10 November 2026
Inspired by the work of Ewan Birney, today’s recipient of the 2026 “Lombardia è Ricerca” Award, Delphi-Lombardia brings together Human Technopole and Centro Cardiologico Monzino to develop an AI model that can map how people’s health changes over time and help identify the risk of cardiac events, including heart attacks. Human Technopole will help develop the model using health data from Lombardy, while Monzino will support its clinical validation through EPIFANIA, a cohort of around 1,300 patients with advanced imaging and multi-omics data.
Delphi-Lombardia aims to use artificial intelligence to analyse the health histories of millions of people and identify those who may have substantial plaque build-up in their coronary arteries at an early stage. Known as coronary atherosclerosis, this chronic inflammatory condition can lead to serious events such as heart attacks. Developed by Human Technopole and Centro Cardiologico Monzino, the project will bring to Lombardy one of the most advanced approaches to using generative AI to predict how people’s health may change over time and explore its potential for cardiovascular prevention.
Delphi-Lombardia will be funded by a share of the 2026 Lombardia è Ricerca Award, presented today to Ewan Birney, Director of the European Bioinformatics Institute (EMBL-EBI). Birney is a leading author of the study introducing DELPHI, an AI model developed by an international team to analyse health trajectories at scale. His work provides the foundation for Delphi-Lombardia. The €1 million award recognises the British scientist’s contribution to computational biology, genomics and AI-based predictive medicine. The portion earmarked for research in Lombardy will support Human Technopole and Monzino in carrying out the project.
The two institutes will bring complementary expertise to the project. At Human Technopole, the research groups led by Andrea Ganna and Francesca Ieva will develop Delphi-Lombardia, adapting the model to help create a risk score for coronary plaque burden. At Centro Cardiologico Monzino, Gualtiero Colombo’s group will contribute EPIFANIA, a cohort of around 1,300 patients with advanced coronary imaging and multi-omics data. This will allow researchers to compare the model’s estimates with measurements from coronary CT angiography.
Andrea Ganna, Senior Research Group Leader at Human Technopole, states: “The Lombardy Region possesses an extraordinary wealth of health data. Through this project, we aim to leverage this asset using secure artificial intelligence methods to better understand the cardiovascular health trajectories of citizens, thereby contributing to the development of increasingly effective and personalized prevention strategies. The core concept of the project is to model a patient’s entire clinical history—encompassing both hospital records and data from general practitioners. Based on this information—much like how generative AI models predict the next word in a sentence—we seek to determine the probability of future clinical events. Lombardy will be the first region in Italy, and one of the first in Europe, to explore the potential of these innovative approaches.”
Gualtiero Colombo echoes this sentiment: “Delphi-Lombardia was established with an ambitious goal: to harness artificial intelligence and the vast availability of regional health data to better understand the progression of diseases—specifically, to improve the early identification of cardiovascular risk. The Centro Cardiologico Monzino will contribute its specialized expertise and a rich dataset—comprising clinical records, coronary imaging, and multi-omic data from one of its prospective cohorts—to the predictive models developed by Human Technopole. The objective is to determine whether these models can be transformed into practical, effective tools for patient prevention and care. In this sense, the ‘Lombardia è Ricerca’ (Lombardy is Research) Award represents much more than just an accolade; it fosters an environment where world-class scientific expertise meets local excellence, strengthening collaborations that generate new knowledge and accelerate its translation into clinical practice.”
Human Technopole: using AI to map health trajectories
Delphi-Lombardia builds on DELPHI, a generative AI model designed to map and predict disease progression and estimate the risk of developing more than 1,000 conditions. The model was trained on data from around 400,000 UK Biobank participants and subsequently validated in 1.9 million people in Denmark. It predicts how different diseases may develop in sequence or coexist over a person’s lifetime, generating possible future health trajectories.
The project will adapt this approach to the Italian healthcare setting, focusing on cardiovascular prevention. Researchers will use routinely collected health information, including previous diagnoses, medication records and demographic data, to develop a score that could identify people with substantial coronary plaque build-up, even before a diagnosis has been recorded.
The aim is to give clinicians additional information to assess risk alongside existing diagnostic tests. In the longer term, the tool could support more targeted prevention and help identify who would benefit most from further investigations, such as coronary CT angiography.
EPIFANIA: clinical validation at Centro Cardiologico Monzino
Centro Cardiologico Monzino will test the model using EPIFANIA, a cohort coordinated by Gualtiero Colombo. It includes around 1,300 patients who underwent coronary CT angiography for suspected coronary artery disease. The scan provides a direct view of atherosclerosis in the coronary arteries, showing the amount and composition of plaque and the extent of any narrowing of the blood vessels.
A key strength of EPIFANIA is the depth of information available on each patient. Clinical records and advanced coronary imaging are complemented by analyses of DNA, gene activity, proteins and metabolites, providing a detailed picture of patients’ biological characteristics. Around 280 participants also had a second CT scan three to five years later, allowing researchers to track the progression of atherosclerosis.
Researchers will use these data to compare the AI model’s estimates with the findings from coronary imaging. They will assess whether the Delphi-Lombardia score is associated with the amount of plaque measured on CT scans and, for patients with a second scan, with disease progression over time.