Artificial Intelligence Alongside Physicians in Canada: Reality and Risks

Abstract

Artificial intelligence (AI) and machine learning (ML) have the potential to revolutionize the healthcare system with their immense potential to diagnose, personalize treatments, and reduce physician burnout. These technologies are highly dependent on large datasets to learn from and require data sharing across organizations for reliable and efficient predictive analysis. However, adoption of AI/ML technologies will require policy imperatives to address the challenges of data privacy, accountability, and bias. To form a regulatory framework, we propose that algorithms should be interpretable and that companies that utilize a black box model for their algorithms be held accountable for the output of their ML systems. To aid in increasing accountability and reducing bias, physicians can be educated about the inherent bias that can be generated from the ML system. We further discuss the potential benefits and disadvantages of existing privacy standards ((Personal Information Protection and Electronic Documents Act) PIPEDA and (Personal Information Protection and Electronic Documents Act) GDPR) at the federal, provincial and territorial levels. We emphasize responsible implementation of AI by ethics, skill-building, and minimizing data privacy breaches while boosting innovation and increased accessibility and interoperability across provinces.