THE PROBLEM

The client tested multiple batches of pharmaceuticals in their laboratory on daily basis.

The laboratory personnel manually compared test results with the standardized ‘certificates of analysis’.

The client wanted to automate the review of assay reports as the manual approach to quality assurance was time-consuming and resulted in loss of productivity.

THE SOLUTION

Unstructured data processing techniques, powered by Natural Language Processing (NLP) were used to benchmark test results against the specifications in certificates of analysis.

The AI system accepted inputs in pdf/scanned formats and published a tabulated comparison of product specifications as output.

The user interface applied a color code to highlight mismatches if any between the various specifications of test batch and the corresponding certificate of analysis.

THE SUCCESS

In a pilot involving 100 certificates of analysis, AI system exhibited 100% accuracy in identifying mismatches between test and standardized results.

It resulted in extending the scope of the project to more complex product lines.