Abstract 3686: Multi-tensor AI/ML uniquely able to identify mechanistically interpretable and actionable predictors from small-cohort and noisy high-dimensional multi-omic clinical data

Two decades after the human genome, prediction in cancer remains limited. The entire multi-ome affects the disease. And, unlike typical artificial intelligence and machine learning (AI/ML), e.g., neural networks and deep learning, our multi-tensor AI/ML is uniquely able to identify predictors from multi-omic clinical data. We have been developing the algorithms, i.e., the “multi-tensor comparative spectral decompositions,” to extend the mathematics that underlies quantum mechanics to overcome the limitations of typical AI/ML. We demonstrated the algorithms in the discovery and validation of novel predictors of survival and response to treatment in, e.g., astrocytoma, including glioblastoma (GBM), brain, lung, nerve, ovarian, and uterine cancers. The algorithms, first, blindly, i.e., without any assumptions or a-priori knowledge, removed batch effects and separated demographic variations from the data. This enabled identifying, e.g., a predictor in ovarian cancer, which is encompassed within the normally varying X chromosome. Second, the algorithms identified the predictors repeatedly, in federated and imbalanced public datasets from as few as 50-100 patients each, showing that the algorithms are batch- and demographics-agnostic, and the predictors are actionable in the general population. Third, the predictors were interpreted in terms of known and new disease mechanisms and drug targets and combinations of targets to sensitize the tumors to the treatment. The predictors are independent from all standard-of-care indicators, additionally showing that this information that they contain was missing from the understanding and management of cancer. Fourth, the GBM predictor, which is the first biomarker to encompass the whole genome, was prospectively and retrospectively experimentally validated to be the most accurate and precise. All other attempts to associate a GBM tumor’s DNA copy numbers with the patient’s outcome failed. For 70 years, the best indicator has remained age at diagnosis, establishing that the algorithms find what all others miss, and the predictors outperform all others where they exist. Here, we present recent experiments demonstrating that, fifth, the GBM predictor correctly identified previously unrecognized drug targets, i.e., genes that are required for the tumor’s cell proliferation and viability, further validating the mechanistic interpretability of the multi-tensor AI/ML. Guide RNAs were designed and a lentiviral CRISPR-Cas9 all-in-one vector was harnessed to knock out the candidate genes. Knockout validation at the protein level was performed using Western blot. Knockout in patient-derived GBM cell lines resulted in significantly attenuated cell viability and proliferation. We conclude that our multi-tensor AI/ML is uniquely suited for personalized medicine. Citation Format: Orly Alter, Marissa Coppola, Tessa O. House, Jessica W. Tsai. Multi-tensor AI/ML uniquely able to identify mechanistically interpretable and actionable predictors from small-cohort and noisy high-dimensional multi-omic clinical data [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2025; Part 1 (Regular Abstracts); 2025 Apr 25-30; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2025;85(8_Suppl_1):Abstract nr 3686.