Composite multivariate quality control using a system of univariate, bivariate, and multivariate quality control...

Area of Science:

• Analytical Chemistry
• Laboratory Science
• Statistical Process Control

Background:

• Multivariate quality control is essential for complex laboratory processes with numerous simultaneously measured variables.
• Existing methods like T2 and principal component analysis have limitations in detecting various error types and correlation changes.
• There is a need for a robust quality control system that can handle missing data and provide interpretable diagnostics.

Purpose of the Study:

• To introduce and validate a novel Composite Multivariate Quality Control (CMQC) system.
• To demonstrate the CMQC system's ability to detect systematic and random errors in individual or multiple variables.
• To evaluate the CMQC system's capacity for identifying alterations in the correlation structure between variables.

Main Methods:

• Development of a CMQC system incorporating univariate, multivariate, and correlation quality control rules.
• Implementation of control statistics and graphical displays for diagnosing analytical error sources.
• Testing the CMQC procedure on a laboratory dataset with 40 measured variables across characterization and unknown analysis runs.

Main Results:

• The CMQC system successfully detected unacceptable trends, systematic/random errors, and changes in correlation structure.
• The system demonstrated tolerance to missing data and flexibility in rejecting single or multiple variables.
• Control charts and statistics provided clear insights into potential analytical error sources.

Conclusions:

• The proposed CMQC system offers a comprehensive approach to multivariate quality control in laboratory settings.
• CMQC provides advantages over T2 and principal component methods by addressing a wider range of potential analytical issues.
• The system's design facilitates improved process monitoring and error detection in complex analytical workflows.