Scientists from Hong Kong, Beijing, South Korea create AI tool to predict brain cancer prognosis

A team of researchers from Hong Kong, Beijing, and South Korea have developed an artificial intelligence (AI) tool and a web portal that can predict the prognosis of brain cancer patients. The tool also provides publicly accessible information on the severity of individual cases. The team, led by the Hong Kong University of Science and Technology, announced on Tuesday that their predictions could help in devising more precise treatments for patients, potentially improving their survival rates. Currently, the available treatments can typically only extend a patient's life by about three months, and recurrence of the cancer is almost inevitable. When the cancer does recur, it often presents a more aggressive and complex situation. The cancer cells in the recurrence are often different from those in the initial occurrence, which poses a significant challenge to clinical treatment. Glioma, a type of tumour that originates in the brain or spinal cord, is the most common primary brain tumour in adults. The research, conducted by HKUST, Beijing Tiantan Hospital, Prince of Wales Hospital at Chinese University, and the Samsung Medical Centre in South Korea, identified early predictors of glioma progression. These include the role of MYC gene amplification and CDKN2A gene deletion in accelerating tumour growth and increasing the risk of recurrence. The study, which involved 544 patients, also revealed disparities in brain tumour evolution among patients of different ethnicities. It was observed that MYC gene amplification was more common among East Asian patients, while Caucasians showed a greater prevalence of a high-glioma risk factor. The researchers believe that these findings underscore the importance of tailored therapies targeting cancers. To predict the development of brain tumours after the initial diagnosis, the research team developed a machine learning model named CELLO2. This AI-powered technology can accurately predict tumour recurrence and identify patients facing higher risks. The researchers also established an interactive and publicly accessible web portal to provide patients and doctors with a platform for uploading personal data and obtaining predictions on the severity of individual cases of brain tumours. The AI model is expected to take at least five years to be clinically validated, as it usually takes three to five years to evaluate the recurrence of gliomas. The median survival for the disease is 15 to 18 months, while some patients can live for five more years. This suggests that the effectiveness of the current treatment varies from person to person, which highlights the significance of precision oncology.