Grant: CIHR Fellowship

Lung cancer is the leading cause of cancer-related death in Canada, resulting in more cancer-related deaths than breast, colorectal, and prostate cancer combined. Lung cancer screening with low-dose computed tomography (CT) has been demonstrated in multiple large scale clinical trials to reduce mortality. However, due to the non-specific appearance of lung nodules on CT imaging, current lung cancer screening approaches have a high proportion of false positives resulting in many unnecessary investigations, increased patient anxiety, and significant healthcare costs. Artificial intelligence (AI) has been shown to have high accuracy for classifying lung nodules as malignant or benign on lung cancer screening CT exams, with strong potential to decrease the number of false positives. However, it is unclear how output from AI algorithms can be used to direct clinical decision making. Additionally, health outcomes and downstream costs relating to the use of AI have not been assessed, limiting clinical translation of this technology. This research will address these knowledge gaps by using a well-validated simulation model to assess the effectiveness and cost-effectiveness of lung nodule management strategies using AI to guide lung nodule management in lung cancer screening programs. This research will provide key evidence for health system leaders and policy makers to potentially support the implementation of AI in lung cancer screening, leading to earlier diagnoses of lung cancer, reduced lung cancer mortality, and reduced healthcare costs. Reducing costs associated with lung cancer screening with the use of AI may enable additional lung cancer screening programs to be developed in Canada, increasing access to screening for Canadians. The research methodology developed through this work, uniquely at the intersection of AI and health services research, will provide a framework for investigating effectiveness and cost-effectiveness of AI in other important clinical domains.