AI/ML-Derived Biomarkers from Small-Cohort and Noisy, High-Dimensional Multi-Omic Clinical Data

Prediction in medicine remains limited. The entire multi-ome affects cancer and other diseases. 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 actionable and mechanistically interpretable biomarkers 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 biomarkers in, e.g., astrocytoma, including glioblastoma (GBM), brain, lung, nerve, ovarian, and uterine cancers. The algorithms identified the biomarkers 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 biomarkers are actionable in the general population. The GBM biomarker, the first to encompass the whole genome, was additionally prospectively and retrospectively experimentally validated to be the most accurate and precise predictor of survival and response to treatment. All other attempts to associate a GBM tumor’s DNA copy-number alterations with the patient’s outcome failed, establishing that the algorithms find what all others miss, and the biomarkers outperform all others where they exist. Recent experiments show that the GBM biomarker 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.