ovarian cancer has spread

How this fits prior evidence

Common questions

More on Ovarian Cancer

Study Details

Machine Learning Model Performance for Predicting Contrast Induced Nephropathy in Coronary Interventions

Machine learning models help predict kidney damage during heart procedures

Clinical research that matters. Delivered to your inbox.

Related in Radiology & Imaging

From Other Specialties