Blab AI
Spain Launches AI Initiative to Predict Survival in Severe Trauma Cases
On 8 July 2026, the Arnau de Vilanova University Hospital, the Lleida Biomedical Research Institute, and the University of Lleida announced a new artificial‑intelligence project to predict survival in patients with severe trauma. The initiative received a 90,000‑euro grant from the Fundación Mútua Madrileña and will use data from more than 22,000 trauma cases recorded since 2015 to train AI models that could improve early mortality predictions in the first hours after admission.
The project builds on the RETRAUCI registry, a national database that collects clinical information on critical trauma patients across Spain. The team will also refine RETRASCORE, a tool originally developed in Lleida to estimate in‑hospital mortality within the first 24 hours of admission. RETRASCORE was built with classic statistical models, which, according to Luis Servià, have limitations in capturing the complex interactions that influence survival after an accident.
Luis Servià, professor at the University of Lleida, physician at Arnau de Vilanova, and researcher at the Lleida Biomedical Research Institute, said the new approach will use AI‑based prediction tools to better understand the factors that determine a patient’s survival. The updated model will incorporate additional clinical variables, including laboratory tests that were not part of the initial design. This expansion is supported by the accumulated experience of the Lleida teams and the collaboration of other centers that also participate in the project.
In addition to the Lleida‑based teams, the project will involve five hospitals outside Catalonia: the 12 de Octubre and San Carlos Clinical hospitals in Madrid, Marqués de Valdecilla in Santander, Virgen del Rocío in Seville, and the hospital in Burgos. The network will allow the registry data to be contrasted with patients treated in various hospitals across the country, providing a broader perspective on trauma outcomes.
The 90,000‑euro grant is part of the Fundación Mútua Madrileña’s annual call for applications, which this year distributed 2.3 million euros among 21 scientific projects in 12 autonomous communities. Catalonia received three of these initiatives and a total of 300,000 euros in aid. The award ceremony was held on Wednesday in Madrid and was attended by Rafael Matesanz, Ignacio Garralda, and Julio Zarco.
The initiative aligns with other ongoing projects in the Arnau de Vilanova environment, such as the MIR position for Emergencies, and reflects a growing interest in applying machine‑learning techniques to clinical decision support. By leveraging a large, longitudinal dataset, the team aims to produce a tool that can be deployed in emergency departments to inform triage and treatment decisions.
At present, the project is in the data‑collection and model‑development phase. The research team plans to validate the AI model externally once the training dataset is complete. If successful, the tool could be integrated into the electronic health record systems of participating hospitals, potentially improving early survival rates for severe trauma patients across Spain.
The initiative is positioned to address a critical gap in trauma care: the need for rapid, accurate prognostic information in the first hours after injury. By integrating additional laboratory variables and leveraging machine‑learning algorithms, the research team hopes to produce a tool that can flag patients at high risk of early mortality. Such a tool could inform triage protocols, prioritize intensive‑care resources, and guide early therapeutic interventions. The project will also adhere to Spanish data‑protection regulations, including the General Data Protection Regulation and national privacy laws, to safeguard patient confidentiality. While the model’s performance will be evaluated through external validation, the study’s design allows for future adaptation to other clinical settings, such as predicting outcomes in different critical‑care populations.
The project builds on the RETRAUCI registry, a national database that collects clinical information on critical trauma patients across Spain. The team will also refine RETRASCORE, a tool originally developed in Lleida to estimate in‑hospital mortality within the first 24 hours of admission. RETRASCORE was built with classic statistical models, which, according to Luis Servià, have limitations in capturing the complex interactions that influence survival after an accident.
Luis Servià, professor at the University of Lleida, physician at Arnau de Vilanova, and researcher at the Lleida Biomedical Research Institute, said the new approach will use AI‑based prediction tools to better understand the factors that determine a patient’s survival. The updated model will incorporate additional clinical variables, including laboratory tests that were not part of the initial design. This expansion is supported by the accumulated experience of the Lleida teams and the collaboration of other centers that also participate in the project.
In addition to the Lleida‑based teams, the project will involve five hospitals outside Catalonia: the 12 de Octubre and San Carlos Clinical hospitals in Madrid, Marqués de Valdecilla in Santander, Virgen del Rocío in Seville, and the hospital in Burgos. The network will allow the registry data to be contrasted with patients treated in various hospitals across the country, providing a broader perspective on trauma outcomes.
The 90,000‑euro grant is part of the Fundación Mútua Madrileña’s annual call for applications, which this year distributed 2.3 million euros among 21 scientific projects in 12 autonomous communities. Catalonia received three of these initiatives and a total of 300,000 euros in aid. The award ceremony was held on Wednesday in Madrid and was attended by Rafael Matesanz, Ignacio Garralda, and Julio Zarco.
The initiative aligns with other ongoing projects in the Arnau de Vilanova environment, such as the MIR position for Emergencies, and reflects a growing interest in applying machine‑learning techniques to clinical decision support. By leveraging a large, longitudinal dataset, the team aims to produce a tool that can be deployed in emergency departments to inform triage and treatment decisions.
At present, the project is in the data‑collection and model‑development phase. The research team plans to validate the AI model externally once the training dataset is complete. If successful, the tool could be integrated into the electronic health record systems of participating hospitals, potentially improving early survival rates for severe trauma patients across Spain.
The initiative is positioned to address a critical gap in trauma care: the need for rapid, accurate prognostic information in the first hours after injury. By integrating additional laboratory variables and leveraging machine‑learning algorithms, the research team hopes to produce a tool that can flag patients at high risk of early mortality. Such a tool could inform triage protocols, prioritize intensive‑care resources, and guide early therapeutic interventions. The project will also adhere to Spanish data‑protection regulations, including the General Data Protection Regulation and national privacy laws, to safeguard patient confidentiality. While the model’s performance will be evaluated through external validation, the study’s design allows for future adaptation to other clinical settings, such as predicting outcomes in different critical‑care populations.
