Can synthetic data be a proxy for real clinical trial data? A validation study