|
Abstract:
|
The signal detection methodology has been widely used in detecting safety signal, but has not been applied to identify promising drug candidates with early efficacy/response rate information. We propose to apply signal detection methodologies to measure specific drug/disease setting response rates against the background response rates to see if they are truly promising to move forward. We have established a subject-level tumor response library using Bristol Myers Squibb internal clinical database, which could be used as a reference to assess future promising drug candidates. To apply suitable prior information to future studies, tumor response rate can be generated from this data library based on a different combination of factors of interest, such as tumor type, age, gender, and race. Furthermore, MAP prior, RMAP prior, conjugate prior, and effective sample size can be obtained directly from the library through Bayesian framework. A flexible R-shiny application has been developed to help integrate the statistical/visualization tools with data library to enhance the internal decision making of future assets.
|
