72. Identification and prediction of flaw, taint, and contamination causes using machine learning and artificial intelligence on sensory data and biochemical and metabolic pathway tracing

Z. BUSHMAN (1), R. Ahn (1), J. M. Cohen (1), Y. Zhou (1); (1) Analytical Flavor Systems LLC., State College, PA, U.S.A.


Latent flaws and contaminations, undetectable by most sensory and chemical based quality control programs, pose a real and existential risk to breweries—how can you detect and fix the cause of a flaw that has not yet developed? In this research, we present a novel approach to predicting latent flaws and tracing their creation back to the root cause in the beer brewing process. Furthermore, we show that the process is robust enough to predict flaws that occur in beer after packaging and distribution, increasing the actionability of any quality control program. At Analytical Flavor Systems we use machine learning and artificial intelligence to build quality control and flavor profiling tools for the food and beverage industry. By applying our algorithms to human sensory data collected with the Gastrograph review mobile application, predictions can be made as to the likelihood of a flaw appearing and how to prevent, delay, or mitigate the flaw.

Zachary Bushman is a chemist at Analytical Flavor Systems (AFS) and an avid home brewer. He received his B.S. degree in chemistry from the University of Wisconsin–Platteville in 2013. He then attended graduate school at the Pennsylvania State University, pursuing an advanced degree in chemistry. Upon withdrawal from graduate school he began working at AFS in State College, PA. AFS is a company dedicated to flavor profiling and quality control in the craft beverage industry. He is now the head chemist at AFS and directs projects on flaw detection and hardware development.

View Presentation