Project 466620

Bladder cancer is the 5th most common cancer in Canada with about 75% of these patients being diagnosed with non-muscle invasive bladder cancer (NMIBC), in which the tumour does not infiltrate the muscle layer surrounding the bladder. NMIBC has about a 65% recurrence rate but unfortunately, up to 20% of NMIBC patients will progress to more advanced stages of disease. While NMIBC can be managed with minimally invasive surgeries to remove bladder tumours or instilling medication into the bladder, higher stages such as muscle invasive bladder cancer requires extensive surgery which includes the removal of the bladder, nearby organs, and the need to create a urinary diversion to allow urine to drain from the kidneys to outside the body.As progression often occurs after multiple NMIBC recurrences, identifying patients at risk of progression is important for both patient counselling and clinical decision-making to guide frequency of follow-ups and the need for more aggressive upfront treatment. Unfortunately, current predictive models are limited due to poor performance and several practical issues. First, these models do not reflect the current standard of care for management of NMIBC, which includes re-resection of the tumour bed when indicated and use of intravesical therapies. Second, these models are time-fixed in that they do not capture changes in the patient's cancer status after initial diagnosis, which includes subsequent follow-ups and tumour recurrences. To address these shortcomings, we plan to develop a time-series forecasting model using artificial intelligence to better encapsulate the complete oncological timeline of contemporarily treated NMIBC patients to improve prognostic accuracy.