A New Graphic Method for Data from Human Drug Abuse Potential Studies

• Activity Number: 377
• Type: Contributed
• Date/Time: Tuesday, August 6, 2013 : 10:30 AM to 12:20 PM
• Sponsor: Section on Statistical Learning and Data Mining
• Author(s): Ling Chen*+ and Yang Wang
• Companies: U.S. Food and Drug Administration and U.S. Food and Drug Administration
• Keywords: Abuse potential ; Crossover study ; Heat maps ; Missing data

Abstract

Aims: This presentation proposes a new way of statistical visualization for displaying the data from clinical abuse potential studies. This type of study has multiple abuse potential measures and multiple treatments. The study is crossover and often uses a Williams square design. Data are collected for treatments from each treatment sequence and during each treatment period at many time points. This presentation shares the methodology of the use of heat maps to display complicated data sets.

Method: To examine individual data, we use heat maps to display data (3 dimensional or 4 dimensional) from clinical abuse potential studies. The Williams square is painted using the colors from transparent to very intense based on Emax, for example from individual subjects. The greater the Emax value, the more intense the color. For each measure, for example we arrange subjects by sequence or by treatment, and use color intensity to show the large or small Emax from each subject to each treatment. For each measure, and each treatment, we also use a heat map to display responses from each subject at all time points (constructing time course response profiles for individual subjects).

Results: This research has been published in the Drug Information Journal. A Graphical User Interface (GUI) in R for heat map displays and other graphic methods that are often used in assessing data from clinical abuse potential studies has been developed and will be shared within and outside of the FDA.

Conclusions: The heat map display for data from clinical abuse potential studies provides a tool to visually examine the data. The tool is not restricted to use for Williams square designs. Such heat map displays have wide applicability and could be used for any crossover studies.