Translated automatically by AI from the Italian original.
SYRACUSE – In chronic and rare diseases, time is a decisive factor: early diagnosis allows prompt intervention, identification of the best therapeutic option and a reduction in the risk of complications. With this goal in mind, AI-PBC-Finder was created, the cutting-edge scientific project designed by the hepatology operational unit of the provincial health authority (ASP) of Syracuse.
The initiative harnesses the potential of artificial intelligence for the early identification of primary biliary cholangitis (PBC), a rare liver disease that is often complex to diagnose in its early stages.
The project, coordinated by Dr Marco Distefano, has been selected and funded under the 15th edition of the Fellowship Program 2026, a national competition promoted by the biopharmaceutical company Gilead Sciences to support excellence in Italian scientific and socio-health research.
What is primary biliary cholangitis and the risks of delayed diagnosis
Primary biliary cholangitis is a rare, chronic autoimmune liver disease that mainly affects women aged between 40 and 65. In Italy, there are almost 28 cases per 100,000 inhabitants, with more than five new diagnoses per year per 100,000 people.
The main problem lies in the initial lack of symptoms or the presence of vague symptoms that are not immediately linked to the liver. This leads to late diagnoses, when the disease has already caused irreversible damage such as:
- liver fibrosis
- cirrhosis
- the need for an organ transplant
“Today the main problem is managing to recognise the disease early in order to make the best use of therapeutic innovation and avoid progression to the most serious forms,” explained Dr Marco Distefano. “Often the diagnosis comes after months or years. With our project we want to move to a proactive approach thanks to the opportunities offered by artificial intelligence.”
How AI-PBC-Finder works: three operational phases
The system will integrate AI both within hospital structures and in the network of GPs across the area, through a development articulated in three phases:
- Data analysis and algorithm: in the preparatory phase, available clinical and blood chemistry data will be examined to create an algorithm capable of detecting early warning signs of the disease.
- Predictive models: based on the information collected, the AI will develop models capable of estimating the probability of PBC among hospitalised patients and the general population.
- Automated clinical integration: the model will enter daily practice by sending automatic alerts to doctors in the presence of suspected cases, immediately initiating dedicated specialist investigations.
A virtuous and replicable model for healthcare
As well as speeding up the start of treatments for patients in Syracuse and optimising the prescribing appropriateness of doctors, the model developed in Syracuse aims to become a national prototype that can be replicated for other underdiagnosed chronic diseases. The scientific results achieved will be shared with the medical community through scientific publications and sector conferences.
