AI Predicts Success of Immunotherapy in Lung Cancer治疗成功

2019-12-02

A study published in Cancer Immunology Research has revealed a research breakthrough in utilizing artificial intelligence algorithms to identify previously unseen patterns in CT scans and determine the benefits of immunotherapy for lung cancer patients. Previously, researchers have pointed out that only about 20% of cancer patients can benefit from immunotherapy. Unlike chemotherapy, immunotherapy uses drugs to help the immune system fight against cancer instead of directly killing cancer cells. Therefore, having an artificial intelligence tool that can match patients with the appropriate immunotherapy method undoubtedly has the potential to significantly reduce medical costs. "Although immunotherapy has revolutionized the entire cancer ecosystem, it is not without flaws. Its most fatal flaw is its high cost, with each patient spending approximately $200,000 per year," said the medical director of a diagnostic center. "Preliminary statistics show that 42% of newly diagnosed cancer patients will deplete their lifelong savings within one year of diagnosis." In the initial study, the research team used CT scans from 50 patients to train the AI algorithm to identify changes in lesions and attempt to find the most relevant patterns associated with successful immunotherapy. Furthermore, the patterns observed in CT scans are closely related to treatment response and overall survival of patients, as well as the immune cell arrangement in their original diagnostic biopsies. This suggests that CT scans actually capture the immune response of tumors to cancer invasion, and tumors with the strongest immune response have the best response to immunotherapy. The algorithm not only determines the size of the lesions but also records changes in the texture, shape, and volume of specific lesions. The researchers stated, "This is important because when doctors judge whether a patient is responding to treatment based solely on CT images, they usually rely on the size of the lesions. However, in reality, changes in texture can better predict treatment efficacy." For example, if blood vessels rupture within a tumor, nodules may appear larger after treatment, but the treatment is actually effective. With this artificial intelligence, it is possible to accurately determine this point, rather than stopping treatment due to surface deterioration of the condition, which could further endanger the patient's life. The research team is extremely proud of this achievement. "Life is priceless. As the developers of this research, we are deeply honored."

RECENT AI NEWS

RECENT AI TOOLS