Noetik is leveraging artificial intelligence to address the high failure rate of cancer clinical trials, which often result from poor patient-drug matching. During a recent discussion, CEO Ron Alfa and VP of Research Daniel Bear highlighted their innovative approach, focusing on their model TARIO-2, which is trained on extensive tumor spatial transcriptomics datasets. This model enhances patient selection for trials by generating valuable spatial transcriptomics predictions directly from routine pathology slides, thereby improving access to advanced tumor profiling and supporting the crucial research pipeline that Noetik has invested in for years.

Noetik: Noetik is an AI-native biotechnology company leveraging advanced machine learning to discover and develop cancer immunotherapies through foundation models of tumor biology. The company focuses on key clinical bottlenecks like patient selection by training large autoregressive transformers on spatial transcriptomics datasets. It recently featured its TARIO-2 model in a podcast and blog post explaining its role in improving cancer trial success rates.
TARIO-2: TARIO-2 is Noetik’s autoregressive transformer foundation model trained on extensive tumor spatial transcriptomics data to predict whole-transcriptome maps from standard H&E slides. This enables routine biopsies to be analyzed in the same biological space as specialized assays, supporting patient stratification and biomarker discovery. The model was detailed in Noetik’s recent blog post and discussed in an AI for Science podcast.
Ron Alfa: Ron Alfa is the co-founder and CEO of Noetik, a Stanford-trained physician-scientist and former Recursion Pharmaceuticals executive. He leads the company’s efforts to decode cancer using AI foundation models, including ongoing tumor sample processing for model training. Alfa recently appeared on the Latent Space podcast to discuss strategies for overcoming cancer trial failures through better patient selection.
Daniel Bear: Daniel Bear is Vice President of AI Research at Noetik, with expertise in machine learning for biological applications. He oversees the development of large cancer foundation models like TARIO-2. Bear co-hosted a recent Latent Space AI for Science podcast episode explaining Noetik’s transformer-based approaches to patient selection in clinical trials.

Trial Challenge: Poor patient-drug matching drives high failure rates in cancer clinical trials, addressed by AI-driven stratification tools.
Model Innovation: TARIO-2 generates spatial transcriptomics predictions from routine H&E pathology slides, broadening access to advanced tumor profiling.
Research Pipeline: Noetik invests in collecting tumor spatial data to enable training of expansive cancer foundation models.