Applied Scientist, Oncology Foundation Model (Intern)

Pathos AI enters collaboration and co-exclusive licensing agreement to advance novel erα PROTAC for ER+ breast cancer into the clinic. * 4 hrs ago Pathos AI, a clinical-stage AI and technology company advancing its own pipeline of cancer therapies, today announced a collaboration and co-exclusive licensing agreement with AstraZeneca. Under this agreement, the companies will advance AZD4241, a preclinical investigational therapy for patients with estrogen receptor-positive, human epidermal growth factor receptor 2-negative (ER+/HER2-) breast cancer, with Pathos assuming responsibility for the early clinical development of the asset. AZD4241 is an investigational oral therapy known as a PROTAC (proteolysis-targeting chimera), specifically designed to degrade the estrogen receptor, targeting both normal and mutated forms of the receptor. Mutated forms of the estrogen receptor are present in up to 30-40% of patients whose disease progresses on standard hormone therapies. Powered By Keir Starmer arrives in Ankara for Nato summit Translating the full potential of this mechanism into patient benefit requires a fundamentally different approach. By harnessing large-scale multimodal patient data and a proprietary AI platform, Pathos redesigns drug development end-to-end. The company's capabilities span early asset identification, data driven portfolio optimization, and AI-native trial monitoring, with the specific early clinical execution for AZD4241 driven directly by its Foundry platform. Accelerated by AI: The Foundry Advantage In complex indications like ER+ breast cancer, identifying the exact patient subpopulation, optimal dose, and ideal treatment setting is a massive data challenge. Foundry was built to solve this problem. Foundry utilizes thousands of AI agents working in parallel, powered by the Pathos Oncology Foundation Model. These agents analyze complex biological, clinical, and real-world data to intelligently design clinical trials, optimize dosing, and match the right patients to the right therapies faster and more safely than traditional methods allow. "The bottleneck in oncology is not finding molecules. It is proving they work in the right patients. AZD4241 has a compelling mechanism. Foundry's job is to design the trial that proves it - matching this drug to the patients whose biology demands it. That is how we compress time. Patients are waiting." - Iker Huerga, CEO of Pathos AI A Growing Mission to Transform Oncology This collaboration and co-exclusive licence agreement to bring AZD4241 into the clinic expands Pathos' clinical pipeline to three major assets across four distinct tumor types. AZD4241, targeting ER+/HER2- breast cancer, joins pocenbrodib, currently in development for metastatic castration-resistant prostate cancer (mCRPC) and relapsed/refractory multiple myeloma (RRMM), and DO-2 for MET-altered non-small-cell lung cancer (NSCLC). Every program in the Pathos pipeline has been identified, evaluated, or accelerated by the Foundry platform, reinforcing the company's commitment to AI-native drug development. About Pathos AI Pathos AI is an AI and technology company advancing its own pipeline of cancer therapies. Its Foundry platform redesigns drug development from first principles: identifying undervalued clinical assets, making portfolio decisions with continuously updated probability-of-success calculations, and executing clinical trials with AI-native monitoring. Pathos's pipeline includes clinical-stage programs in mCRPC, multiple myeloma, and MET-altered NSCLC. Forward-Looking Statements This press release contains forward-looking statements. These statements involve risks and uncertainties that may cause actual results to differ materially. Pathos AI undertakes no obligation to update any forward-looking statement. Media gallery Sections.

GSK, Relation to build cellular datasets for AI drug targets. GSK agreed a research collaboration with Relation Therapeutics worth up to $110 million for large-scale human cell datasets and AI model training to identify drug targets. GSK has signed a research collaboration with Relation Therapeutics worth up to $110 million. Relation will generate large-scale datasets that measure how human cells respond to genetic changes and drug interventions, and will use those data to train AI models for drug-target discovery. Under the agreement, Relation will produce human cellular datasets and apply them to models within its MORGAN platform. The company's Lab-in-the-Loop approach links computational analysis with laboratory experiments. The planned work includes tissue profiling, single-cell and spatial transcriptomics, sequencing, functional assays and perturbation experiments that record how specific genetic changes alter cell states relevant to disease. The collaboration builds on earlier GSK-Relation projects focused on fibrotic diseases and osteoarthritis that created two functional disease datasets. Relation has also developed Osteomics, a proprietary functional single-cell bone atlas that combines patient-derived samples with single-cell and spatial omics, imaging, genomics, proteomics and clinical phenotype data. Hospitals and research partners in the UK and Australia are participating in that study. Public repositories provide large volumes of single-cell data, but differences in sampling, sequencing and processing pipelines can introduce technical noise and batch effects. Dataset overlap across resources can create risks of data leakage when training and test sets share similar cells. A recent study that trained 400 models on a corpus of 22.2 million cells and evaluated them across thousands of experiments reported that performance often plateaued after training on a fraction of the data and did not show the clear data-scaling behavior seen in some other AI fields. A separate assessment of two single-cell foundation models reported inconsistent performance gains over simpler approaches and highlighted ongoing challenges from batch effects and dataset composition. Pharmaceutical companies are seeking disease-specific, curated datasets through partnerships, licensing and joint development. Examples include a separate GSK agreement with Ochre Bio for human liver single-cell and perfused-organ data and a multi-party deal involving AstraZeneca, Pathos AI and Tempus to develop oncology models using de-identified clinical, genomic and imaging data. Limited access to suitable training data has been identified by industry participants as a constraint for AI applications in drug discovery. The GSK-Relation collaboration combines biological data generation with model development so the experimental datasets produced by Relation can be used directly to train and test AI models for target identification, prioritisation and validation. The companies plan to produce disease-focused single-cell datasets alongside machine-learning workflows to align laboratory measurements with computational predictions for drug-target discovery. Content on BlockPort is provided for informational purposes only and does not constitute financial guidance. Blockport Company strive to ensure the accuracy and relevance of the information Blockport Company share, but Blockport Company do not guarantee that all content is complete, error-free, or up to date. BlockPort disclaims any liability for losses, mistakes, or actions taken based on the material found on this site. Always conduct your own research before making financial decisions and consider consulting with a licensed advisor. For further details, please review its Terms of Use, Privacy Policy, and Disclaimer.

Tempus and Bristol Myers Squibb expand AI collaboration across oncology and neuroscience programs. May 20, 2026 Tempus AI (Nasdaq: TEM) and Bristol Myers Squibb (NYSE: BMY) announced an expansion of their existing partnership, applying Tempus's Lens AI-enabled analytical platform and multimodal real-world data library to five initial BMS clinical trial programs across solid tumor oncology and Alzheimer's disease. The Tempus Bristol Myers Squibb collaboration builds on a prior arrangement centered on the Next Pathways program for advanced non-small cell lung cancer and now extends the analytical scope to lung, colon, and prostate cancers, as well as neuroscience. The expanded agreement tasks Tempus with applying its Lens platform to optimize clinical trial design and improve the Probability of Technical and Regulatory Success across BMS's development programs. Lens draws on Tempus's library of de-identified multimodal records - combining clinical data with molecular and genomic profiles - to characterize patient populations, pressure-test trial assumptions, validate control-group parameters, and identify patient segments with the highest likelihood of responding to investigational therapies. The collaboration covers five initial BMS clinical programs. In oncology, the work spans solid tumor indications including lung, colon, and prostate cancers. In neuroscience, the arrangement extends to Alzheimer's disease drug development, making this one of the broader therapeutic applications of the Tempus platform disclosed to date. No specific BMS drug candidates, program names, trial identifiers, or ClinicalTrials.gov NCT numbers were disclosed in connection with the agreement. Tempus describes Lens as an AI-enabled analytical environment built on what the company characterizes as one of the largest multimodal real-world data libraries in healthcare. The platform links de-identified clinical records to molecular subtypes, enabling analyses that the company says go beyond what traditional trial-design methods support. In the context of the BMS collaboration, Lens is being used not to generate drug candidates or run wet-lab experiments, but to inform development decisions - specifically, to assess whether trial assumptions are well-founded before studies are initiated or advanced. The use of real-world data in pharma development to validate control-arm assumptions and refine patient-selection criteria is an established practice, but the integration of clinicogenomic data at the scale Tempus describes represents a more recent application of this approach. Ryan Fukushima, CEO of Data and Apps at Tempus, said the platform allows the companies to "connect the dots between clinical records and molecular subtypes" and to "uncover the hidden biological signals of unmet patient need." Bryan Campbell, Senior Vice President of Drug Development Strategy and Innovation at BMS, said the collaboration allows the company to "rigorously pressure-test trial assumptions, better understand patient heterogeneity, and design studies with a higher probability of technical and regulatory success." Neither company disclosed financial terms associated with the expanded agreement. Prior collaboration and deal context. The current expansion builds on an earlier Tempus-BMS arrangement involving the Next Pathways program, which deployed Tempus's platform across 13 community-based health systems to identify care gaps for patients with advanced non-small cell lung cancer. That prior work appears to have served as the foundation for the broader clinical-development analytics relationship now being disclosed. Tempus has been accelerating its partnering activities in recent years, including a partnership with AstraZeneca and Pathos AI in 2025 to co-develop a multimodal oncology foundation model, with Gilead in April 2026 focused on oncology assets, and with Daiichi Sankyo in March 2026 for antibody drug conjugate biomarker discovery. This article was generated with AI assistance and reviewed and edited by the AllSci editorial team Explore more at AllSci News: https://allsci.com/news/