Speaker: Khaled El Emam GM and SVP, Replica Analytics; Professor, University of Ottawa
Abstract : With the growing interest in generative AI across the health industry, one type of generative AI is synthetic data generation (SDG). This important set of technologies can help solve challenging privacy problems by enabling the access and sharing of health data in a responsible way. In addition to that SDG can also allow us to tackle data bias, data augmentation, and clinical study accrual problems. Solving these problems can potentially bend the trajectory on equity, clinical research, and decision making in the discovery of new drugs, the development of devices, and the delivery of care. This presentation will provide an overview of the technology, followed by sharing ten important lessons learned distilled from experiences over the last four years about what works and what does not work in practice. We have been developing and applying SDG technologies across a heterogeneous set of data, including real-world data and clinical trial data, and in multiple jurisdictions. The lessons can make the difference between successful efforts to deploy SDG and get business value from the technology and the methodologies around it use (which are just as important).