Tempus and Daiichi Sankyo Partner on AI-Guided Cancer Biomarker Discovery
Tempus and Daiichi Sankyo have entered a collaboration to apply multimodal AI models and real-world clinical data to an oncology antibody-drug conjugate (ADC) program. The effort centers on improving patient selection and identifying biomarkers that can differentiate treatment response, aiming to increase clinical trial success rates and refine trial design.
The collaboration combines Daiichi Sankyo’s clinical and preclinical datasets with Tempus’ large-scale oncology database, which includes linked pathology, molecular, and clinical records. At the core is Tempus’ PRISM2 model, a multimodal system trained to integrate histopathology images with clinical data to generate predictive signals related to disease progression and treatment response.
PRISM2 builds on a model developed and open-sourced in 2024 by Paige AI, the company that Tempus acquired for $81 million in 2025. Earlier this year, Tempus also launched Page Predict, an AI-based digital pathology system that predicts biomarker presence across 16 cancer types and guides molecular testing decisions.
Chicago-based Tempus is steadily expanding its use of AI across precision diagnostics and treatment workflows, with recent developments including the launch of the olivia health data assistant, Loop, a platform integrating real-world data with organoids and CRISPR screening, and xM for TRM liquid biopsy test for immunotherapy monitoring.
See also: Tempus, AstraZeneca, and Pathos Partner to Build Oncology Foundation Model Using Multimodal Data
The company also made several recent strategic acquisitions, including the acquisition of Deep 6 AI for clinical trial matching, and Ambry Genetics for $600 million to extend its capabilities in hereditary and clinical genetic testing.
Recently, Tempus and Merck expanded their multi-year collaboration to apply AI to multimodal clinical and molecular datasets for biomarker discovery, using Tempus’ Lens platform and data infrastructure to support oncology research and early-stage drug development.
Daiichi Sankyo is a Japan-origin global pharmaceutical company, focused on developing innovative and generic medicines across diverse therapeutic areas, with more than 120 years of scientific experience. The company provides access to clinical study datasets and oncology research resources, including T-DXd (a highly effective, next-generation antibody-drug conjugate (ADC) designed to treat HER2-expressing cancers) data, through platforms such as DataSource and Vivli, which host publications, congress materials, and patient-level clinical datasets.
Building Data-Driven Systems for Oncology Trials
The companies plan to develop proof-of-concept AI models that map how different patient subgroups respond to a potential novel ADC. These response maps are intended to support novel biomarker discovery and allow for more granular patient stratification, a key constraint in ADC development where efficacy and toxicity can vary significantly across populations.
The models will be used to simulate and benchmark potential control arms, which could inform trial design and reduce reliance on broad comparator groups.
See also: Companies Applying AI to De-Risk Clinical Trials: 2026 Watchlist
Tempus-Daiichi collaboration targets two recurring challenges in oncology trials.
- First, identifying predictive biomarkers early enough to guide enrollment.
- Second, improving trial efficiency by choosing patients more likely to respond, which can affect both statistical power and cost. ADCs in particular require careful balancing of efficacy and off-target toxicity, making patient selection a central variable.
Cancer AI is increasingly used to link pathology, molecular data, and clinical outcomes to guide biomarker discovery and patient selection. Many groups are building systems that support trial design, subgroup definition, and treatment matching across datasets, as cancer research becomes more data-driven end to end.
Topic: AI in Bio
