Healthcare needs evidence that survives peer review.
When a journal or an investor asks how sure you are, your evidence holds up. Study design, patient outcomes and testing of clinical AI.
The questions we answer.
How many patients does this study need?
You know the right number before the first patient enrolls, and you can show why.
Power analysis, simulation
How should the trial be designed?
You agree the rules before any data arrives, so nobody can argue the result away.
Experimental design, statistical analysis plans
How long until the event, and what changes it?
You get time to outcome measured correctly, even when patients drop out.
Survival analysis, Cox models, competing risks
Does the AI do better than the clinicians?
You get an honest comparison with your experts, on patients the model has never seen.
Inter-rater agreement, calibration, external validation
Can existing patient data answer this?
When a trial is not possible, you still get a careful estimate, with its limits stated.
Causal inference, propensity methods, sensitivity analysis
- First author
- paper on embryo assessment, Oxford University Press on behalf of ESHRE
- Co-author
- preliminary model of mutational burden in retinal dystrophy, 527 subjects, Journal of Ophthalmic and Vision Research
- IVF
- statistical simulations of AI tools across a patient's full treatment cycle
- Trials
- experimental design of clinical trials for an AI-driven fertility company
Our work here draws on clinical trial design and statistical simulation in IVF, first-author research published by Oxford University Press, and a preliminary model of mutational burden in inherited retinal disease.
Bring us the study before it starts.
The cheapest time to fix a study is before the first patient. Tell us the question and we will tell you what it takes to answer it.