 The ongoing COVID-19 pandemic has highlighted the importance of streamlining regulatory approval of medical products and technologies. Computational medicine and emerging field integrating computational imaging and modelling offers a pathway to refine, reduce or replace otherwise costly and lengthy clinical trials, allowing them to be performed computationally or in silico. But reduced costs and time are only two benefits of in silico trials. By performing trials on virtual populations, investigators can thoroughly explore extreme but plausible conditions that would not be feasible or ethical to consider in conventional clinical trials. They can also reduce the risk of human harm and the need for animal experiments. However, because in silico trials are a new approach, the question of whether they can genuinely first replicate and then expand on conventional trial results remains. To address this question, the Royal Academy of Engineering sponsored researchers from the University of Leeds collaborating with an interdisciplinary team from Oxford, Luven and Nijmegen to undertake one of the most comprehensive in silico clinical trials to date, comparing the results against the findings of three conventional clinical trials using the model of flow-diverted treatment of brain aneurysms. Flow-divertors revolutionized the treatment of brain aneurysms when the US FDA approved them in 2011. These self-expanding, stent-like devices redirect flow towards the mainstream, reducing blood flow into the aneurysm and ultimately cutting it off from the circulation or occluding it. Conventional clinical trials have demonstrated the safety and efficacy of these devices for many aneurysm types, but a better understanding of why flow diversion sometimes fails is needed. The team generated an in silico trial cohort with 82 patient-specific 3D anatomic surface models of secular intracranial aneurysms, ensuring that the cohort demographics and aneurysm characteristics matched those in conventional trials. They evaluated whether the aneurysms would undergo complete occlusion, defined in their model as a post-treatment reduction in space and time-average velocity in the aneurysm of at least 35%. The results were then compared with those of three previously published clinical trials. The occlusion rates observed in the in silico trial matched those observed in the previously published clinical trials, demonstrating that in silico trials can successfully replicate the findings of conventional trials. The in silico trial also allowed for detailed subgroup analysis not possible in the conventional trials, showing a higher risk of incomplete occlusion for aneurysms with a branch vessel emerging from the aneurysm sac and patients with hypertension, but no significant increased risk in aneurysms larger than 10 mm or with an aspect ratio greater than 1.6. Finally, by modelling clot formation and stability in a subgroup of aneurysms, the in silico trial explained the elevated risk of post-treatment ischemic and hemorrhagic stroke in certain patients. The findings of the in silico trial successfully replicated and expanded upon the results of conventional clinical trials. In silico trial insights such as these can revolutionise the design of medical devices, making them safer and more effective, while simultaneously reducing cost, human and animal harm and time to market.