Options and Considerations for Adaptive Laboratory Experiments.

Area of Science:

• Biostatistics
• Experimental Design
• Oncology Research

Background:

• Laboratory experiments often fail to achieve statistical significance on the first attempt.
• Unanticipated patient variance and unpredictable effect sizes complicate experimental design, especially in leukemia research.
• Achieving significance in the initial experimental run simplifies decision-making processes.

Purpose of the Study:

• To develop a small sample size approach for designing subsequent experiments.
• To control overall type I error and achieve sufficient conditional power when initial experiments lack significance.
• To address challenges in leukemia cell experiments, including Chronic Lymphocytic Leukemia.

Main Methods:

• Developed a small sample size design for sequential experiments.
• Focused on controlling overall type I error and maximizing conditional power.
• Utilized combination testing within an adaptive laboratory experiment framework.

Main Results:

• Demonstrated that various futility cut-offs, effect size estimations, and significance levels yield similar power and sample size.
• Showcased the effectiveness of the approach using a Chronic Lymphocytic Leukemia example.
• Highlighted the benefits over naive procedures combining data from unplanned subsequent experiments.

Conclusions:

• The proposed small sample size approach effectively manages statistical power and error rates in sequential experiments.
• This method simplifies adaptive trial design by emphasizing significance in the first experimental run.
• The strategy is particularly relevant for complex biological systems like leukemia cells where variance is high.