Authors
Sanjeev Herr, Niels Olshausen, Melike Pekmezci, Jasleen Kaur, Youssef Sibih, Vardhaan Ambati, Abraham Dada, Katie Scotford, Amit Persad, Thiebaud Picart, Akhil Kondepudi, Gabrielle Malte, Gabriella Vulakh, Johanna Pechmann, Jessica Makolli, Nancy Ann Oberheim-Bush, Albert Kim, Jacob Young, Mitchel S Berger, Ammar Mallouhi, Barbara Kiesel, Georg Widhalm, Lisa Körner, Madhumita Sushil, Todd Hollon, Shawn L Hervey-Jumper
Published in
Science advances. Volume 12. Issue 39. Pages eaec8202. Sep 25, 2026. Epub Sep 25, 2026.
Abstract
A hallmark of glioblastoma (GBM) is disease recurrence that occurs in all patients despite resection, radiation, and chemotherapy. A critical challenge in GBM treatment is the management of recurrent disease for which no standard of care exists. Predicting the location of GBM recurrence may improve the efficiency of advanced-stage therapies. We present an artificial intelligence (AI)-based model to predict the recurrence risk of unprocessed surgical tissues at initial resection. AI-informed label-free optical microscopy was used to generate a normalized tumor infiltration value (AI-infiltration) for optical images of samples taken from resection cavity margins. These values, in combination with clinical, radiographic, and molecular variables, were used to build a predictive model of focal recurrence. In a cohort of 80 patients, comprising 367 samples and 133,454 unique images, GBM infiltration was significantly higher in margin samples from recurrent sites (P = 0.03) compared with those from nonrecurrent sites. A random forest machine learning classifier predicted site recurrence with an average area under the receiver operating characteristic curve of 87% ± 10.0 for the training cohort and 80% (95% confidence interval: 0.64 to 0.97) for the validation cohort. AI-infiltration was the strongest contributor to recurrence prediction, outperforming tumor molecular features. Model performance remained high regardless of tumor location, resulting in random forest model predictions of recurrence at 5 and 10 millimeters of each sample. These findings represent the potential of AI to predict sites of tumor recurrence, thereby improving accessibility to targeted, precision, and multimodal therapy for the highest-risk areas of disease.
PMID:
42789696
Bibliographic data and abstract were imported from PubMed on 26 Sep 2026.
Read full publication at:
Please sign in
to see all details.
Advertisement
Stats
- Recommendations n/a n/a positive of 0 vote(s)
- Views 26
- Comments 0