Faisal Mahmood

⚡️🔬📣 Excited to share our new @Nature article building and evaluating PathChat, a multimodal generative AI copilot and chatbot for human pathology. Article: nature.com/articles/s4158… Open Access Link: rdcu.be/dKC0r We leverage our previous success in building foundation models for computational pathology such as UNI / CONCH and combine it with the advancements of large vision language models and generative AI to enable PathChat to answer diverse pathology-related queries. We assessed PathChat using both multiple choice diagnostic questions and open-ended questions. Congratulations to @MYLu97 @chenbowen118 @DFKW_MD @richardjchen and everyone else who contributed to this work. Also see blog post from @MYLu97 about this work: linkedin.com/pulse/towards-… , also teasing the development and preview of PathChat 2, a successor to PathChat 1 bringing new capabilities and substantially improved performance to the state-of-the-art.

Always grateful for my academic family!

Excited to share our latest work, "Mixture of Mini Experts: Overcoming the Linear Layer Bottleneck in Multiple Instance Learning," to appear at ICLR 2026 @iclr_conf This study tackles an overlooked but critical bottleneck in computational pathology: the linear layer that connects foundation models and multiple instance learning. The Problem: Standard MIL pipelines rely on a familiar three-step process: patch feature extraction, feature transformation with a linear layer, and aggregation for slide-level analysis. While we have made substantial progress on the first and last steps, we identify the intermediate linear layer as a major, hidden bottleneck. Tissue morphology is highly complex, and processing this heterogeneity through a "one-size-fits-all" linear transformation severely limits model performance. Our Solution: Mammoth 🦣. We replace the single linear layer with a parameter-efficient, low-rank Mixture of Experts (MoE). Mammoth uses a soft routing mechanism to assign each patch to a combination of mini-experts tailored specifically to its phenotype. Key Highlights: 🔹 Plug-and-Play: Mammoth is a drop-in replacement for the first linear layer in any MIL method (e.g., CLAM, ABMIL, etc). 🔹 Consistent Gains: Evaluated across 8 MIL methods and 19 classification tasks, Mammoth improved performance in 130/152 configurations with an average boost of +3.8%. 🔹 Parameter efficiency: By utilizing factorized (LoRA-style) linear layers, the parameter count stays comparable to a standard linear layer. 🔹 Structured features: The MoE routing yields a highly structured feature space where experts naturally specialize in diverse morphological phenotypes. Incredible work by Daniel Shao, co-advised with Andrew H. Song and myself, and the rest of the team: Joel Runevic, Richard J. Chen, Drew F. K. Williamson, Ahrong Kim. Read the full paper and try it out in your own MIL pipelines below 👇 📄 Paper: openreview.net/pdf?id=S5Io33p… 🔬 OpenReview: openreview.net/forum?id=S5Io3… 💻 Code: github.com/mahmoodlab/mam…

We are thrilled to host the 3rd Workshop on Foundation Models for Medical Vision (FMV) at #CVPR2026. @CVPR fmv-cvpr26workshop.github.io Join us as we explore the next frontier of medical foundation models and agentic systems with an incredible lineup of visionaries: Dr. Jakob Nikolas Kather @jnkath Dr. Faisal Mahmood @AI4Pathology Dr. Hoifung Poon @hoifungpoon Dr. Pranav Rajpurkar @pranavrajpurkar Can’t wait to connect with the community this summer! Special thanks to the organizing team: @JunMa_AI4Health @yuyinzhou_cs @vishalm_patel

#REVIEW 🚨 #AI agents perform complex tasks with minimal human oversight. This Review by @jnkath, @AI4Pathology, @james_y_zou et al. discusses how AI could transform oncology & the challenges to address for efficient & safe use. 👇 📖 dlvr.it/TQRQbw

Apply and join us for the Ascona workshop "Statistical and AI methods for multi-modal multi-scale modeling of biological systems", 28 June - 3 July, 2026, on Monte Verità, Lago Maggiore, at the foot of the Swiss Alps. LINK: ascona2026.sciencesconf.org Our invited speakers: Maria Brbic @mariabrbic Charlotte Bunne @_bunnech Faisal Mahmood @AI4Pathology Dana Pe’er Barbara Engelhardt @BeEngelhardt Caroline Uhler @CarolineUhler Julien Gagneur Marinka Zitnik @marinkazitnik Julie Josse Basile Wicky @BasileWicky Fabian Fröhlich Enjoy a great program and a beautiful view of the lake in the Swiss Alps! Organized by Niko Beerenwinkel, Wolfgang Huber, Peter Bühlmann, and me.

.@modella_ai the very first startup to spin out of our lab has been acquired by @AstraZeneca Congratulations to @richardjchen @MYLu97 and everyone else from the Modella team. reuters.com/legal/litigati…

We're excited to host a distinguished lineup of speakers at @CellSymposia #CSHallmarks2026, including keynotes Johanna Joyce @Joycelab & Faisal Mahmood @AI4Pathology View the full speaker lineup: hubs.li/Q03VXZll0

Deadline for PhD program for AI in Medicine @HarvardMagazine @HarvardDBMI is December 1st bmiphd.hms.harvard.edu/aim-track/welc…

From single-cell AI-driven 3D spatial transcriptomics to whole-block volumetric pathology to Apollo’s patient-level phenotyping — he somehow makes incredibly #complex ideas feel #intuitive. Real translational AI that we all aspire to build.🚀 @myESMO @AI4Pathology #ESMOAI25 (2/2)

Absolutely phenomenal keynote by @AI4Pathology at #ESMOAI25 today. What stands out is not just the science — it’s the ⚡️full-stack translation. To build models, validating them, open-sourcing, running prospective clinical trials, and taking it to regulatory and FDA @myESMO (1/2)

Injection of patient-aligned AI bill reviewing will create new incentives that will generate new market forces. Opportunity: Regulations that widen the aperture for AI-driven patient-aligned value/$ perspective. CC @IAmSamFin @zittrain

Starting now (9:00am - 5:30pm)!! Join us for the @ICCVConference full-day workshop Computer Vision for Automated Medical Diagnosis (CVAMD) (cvamd.github.io/CVAMD2025/) 📅 Oct 19 | 9:00–18:00 📍 Hawaii Convention Center, Room 314 Featuring leading voices in medical AI: @ShengLiu_ @Dr_ASChaudhari @AI4Pathology @lx2015 Daguang Xu @yeung_levy @YalaTweets @james_y_zou @kyrie0109 Qingyue Wei, Yi Lin, @ylqzd2011 @yang_yuzhe, Angelica Aviles-Rivero, Haochen, @p_tingying, @pengyifan, @VITAGroupUT 🔗 Full schedule: cvamd.github.io/CVAMD2025/

Congratulations 🥳 to our superstar @GreatAndrew90 he’ll be starting his own research group at MD Anderson Cancer Center. Andrew pushed the frontier of computational pathology and medical AI in our lab with work including: • TriPath - Cell (2024): cell.com/cell/fulltext/… • PANTHER - CVPR 2024: arxiv.org/pdf/2405.11643 • MMP - ICML 2024: arxiv.org/pdf/2407.00224 • TITAN - preprint: arxiv.org/pdf/2411.19666 • VORTEX - preprint: arxiv.org/pdf/2502.17761 Really proud of everything you accomplished and can’t wait to see what you do next at MD Anderson. #ComputationalPathology #MedicalAI #CancerResearch #AIinHealthcare

It was a pleasure hosting @hoifungpoon at the lab today!

📣 If you're at @icmlconf ICML 2025, our superstar graduate student @Daniel__Shao will be presenting our work on transferability of MIL models and supervised foundation models for computational pathology. Poster session: 11 am - 1:30 pm PST at West Exhibition Hall B2-B3 #W-314 Read the article: arxiv.org/pdf/2506.09022 Code and models: github.com/mahmoodlab/MIL…

Happy 4th of July 🇺🇸 Celebrating our ideals of freedom, unity & opportunity that bind us all. To our service members, veterans & first responders who safeguard our freedoms, thank you for your dedication. Wishing everyone a safe, joyful Independence Day! #FourthOfJuly

We just dropped a new FL library to support researchers and organizations grappling with Federated Learning projects. Syft_Flwr combines Flower’s flexibility with the privacy-preserving networking capabilities of SyftBox. Links in the thread ↓

Always grateful for my academic family!

From protein structure and disease modelling to drug discovery and clinical imaging, #AI is redefining how we approach many biological challenges. Hear from experts working at the forefront of #AIforBiology at #AIxBIo25! ⏰Register by 27 May: bit.ly/4gh164E