Data-Driven Approach toward Long-Term Equipment Condition Assessment in Sterile Drug Product Manufacturing.

Area of Science:

• Pharmaceutical Manufacturing
• Data Science
• Predictive Maintenance

Background:

• Equipment condition assessment is crucial for reliable drug production.
• Existing methods may struggle with heterogeneous data sources and the need for specialized sensors.

Purpose of the Study:

• To develop and validate a data-driven methodology for long-term equipment condition assessment.
• To leverage existing process monitoring data for predictive maintenance in pharmaceutical manufacturing.

Main Methods:

• A two-stage methodology processing heterogeneous sensor data.
• Anomaly detection using Independent Component Analysis (ICA) to identify outlier patterns.
• Condition monitoring via Principal Component Analysis (PCA) to detect shifts in operational baselines.

Main Results:

• ICA successfully identified clusters of outliers, providing insights into maintenance needs.
• PCA revealed shifts in operational baselines correlated with equipment faults.
• Combined ICA and PCA demonstrated potential for effective equipment condition monitoring.

Conclusions:

• The presented methodology offers a viable approach for predictive maintenance in pharmaceutical settings.
• It effectively utilizes existing process data, reducing the need for new sensor installations.
• This data-driven approach enhances equipment reliability and operational efficiency in drug product manufacturing.