2023-12-01 16:12:00
Precisely evaluating the type of a tumor, starting from its being benign or malignant up to its lesser or greater aggressiveness, is a fundamental step in guiding the urgency and type of therapeutic choices to be implemented. In the case of brain tumors, much information may not be available before surgery, but new applications of artificial intelligence can be deployed to get a clearer picture of the situation in advance. This is the objective of research published in the scientific journal ‘Frontiers in Oncology’ and conducted by the Neurosurgery Unit and the Department of Epidemiology and Prevention of the Irccs Neuromed of Pozzilli (Isernia), in collaboration with the Clinica Mediterranea of Naples , the University of Insubria of Varese-Como and the Umberto Veronesi Ets Foundation.
The researchers – we read in a note – studying 90 patients with suspected tumors of the central nervous system, explored the possibility of early identifying the malignancy of the tumor itself using exclusively the information already available. To achieve this result, a machine learning technique (a branch of AI) called clustering was used. “We have collected – explains Alessandro Gialluisi, researcher of the Department of Epidemiology and Prevention of Neuromed and of the Free Mediterranean University (Lum) ‘Giuseppe Degennaro’ of Casamassima (Bari) – a series of data that were easily available before the surgery thanks to a simple medical history. For example: the patient’s family and clinical history, the environment in which he lives, his lifestyle, level of education, psychological situation and socioeconomic status. Subsequently, all this information is processed by an algorithm that created clusters (groups of patients with similar characteristics, ed.), which were finally compared with the real results of the investigations carried out following the intervention”.
AI has made it possible to identify which elements of everyday life can be correlated with the presence of a malignant rather than benign tumor, and possibly with its aggressiveness.
“Naturally, this is a preliminary study – comments Vincenzo Esposito, head of the Neurosurgery Unit II and director of the Neuromed Neurosurgery Department – which will need confirmation and further research on a greater number of patients. Our results however indicate that machine learning algorithms, based on a combination of clinical and biological characteristics, could be useful in helping doctors decide on individual patient priorities for surgery or other treatments.”
“Thanks to this personalized medicine approach – adds the specialist – we could make up for the lack of information, both histopathological and molecular, which is not available until the patient undergoes surgery”.
#Tumors #improve #diagnosis #treatment #brain #tumors #Italian #study