OMNIMED: Virtual clinical copilot for assisted, safe and explainable diagnosis
INNOGLOBAL Programme · International technological cooperation R&D projects · CDTI · FEDER Funds 2021-2027 · File No. INNO-20251072
Research in generative AI to develop a clinical copilot that supports healthcare professionals in diagnosis and decision-making, reducing bias and hallucinations in the context of cardiac conditions.
Project funded by CDTI with FEDER Funds 2021-2027 through the aid for international technological cooperation R&D projects (INNOGLOBAL). “Europa se siente”.
Research in Generative Artificial Intelligence to develop a clinical copilot that supports healthcare professionals in diagnostic and decision-making processes. To this end, research, the design and the validation of an architecture able to adapt to new contexts through novel techniques that reduce bias and hallucinations in the context of cardiac conditions will be required.
The methodological approach is based on the research and comparison of different generative artificial intelligence architectures and techniques, with a special focus on Domain Adaptation of Large Multimodal Models (LMM) applied to the clinical domain, such as fine-tuning, advanced contexts and prompt engineering. The objective is to identify the most suitable strategy to integrate, in a balanced way: (i) robust mechanisms for the retrieval and use of medical evidence, (ii) models adapted to the linguistic and clinical context of real healthcare practice, and (iii) the generative ability to offer explainable and traceable recommendations, always grounded in verifiable evidence. The research will make it possible to adapt large generative models to highly specific tasks, such as clinical diagnosis targeted at rural populations in Colombia, thereby reducing the likelihood of hallucinations and problems arising from bias.
Avenida Cataluña, 11 Entlo. (Valencia)
Execution period01/01/2026 to 31/12/2027
€344,397.00
€605,093.00
