New Software Predicts Ovarian Cancer Prognostics 4x Better than Standard Methods
TEXLab, a mathematical AI software created by scientists at Imperial College London and the University of Melbourne, can predict survival rates of patients with ovarian cancer more accurately than any current method, a trial published in Nature Communications has found.
Ovarian cancer is the sixth most common female cancer, and often affects women after menopause, or those with family history of the disease. The survival rate of the cancer is 35-40%, due to most diagnoses occurring at a late stage when symptoms become evident.
TEXLab examined four biological tumour characteristics which heavily influence survival rates: structure, shape, size and genetic makeup. Patients are then given a Radiomic Prognostic Vector (RPV) score, from mild to severe, setting out how severe the disease is.
The technology’s results were compared with blood tests and currently used prognostic scores, with the researchers finding that their software was up to four times more accurate for fatality prediction than other standardised methods.
Five percent of those diagnosed with a high RPV score were found to have a survival rate of less than two years. Alongside this, a high RPV score indicated chemotherapy resistance and poor surgical outcomes.
The new technology could be used to administer the best treatments for every patient individually, and to stratify ovarian cancer patients into groups based on the differences in the texture of their cancer on CT scans, rather than classifying them based on the types of cancer they have.
From here, the researchers said they would carry out a larger study to judge the accuracy of the technology regarding surgical and drug therapy outcomes.
Lead author Professor Eric Aboagye at Imperial College London stated that those with advanced ovarian cancer have poor long-term survival rates, despite recent advances in treatment: “There is an urgent need to find new ways to treat the disease. Our technology is able to give clinicians more detailed and accurate information on the how patients are likely to respond to different treatments, which could enable them to make better and more targeted treatment decisions.”