Real-world clinical performance of a combinatorial functional precision platform, Optim.AI™, in hematological malignancies

Background: Ex vivo drug sensitivity testing has been explored for nearly five decades as a strategy to individualize treatment, especially for patients without actionable biomarkers or who have exhausted standard options. Historically, adoption has been limited by challenges such as scarce tumor samples, difficulty recreating indication-specific microenvironments, and inconsistent correlation with clinical outcomes. Advances in functional precision medicine now enable more reliable translation. Optim.AI™, a combinatorial functional precision medicine platform, has shown prospective clinical utility in hematologic malignancies and sarcoma. Here, we report real-world performance data from its use in a clinical-certified laboratory for hematological cancers. Methods: Tumor cells from solid tissue, peripheral blood, or bone marrow aspirates were isolated and exposed to 12 FDA-approved chemotherapy and targeted agents in combinatorial formats. Post-treatment cell viability was quantified to generate Optim.AI™ rankings of actionable treatment combinations. Clinical outcomes were retrospectively assessed for patients who subsequently received Optim.AI™-guided therapies. Results: Among 154 hematological samples, 91% yielded sufficient cells for testing and 94% of these produced successful reports. Concordance with prior resistance patterns was high: 88% of cases demonstrated predicted normalized cell viability (NCV) > 0.6, consistent with clinically observed resistance. Retrospective analysis of five acute myeloid leukemia (AML) patients treated according to top-ranked Optim.AI™ recommendations showed that the platform was clinically useful in all five cases. Three patients responded to Optim.AI™-guided therapy, including two complete remissions-one successfully bridged to transplant. For the two patients who did not respond, the Optim.AI™ profiles accurately predicted non-response (NCV > 0.6) and supported timely decisions to limit further futile therapy, including transition to palliative care. Across these cases, NCV < 0.3 effectively stratified responders from non-responders. Conclusion: Consistent with prior prospective studies, these real-world results validate Optim.AI™'s ability to predict treatment responses in hematological malignancies. Importantly, the platform demonstrated clinical utility in all evaluated patients-by identifying effective therapeutic options when available and by guiding physicians toward appropriate palliative approaches when further intensive therapy was unlikely to help. Larger prospective studies across diverse indications will strengthen its path toward broader clinical adoption.