Dr. Weixian Shi

With modernization in pharmaceutical processing, data that was originally sparsely recorded on paper becomes more easily and frequently recorded in electronic format for the purpose of real time monitoring and control of process, especially with use of process analytical technology (PAT) tools. The availability of such huge amount of data opens up paramount of opportunity for extracting valuable process information which we were not able to do before with limited data recorded on paper. Artificial Neural Net or ANN is one of the most powerful tools that can assist us in better use of the data. In fact, ANN has a wide area of application where high volume of data exchange occurs, such as face identification; object recognition; pattern, gesture, speech, or text recognition; weather forecasting; and stock performance prediction.

Roller compaction is used in a variety of industries, such as foods, chemicals, and pharmaceuticals, where dustless free flowing granules or shaped briquettes of materials are needed. In order to understand the relative performance of ANN compared to the more traditional type of data analysis, we have derived empirical models with a set of Design of Experiment data from a roller compaction study using both ANN and the statistical method. The result suggests that ANN has similar performance in term of ranking factors for importance and the quality of the data model. The advantage of ANN over the statistical method is that it does not require data transformation or parameter adjustment before a meaningful model can be identified. This is especially valuable when the data is generated in free form experiments (i.e., combination of data from various DOEs with different factors) or the data pool may violate certain statistical rules, such as normality.

Biography:

Weixian Shi has a BE and MS in Chemical Engineering from Tsinghua University and a Ph.D. from University of Michigan, Ann Arbor. He’s a member of AICHE and chairs the Mixing and Segregation Sessions in AICHE annual meeting. He has extensive experience in solid dosage forms covering formulation selection and optimization, powder mixing, roller compaction, tabletting and film coating. His expertise includes process design, control, transfer and scale up using risk assessment and QbD principles. He has presented his work in various forums, such as AIChE, AAPS and IFPAC. Recently, he published an article on Powder Technology regarding a novel method to characterize cohesiveness of material and to measure blendability of powder mixing apparatus.

Dr. Weixian Shi

Biography: