Nucs AI and Segmed to advance AI-driven precision oncology with real-world data

Nucs AI and Segmed have announced a partnership to combine large-scale real-world imaging and clinical data with AI to support the development and validation of predictive imaging biomarkers for precision oncology and theranostics.

The Baseline

• Nucs AI and Segmed are partnering to combine large-scale real-world imaging and clinical data with AI to support the development and validation of predictive imaging biomarkers for precision oncology.
• The collaboration will use longitudinal, multi-institutional real-world data (RWD) to improve prediction of treatment response and support the generation of real-world evidence (RWE).
• The partnership also includes the co-development of disease registries, validation frameworks, and research resources to advance precision oncology and theranostics.

AI technology firm Nucs AI has announced a collaboration with Segmed, a clinical data and imaging company, to address a longstanding challenge in precision oncology: developing AI models capable of predicting which patients are most likely to benefit from a particular therapy. As part of the agreement, Segmed will become Nucs AI's preferred oncology data partner, providing access to one of the largest collections of longitudinal imaging and clinical RWD available for AI development. The collaboration will also include a strategic investment by Segmed to support the co-development of disease registries, validation frameworks, and RWE initiatives.

The goal is to develop predictive imaging biomarkers that can help identify which patients are most likely to respond to a given therapy, supporting more personalized treatment decisions and reducing exposure to ineffective treatments. The companies point to radioligand therapy as an example of the difficulty in accurately predicting treatment response, noting that the landmark VISION trial found more than half of patients treated with lutetium-177 prostate-specific membrane antigen (PSMA) therapy did not achieve a prostate-specific antigen (PSA) response. According to the companies, improving these predictions requires AI models trained on imaging, clinical information, and patient outcomes from large, diverse real-world populations. They argue that models developed using narrower or single-site datasets may capture characteristics specific to one clinical setting and perform less consistently across broader healthcare populations.

To support this approach, Nucs AI will gain access to Segmed's network of de-identified imaging and clinical data spanning more than 2,800 healthcare partner sites and approximately 150 million imaging studies across multiple cancer types, imaging modalities, treatment settings, and geographic regions. Unlike many imaging datasets, these longitudinal data follow patients over time, linking diagnosis, treatment, and follow-up over time within a single longitudinal record. This longitudinal view is particularly valuable for theranostic applications, where understanding the relationship between imaging, therapy, and patient outcomes is essential.

In addition to supporting the development of predictive imaging biomarkers, the companies plan to jointly develop disease-specific registries, curated research cohorts, and validation frameworks to support the generation of RWE for healthcare providers, researchers, pharmaceutical companies, and patients. The partnership will also extend Segmed's data infrastructure into molecular imaging, radiopharmaceuticals, and theranostics.

Commenting on the collaboration, Nijat Ahmadov, Chief Executive Officer of Nucs AI, said access to diverse, outcomes-linked data is fundamental to developing predictive AI models for precision medicine:

"Models are not the bottleneck in precision medicine; data is. Specifically, diverse, multi-institutional data that pairs imaging with real clinical context and outcomes. Train on that, and you unlock accuracy and patient-level specificity that imaging alone simply leaves on the table."

Ahmadov added that access to Segmed's data resources would allow Nucs AI to address clinically important research questions rather than being constrained by the availability of suitable datasets.

Nucs AI has already trained its existing models using data from more than 72,000 lesions. Access to Segmed's broader real-world datasets is expected to support expansion across additional cancer types, imaging modalities, and clinical applications.

Commenting on the partnership, Jie Wu, PhD, Co-founder and Chief Data Officer at Segmed, said the company's focus has been on building the data infrastructure needed to support precision oncology:

"We have spent years building the data foundation that precision oncology requires. The next step has always been finding partners who know how to use it. Nucs AI is asking precisely the right clinical questions, ones that demand the kind of diverse, multi-institutional, outcomes-linked data we have built."