(238b) Estimation of Product Robustness: Prediction of Manufacturing Variability

Tabora, J. E., Bristol-Myers Squibb
Domagalski, N., Bristol-Myers Squibb
Albrecht, J., Bristol-Myers Squibb
Behling, R., Bristol-Myers Squibb
Li, J., Bristol-Myers Squibb
Sentveld, G., New Brunswick
Crison, J. R., Bristol-Myers Squibb
Walsh, J., Bristol-Myers Squibb

This presentation will describe an approach to leverage process models and historical data of potential

sources of variability (materials, process parameters, and analytical measurements) to estimate the

process performance under defined manufacturing control strategies.

Using Monte Carlo tools, the process performance is estimated as a distribution of the possible outcomes

of the measured process quality attributes. The outcomes are generated from a Bayesian network that

integrates all the sources of variability associated with the process.

Applied to process models generated during process development, this framework will be useful in

providing a statistically sound, quantitative level of risk associated with process performance under

different control and monitoring strategies as well as informing technology transfer/validation and

process robustness decisions.