A business analytics approach to augment six sigma problem solving: A biopharmaceutical manufacturing case study

• Published in: Computers in industry 2020-04, Vol.116, p.103153, Article 103153
• Main Authors: Fahey, Will, Jeffers, Paul, Carroll, Paula
• Format: Article
• Language: English
• Subjects: Analytics; Biopharmaceutical; Manufacturing; Six sigma; Time series
• ISSN: 0166-3615; 1872-6194
• DOI: 10.1016/j.compind.2019.103153

•A business analytics methodology to complement the Six Sigma defect reduction framework.•Adapt symbolic representation schemes for large biopharma manufacturing datasets.•A novel approach to identifying and characterising influential variables in complex manufacturing time series data.•Translation of model findings into manufacturing rules which improve process robustness. Biopharmaceutical manufacturers are required to collect extensive observational data sets in order to meet regulatory and process quality monitoring requirements. These datasets contain information that may improve the performance of the production process. Analytics provides a means of extracting this information while Six Sigma provides a means for the insights to be incorporated into production practices. We present a novel framework which combines Six Sigma and Business Analytics. This approach mines large volumes of inline and offline biopharmaceutical production data, allowing the entire production process to be analysed and modelled. The recommendations of the model are represented as manufacturing rules which give actionable insights to improve the performance of the process. The integrated approach delivers promising results from synthetic experiments as well as being applied in practice to a cell culture process.