AI Based Predictive Modelling for Internal Quality Control: A Machine Learning Approach Using Altair RapidMiner

Area of Science:

• Clinical Laboratory Science
• Data Science
• Quality Management

Background:

• Traditional Internal Quality Control (IQC) methods are reactive and threshold-based.
• These methods often fail to detect subtle process deviations promptly, risking compromised laboratory quality.

Purpose of the Study:

• To develop and validate a machine learning (ML) model for early detection of IQC deviations.
• To enhance proactive quality management in clinical laboratories using predictive analytics.

Main Methods:

• A retrospective study analyzed 4,572 IQC records across 8 analytes and multiple instruments.
• Three ML classification algorithms (Decision Tree, Gradient Boosted Trees, Random Forest) were built using Altair RapidMiner.
• Models were evaluated using 10-fold cross-validation with metrics including accuracy, precision, recall, F1-score, and ROC-AUC.

Main Results:

• The Random Forest model demonstrated superior performance with 92.0% accuracy, 91.0% precision, 89.4% recall, and 0.932 AUC.
• Key predictors included analyte type, control level, reagent lot, and operator ID.
• The model successfully predicted 68% of future out-of-control events within 24 hours, enabling preventive actions.

Conclusions:

• ML, specifically Random Forest, effectively enhances IQC through predictive monitoring.
• Altair RapidMiner provides an accessible, no-code platform for advanced laboratory analytics.
• This data-driven approach supports Quality 4.0 and real-time decision-making in laboratory quality assurance.