IoT-Enabled Global Process Validation System
all the previous batches and factors contributing to the drug quality including raw materials , process parameters , plant equipment performance , environment parameters etc ., along with their IPQC test results to predict the results of IPQC tests in real-time while the batch is still under production . Therefore , the predictive streaming analytics enables customers to identify the potential low-quality batches in the making and provides them an opportunity to scrap and discard them well in advance without having the need to perform impending IPQC tests .
Fig . 5 : Tablet Formulation Manufacturing with streaming analytics and predictive analytics
For example , in the above Figure 5 , during Operation 1 Milling , if IPQC 1 test is predicted to fail , the batch can be immediately scrapped , saving the time required to perform actual IPQC 1 test . Similarly , in case both the predicted and actual IPQC 1 test pass , then the actual IPQC2 test is performed only if the predicted IPQC2 after Operation 2-Blending lead to a positive test result , else the batch can be scrapped after Operation 2-Blending without having the need to perform actual IPQC2 test and so on .
Note : Prediction accuracy plays a very critical role and therefore appropriate model governance mechanism is essential to ensure that the right model is working in the background .
- 46 - November 2020