The use of Artificial Intelligence in Multimorbidity Treatment and Research
A project that – thanks to the use of Artificial Intelligence – aims to improve treatment strategies for patients suffering from multimorbidity and polypathologies. On behalf of the Colleagues of the Units involved in the Immuno Center of Humanitas, Carlo Selmi – Head of Rheumatology and Clinical Immunology – introduces the project to us.
About 5% of the world’s population suffers from immune-mediated inflammatory diseases (IMID). Thats about more than a hundred diseases characterized by a chronic inflammatory process – in many cases a disabling one – which often have common symptoms, same therapies and may be present at the same time. This coexistence of pathologies is defined as multimorbidity. It occurs when, for example, in the same patient we find an inflammatory arthritis that coexists with intestinal inflammation or psoriasis.
Therefore, it is important that the clinician understands how a multimorbidity state can interfere with the patient’s clinical picture, for example by creating problems in the treatment of the most serious pathologies that afflict him/her. This aspect is crucial to improve our comprehension of the multimorbidity itself and to outline better treatment strategies that will benefit the patient’s health.
Research and Artificial Intelligence: a new frontier in the treatment of multimorbidities
The specialists and researchers of the Immuno Center of Humanitas are working together with the data scientists of the Artificial Intelligence Center with the aim of evaluating new ways to treat multimorbidities. The first step is to combine the unique clinical resources of the facility and the Research Biobank at its disposal with the use of innovative tools that – thanks to Artificial Intelligence – aim to identify the physiological markers of these pathologies.
The aim of this research is to use new learning models based on the data available – about thirty thousand patients until 2018 – in order to identify the most common patterns of multimorbidity and to assess the correlation between the data of patients affected by one or more IMIDs and the treatment possibilities.
The set of patterns will be tested through a cross-validation mechanism. The deep learning networks will also be testeand validated in order to improve the adequacy of the data.
How many researchers are involved and how long will the project last?
The project involves an interdisciplinary team and clinical and research staff working in the four operational units – Rheumatology, Allergology and Personalized Medicine with Walter Canonica and Francesca Puggioni, Dermatology with Antonio Costanzo, Intestinal Inflammatory Diseases with Silvio Danese, the scientific coordinator of the ImmunoCenter. These different areas will analyse to a large number of patients that come to the Center for different clinical reasons. Therefore, the research goes beyond the limits of a single specialist area. The first extraction of keywords from the clinical database is currently in progress in order to identify a first number of patients that will educate the deep learning system. The next step that will develop over a few months will be the creation of algorithms predicting the course of chronic inflammatory diseases.
How are the medical skills and the expertise of data scientists integrated?
A close collaboration between the medical staff of the ImmunoCenter units – the data producers – and those who will analyze the data after making it available is obviously essential to obtain reliable data that will then lead to an algorithm that can be verified and used in the future clinical activity.