Your Guide to the Platform

Step-by-step walkthroughs to help you design and analyze clinical trials, no statistics background required.

Choose your trial design

Follow the decision tree, or answer three questions to get a recommendation.

Prefer a guided route?

Once you know your design, follow a walkthrough to size the trial or analyze your data.

“How do I calculate sample size?”

“How do I analyze my trial data?”

Compare all five designs at a glance, with the pitfalls to watch for.

Design	Use when	Watch out for	Guides
Parallel	Independent treatment vs control groups in one population.	With small N and binary or GLMM outcomes, analytic power runs optimistic. Prefer simulation.	Power Analysis
Crossover	Each participant can take both treatments; stable condition, washout feasible.	A progressive disease or a long-half-life repurposed drug makes carryover invalidate it.	Power Analysis
N-of-1	A single participant, rare-disease or personalized decision, over repeated cycles.	The effect must be reversible and show within one cycle.	Power Analysis
Umbrella	One disease split by biomarkers, each subgroup gets a matched treatment.	Shared controls save participants but need comparable subgroups. Plan multiplicity across arms.	Power Analysis
Basket	One repurposed drug tested across several subtypes or biomarker groups.	Borrowing misleads if baskets aren't biologically related. Favor EXNEX over BHM for heterogeneous effects.	Power Analysis

Parallel

Use when: Independent treatment vs control groups in one population.
Watch out for: With small N and binary or GLMM outcomes, analytic power runs optimistic. Prefer simulation.

Crossover

Use when: Each participant can take both treatments; stable condition, washout feasible.
Watch out for: A progressive disease or a long-half-life repurposed drug makes carryover invalidate it.

N-of-1

Use when: A single participant, rare-disease or personalized decision, over repeated cycles.
Watch out for: The effect must be reversible and show within one cycle.

Umbrella

Use when: One disease split by biomarkers, each subgroup gets a matched treatment.
Watch out for: Shared controls save participants but need comparable subgroups. Plan multiplicity across arms.

Basket

Use when: One repurposed drug tested across several subtypes or biomarker groups.
Watch out for: Borrowing misleads if baskets aren't biologically related. Favor EXNEX over BHM for heterogeneous effects.

Before you choose

Biomarker groups are not automatically exchangeable. Basket borrowing needs a biological reason.
Small N or low event rates favor simulation over analytic formulas.
Crossover and N-of-1 need a reversible effect; they fit poorly for progressive disease.
Umbrella shared controls trade participants for interpretation risk.
Subgroup and basket comparisons multiply. Label exploratory findings as exploratory.