We use publicly available data to constrain our models, training to multiple clinical trials, drug targets/compounds, doses, and routes of administration. This process imparts a high level of confidence in our predictions.
We can also incorporate proprietary data from our collaborators to fine-tune model prediction accuracy.
- We work closely with our clients to ensure that models are updated to reflect the current understanding of the biology
- We can run virtual clinical trials across a range of inclusion criteria, doses/schedules, and routes of administration, as well as clinical endpoints, thereby predicting the success of target modulation under various conditions
- If significant clinical efficacy fails in simulation, we provide advice on alternate study design
- We can also filter our large virtual population to select subgroups that represent alternate trial enrollment criteria and phenotype-specific characteristics. These populations can then be used to investigate alternate trial designs
- Informed clinical trial predictions can be achieved using limited data, including: patient-level baseline data alone, placebo trajectory data alone, phase 2 data alone (to predict phase 3 results), or via inclusion of only published in vitro and preclinical data