SAM

SAM implements self-adapting mixture priors to integrate historical data with current clinical trial data, dynamically adjusting information borrowing to mitigate prior-data conflict.

Key Features:

• Self-Adapting Mixture Priors: Mixture priors combine informative and noninformative components with mixing weights determined dynamically using likelihood ratio test statistics or Bayes factors.
• Dynamic Information Borrowing: Mixing weights are adjusted in a data-driven manner to increase or decrease borrowing from historical datasets according to the degree of prior-data conflict.
• Sample-Size Robustness: Framework applies to both finite and large sample sizes, supporting consistent information borrowing across varying trial scales.

Scientific Applications:

• Clinical Trials: Integrating historical data into current efficacy and safety analyses while controlling the influence of prior information when conflict is detected.
• Sequencing Analysis: Applying the SAM prior principles to analyses that utilize alignment information recorded in Sequence Alignment/Map (SAM) format files.

Methodology:

Mixing weights for informative and noninformative prior components are computed using likelihood ratio test statistics or Bayes factors and are adjusted according to observed prior-data conflict to control information borrowing.