Power Optimization Patterns

Power Optimization Patterns

When to Use This Skill

• Optimizing sample size for target power
• Selecting design parameters (randomization ratio, event count)
• Trading off between competing objectives
• Performing sensitivity analysis
• Finding optimal regions across scenarios

Clinical Trial Optimization Framework

Problem Formulation

Components:

• Data Model D(θ): Parameterized by θ (treatment effects, rates, etc.)
• Analysis Model A(λ): Parameterized by λ (sample size, events, etc.)
• Criterion ψ(λ | θ): Power or other metric

Objective: Find λ* that optimizes ψ(λ | θ) subject to constraints.

Direct Optimization

Sample Size Determination

Objective: Find minimum n such that Power(n) ≥ target

Binary Search Algorithm:

find_sample_size  5) {
    n_mid = target_power) {
      n_high = 0.90))]

Tradeoff-Based Optimization

Additive Criterion

Formula:

ψ_combined(λ) = w₁ × ψ₁(λ) + w₂ × ψ₂(λ)

Example: Power vs Sample Size

# Weights: 70% power importance, 30% sample size (negative for minimization)
w1 = 0.90)
optimal_w = 80%):", mean(bootstrap_power >= 0.80), "\n")

Optimal Intervals and Regions

Optimal Interval

The η-optimal interval contains all λ values within η% of optimal power.

# Define optimal interval
eta = threshold]
cat("95%-optimal interval: [", min(optimal_interval), ",", max(optimal_interval), "]")

Joint Optimal Region

Intersection of optimal intervals across scenarios.

# Find joint optimal region
intervals  subset(w1 + w2 == 1)

# Evaluate each weight combination
# ... and select optimal weights

Best Practices

1. Define Clear Objective: Power, sample size, cost, or combination
2. Consider All Scenarios: Use qualitative sensitivity for range
3. Validate Constraints: Check FWER, minimum power thresholds
4. Document Trade-offs: Explain rationale for chosen parameters
5. Report Robustness: Show performance across scenarios
6. Pre-specify Optimization: Define optimization strategy in SAP

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: choxos
• Source: choxos/BiostatAgent
• License: MIT

Install and usage instructions live in the source repository linked above.