OWKIN

Just published in Computers in Biology and Medicine, our latest collaborative study with Memorial Sloan Kettering Cancer Center harnesses multimodal ML to enhance survival predictions in intrahepatic cholangiocarcinoma (iCCA). 🔬 Combining clinical, histological, radiological, and genomic data leads to superior outcome predictions. Our models achieved C-Index up to 0.70 ± 0.13, outperforming traditional methods. 📈 These models with interpretable features may enhance understanding of the disease and guide treatment selection. This research effort highlights the power of integrating diverse data sources for superior prognostic accuracy. Read the paper: https://lnkd.in/e-AiUjuM Thanks to the authors: Eliott Brion, Valérie Ducret, Naaz Nasar, Benoît Sauty, Sarah McIntyre, Remo Alessandris, Carlie Sigel, Mala Jain, Umesh Bhanot, MD PhD 🇺🇸 🇮🇳, Jordana Ray-Kirton, Joachim Silber, Caroline Hoffmann, Charles Maussion, Jayasree Chakraborty, Benoît Schmauch, William Jarnagin

🚀 Our latest newsletter is here! Discover the launch of K Pro at ESMO, our new State of the Nation report with STAT, an inside look at the Decoding Biology Hackathon, and our workshop with Gustave Roussy kicking off the K Pro trial.

Preprint on bioRxiv: Unveiling the cellular dialogues within tumors using COMPOTES. 🧬 🔬 Analyzed 1,084 cancer samples from MOSAIC research to reveal conserved and cancer-specific cell communication patterns. 📈 Identified 45 communication programs, highlighting potential therapeutic targets in cancer. This research enhances understanding of tumor dynamics and immune evasion. Read the pre-print: https://lnkd.in/eJCkhKRZ Thanks to our MOSAIC partners Gustave Roussy, CHUV | Lausanne university hospital, Universitätsklinikum Erlangen, Charité - Universitätsmedizin Berlin, University of Pittsburgh, UPMC Hillman Cancer Center for their contribution with the data generation.

📄 Where can Agentic AI add the most value? We asked 202 pharmaceutical industry executives and scientific specialists this question, as part of our State of The Nation report - Biopharma’s agentic AI moment - in conjunction with STAT. 41.6% of respondents believe agentic AI adds the most value by addressing data challenges like harmonization, quality control, and interoperability. Explore the full findings in the report: https://lnkd.in/evWFRqMJ We conducted this research as part of the launch of Owkin’s enterprise-grade agentic software, K Pro. You can book a K Pro demo here: https://lnkd.in/enMB9mRv

🎀 In honor of #BreastCancerAwarenessMonth, Elodie Pronier, our VP of Biomedical Discovery and a member of Jeune et Rose, a community supporting young breast cancer patients, is raising awareness. At a workshop hosted at Owkin this week, she walked through self-examination steps, discussed common and uncommon symptoms, and encouraged open questions.

Our latest collaborative research with Greater Paris University Hospitals - AP-HP integrates machine learning with clinical data, histological slides, and spatial transcriptomics to uncover novel biomarkers in glioblastoma survival. 🔬 We analyzed histological data from 748 patients to identify tissue regions predicting survival. 📈 Our model achieved an AUC of 0.85, outperforming clinical and histology-only models. Model interpretation and detailed biological insights were obtained from the glioblastoma cohort of the MOSAIC research dataset. Thanks to Universitätsklinikum Erlangen and CHUV | Lausanne university hospital for their support on data generation and annotation. These findings improve our understanding of glioblastoma biology and could enhance risk stratification. Read the pre-print on arXiv: https://lnkd.in/g49UrFeb

💬 Great feedback on K Pro from Ben Mellows Ph.D, Managing Director at MICREGEN LTD: “We wanted to get our product pitch across to investors. That process could take weeks to work through with all of our experts, but Owkin K really helped. It assimilated all the information we provided evidencing how our regenerative stem cell-derived platform and products could have a major impact on healthcare and delivered a clear story tailored to investors from different backgrounds. The literature set is huge. I would say the task would have been nearly impossible to achieve at the scale that Owkin K handled in just a few work sessions.” Check out our latest blog where Ben Mellows Ph.D, shares how Owkin K is transforming biotech workflows — from overcoming research challenges to streamlining collaboration and identifying who can benefit most from the platform. Read the interview here: https://lnkd.in/e5x-TxuX

We’re excited to be sponsoring the Research & Innovation Healthcare Conference next week Hear from Matthew Howe, PhD, our Principal Partnerships Manager for North America, and Charley Macedo from Amster Rothstein & Ebenstein LLP to discuss how AI is transforming healthcare and driving smarter, more cost-effective decisions. 📅 Date & Time: Monday, October 27, 2025 at 11:10am EDT 📌 Topic: AI in Healthcare - From Data to Decisions: How AI is Saving Lives and Lowering Costs 📍 Location: Mass General Brigham, Somerville, MA Register now: https://lnkd.in/eR_evNNa

In our new preprint on bioRxiv, we introduce SpaMIL, a novel Multiple Instance Learning (MIL) framework for predicting patient-level outcomes from spatial transcriptomics data. Applying SpaMIL to the glioblastoma MOSAIC dataset, we demonstrate improved survival prediction and show that: 🔬 abMIL on spatial transcriptomics outperformed all other data modalities with a median C-index of 0.72. 📈 MabMIL (Multimodal abMIL) enhances H&E-only models by incorporating spatial transcriptomic signals. These methods are promising avenues for developing robust, interpretable and clinically applicable prognostic tools that harness spatial omics data. Read the pre-print: https://lnkd.in/ex_bigDK Thanks to the MOSAIC research Consortium and in particular to Universitätsklinikum Erlangen and CHUV | Lausanne university hospital for their contribution on the data generation.

At the ESMO annual congress, we showcased our advancements in ADC development through spatial biology and AI in our e-poster presentation. 🔬 Utilizing single-cell and spatial features, the MOSAIC dataset not only recalls known ADC targets but also uncovers new candidates. 📈 Our newly launched agentic AI co-pilot for biopharma, K Pro, pinpoints bladder cancer patients likely to benefit from Enfortumab Vedotin. This research underscores the potential for improved patient selection in ADC trials. We extend our gratitude to MOSAIC research collaborators—Gustave Roussy, CHUV | Lausanne university hospital, Universitätsklinikum Erlangen, Charité - Universitätsmedizin Berlin, University of Pittsburgh. As ESMO 2025 ends, you can explore K Pro further—If you missed the chance to see it in action, schedule a demo with us here: https://lnkd.in/enMB9mRv.