Waiv, formerly Owkin Dx

Conventional IHC tells you whether a target is expressed. It says far less about whether a patient will respond. That gap is where AI pathology is starting to earn its place in drug development: reading more from the H&E and IHC slides teams already have, without burning scarce tissue. But where is it actually changing decisions today? And what does it take to turn a promising model into a regulatory-grade companion diagnostic? We're bringing together pharma leaders from Amgen, Regeneron and Modella AI/ AstraZeneca for a live panel with Pathology News, moderated by our CEO Meriem Sefta 🎙️ Advances in Multimodal AI: de-risking drug development from discovery to CDx 📅 Tuesday, November 3 🕔 5pm CET | 4pm GMT | 11am EST We'll cover: → What AI can read beyond target expression → The evidence bar a computational biomarker has to clear → Closing the gap from model to clinical test 👉 Register here: https://lnkd.in/ePnDkvaC #DigitalPathology #AIPathology #DrugDevelopment #Biomarkers #CompanionDiagnostics #PrecisionMedicine #Oncology #Waiv

📢 LIVE PANEL | Advances in Multimodal AI: de-risking drug development from discovery to CDx AI pathology is moving fast, but how much of it is translating into better decisions in drug development? Our CEO & Co-founder Meriem Sefta sits down with precision medicine leaders from Amgen, Regeneron and Modella AI (a member of the AstraZeneca Group) for an open conversation on what's working today, what's still a work in progress, and what comes next. 🔍 On the agenda: - Where AI pathology delivers today, and where it's still maturing - Use cases across ADCs, IO and bispecifics - What it takes to go from an AI biomarker to a CDx 🎙️ Panelists: Michael C. Montalto, PhD - Vice President of Precision Medicine, Amgen Jennifer Davidson Hamilton - VP & Head, Precision Medicine, Regeneron Jill Stefanelli - President & CEO, Modella AI (a member of the AstraZeneca Group) Meriem Sefta - CEO & Co-founder, @Waiv, formerly Owkin Dx 🗓️ Tuesday, November 3 🕔 5pm CET | 4pm GMT | 11am ET | 10am CT | 8am PT Register here 👉 https://lnkd.in/gAkvhJDM #MultimodalAI #DigitalPathology #PrecisionMedicine #CompanionDiagnostics #DrugDevelopment #Waiv

🎙️ We're really excited about next week's panel. We have four expert precision medicine leaders and a list of questions we’ve been dying to ask... 👉Swipe through to meet the panel for our webinar with Pathology News, Advances in Multimodal AI: De-risking Drug Development from Discovery to CDx: Michael C. Montalto, PhD, VP Precision Medicine at Amgen Jennifer Davidson Hamilton, PhD, VP & Head, Precision Medicine at Regeneron Jill Stefanelli PhD, President & CEO at Modella AI (an AstraZeneca Group company) Meriem Sefta, PhD, our very own CEO & Co-founder at Waiv, formerly Owkin Dx 💡 Here’s a taste of what we’ll get into: What can AI read from a routine H&E or IHC slide that conventional pathology can’t? Where is AI pathology already reducing risk in clinical trials, and where is it headed? How is an AI-derived biomarker used to enrich or stratify a trial? When should you build AI pathology capabilities in-house, and when should you partner? Got a question of your own? Bring it along and we’ll put it to the panel. 📅 Tuesday, November 3 🕔 5pm CET / 11am ET / 8am PT 💻 Online - live panel 🔗 Save your spot: https://lnkd.in/e_7BkrEC #DigitalPathology #DrugDevelopment #PrecisionMedicine #Waiv #CompanionDiagnostics

Our poster is live at ESMO! 🎉 Estelle Hocquet is presenting Poster 325P today, sharing new data from TAILORx, the pivotal trial that established the 21-gene Recurrence Score as the standard test for chemotherapy decisions in HR+/HER2− node-negative early breast cancer. In a prospective-retrospective analysis of 7,585 patients, the H&E-only component of RlapsRisk® BC was compared head-to-head with the 21-gene Recurrence Score, and in combination with it, for predicting distant recurrence: ✅ From the routine H&E slide alone, the AI biomarker showed prognostic performance at least equivalent to the 21-gene Recurrence Score (C-index 0.676 vs 0.673) ✅ The two scores provide independent, complementary information, and combining them gave the strongest prognostic performance (C-index 0.701) ✅ The prognostic value was consistent in both endocrine-only and chemotherapy-treated patients These results support combining AI pathology with genomic testing to refine risk assessment in HR+/HER2− early breast cancer. A huge thank you to our co-authors, ECOG-ACRIN, and the International Immuno-Oncology Biomarker Working Group for making this work possible. 📍 Poster 325P: Head-to-head and combined performance of an AI-based H&E biomarker and the 21-gene recurrence score for distant recurrence in HR+/HER2− early breast cancer: validation in TAILORx 🔗 Read the poster: https://lnkd.in/eb8t9wz9 Want to talk through the data? Find us at Booth #1007 until Monday evening.

Germline BRCA1/2 testing in HR+/HER2− breast cancer is typically offered based on age, family history, and tumor subtype. Population studies estimate that these criteria miss approximately half of carriers. These patients may lose access to PARP inhibitor therapy and whose relatives remain unaware of their own risk. Our paper on BRCAura, published last week in npj Precision Oncology, asked whether the diagnostic H&E slide every patient already has could serve as a pre-screening tool to prioritize who should undergo germline sequencing. BRCAura is a multiple instance learning model built on an in-house foundation model that reads routine H&E slides and flags cases unlikely to carry a germline BRCA1/2 mutation. We validated BRCAura on 1,025 patients from six centers in Europe and the US. To our knowledge, this is the largest external validation of an AI-based germline BRCA classifier in early breast cancer. Key results: 🔹 Sensitivity above 93% while ruling out about one third of patients from unnecessary sequencing, 🔹 High analytical repeatability across staining protocols (ICC 0.99) and strong inter-scanner agreement of prediction scores, 🔹 Significantly outperformed established clinical risk models, including CanRisk, using the H&E slide alone. Beyond performance, we conducted an interpretability analysis mapping which regions of each slide drove predictions. High-scoring regions matched known features of BRCA-associated tumors, including high cellularity, increased mitotic activity, and dense lymphocytic infiltration. Higher tumor cell and lymphocyte density was confirmed across all six cohorts. Thank you to my co-authors, our partner centers, and the team at Waiv, formerly Owkin Dx. Elin Samuelsson, Damien Jacobs, Axel Boulet, Katharina von Loga, MD, Estelle Hocquet, Paraic Kenny, Thomas Holdbrook, Javier Hernández Losa, Sheeba Irshad, ferroudja daidj, Andrew Whittum, Cecile Badoual, Séphora Campoy, Christine LASSET, Catherine Chassagne-Clément, fabrice andre, Ingrid Garberis, Olivier Caron & Magali Lacroix-Triki Gustave Roussy, Centre Léon Bérard, Vall d’Hebron Institute of Research (VHIR) - Vall d'Hebron Institut de Recerca, King's College London, Guy's and St Thomas' NHS Foundation Trust, Gundersen Health System, Cooper Norcross Health 📝 Blog: https://lnkd.in/ecRySAZ9 📄 Paper: https://lnkd.in/eBQ7ymwW BRCAura™ RUO is for research use only and is not for use in diagnostic procedures. #DigitalPathology #BreastCancer #AI #PrecisionOncology #Waiv #OctobreRose

October is Breast Cancer Awareness Month. For patients who carry a germline BRCA1/2 mutation, knowing their genetic status can shape treatment decisions and alert relatives to their own risk. Genetic testing is usually offered based on age, family history, and tumor subtype. These criteria are estimated to miss about half of carriers, particularly in HR+/HER2− breast cancer. Our team, led by Oussama Tchita, recently published a study in npj Precision Oncology on BRCAura, a deep learning model that analyzes the routine H&E slide taken at diagnosis. It identifies slides unlikely to carry a mutation, and the rest are flagged for confirmatory sequencing. In a retrospective study of 1,025 patients from six centers in Europe and the US, BRCAura kept sensitivity above 93% while ruling out about one third of patients from sequencing. Thank you to the co-authors and partner centers who made this work possible. 📝 Read the blog and paper here : https://lnkd.in/eFnXFtsq #BreastCancerAwarenessMonth #OctobreRose #DigitalPathology #BreastCancer #PrecisionOncology BRCAura™ RUO is for research use only and is not for use in diagnostic procedures.