Artificial intelligence at the service of medical research in Hepatology. A research project conducted by the Autoimmune Liver Diseases Centre of Milan-Bicocca University at San Gerardo Hospital in Monza, and the Rulex Data Science Team in Genoa, has allowed the identification of four new subtypes of Primary Biliary Cholangitis (PBC) based on the clinical data of more than 12 thousand subjects from all over the world. The new algorithm joins the existing prognostic scores and makes it possible to improve the prognostic assessment of patients at the time of diagnosis.
“This study is very important to us patients, considering the high number of Italian patients included and the potential of innovation based on artificial intelligence,” explained Davide Salvioni, chair of AMAF Onlus, the Italian patient advocacy group dedicated to autoimmune liver diseases. Better knowledge of these diseases will undoubtedly have positive repercussions on doctors’ ability to manage them more effectively.”
PBC is a liver disease that, although rare, in Italy affects more than 10,000 people, especially women over 40 years of age. Over the last decade, there has been a progressive improvement in the prognostic stratification of patients with PBC, due to the development of scores and calculators.
Artificial intelligence and machine learning were recently applied with positive results in the study of several common diseases, from infections to cardiovascular diseases, from breast cancer to bowel cancer. Nevertheless, in the rare diseases setting, and that of PBC in particular, there is a paucity of experimental evidence on these new technologies and their applications.
The team of the Autoimmune Liver Diseases Centre in Monza, led by Professor Pietro Invernizzi, used Rulex, an innovative data analysis tool using a sophisticated artificial intelligence algorithm developed by the Rulex Research & Development team, headed by its Managing Director, Marco Muselli, and based on a theoretical model developed by the Institute of Electronics, Engineering, Information and Telecommunications of the National Research Council in Genoa.
The study, which was published in the journal Liver International, collected the largest cohort of patients with PBC ever studied at international level, including patients from Europe, Japan and North America (DOI: 10.1111/liv.15141). The aim of the study was to exploit this huge dataset to improve the stratification of risk in this rare condition. Four disease subgroups were identified, in order of increasing clinical severity, based on just three laboratory values: albumin, bilirubin and alkaline phosphatase.
“The Rulex team headed by Damiano Verda classified patients with PBC in a completely novel way and devised rules that are very easy to apply in clinical practice to classify new patients as soon as they are diagnosed,” explained Doctor Alessio Gerussi, the first author of the study and researcher at the Autoimmune Liver Diseases Centre in Monza.
“Our work doesn’t end here: future studies will focus on combining the clinical data with data from gene sequencing, radiodiagnostic techniques and digital scans of histology sample slides,” Gerussi pointed out. The ultimate aim is to describe the heterogeneity of the disease in a more refined manner than has been possible to date, in order to provide patients with personalised treatment, which is the ultimate purpose of Precision Medicine.”