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

Background: Internal Quality Control (IQC) ensures accuracy and reliability in laboratory testing but traditionally relies on reactive, threshold-based methods. These approaches often fail to detect subtle process deviations in time, potentially compromising quality. Objective: To develop and validate a machine learning-based predictive model for early detection of IQC deviations using Altair RapidMiner, enhancing proactive quality management in clinical laboratories. Methods: A retrospective analytical study was conducted using 4,572 IQC records from Meenakshi Labs, covering 8 analytes across multiple instruments. Data preprocessing included cleaning, feature engineering, and encoding. Three classification algorithms - Decision Tree, Gradient Boosted Trees, and Random Forest - were developed using Altair RapidMiner's no-code environment. Models were evaluated via 10-fold cross-validation using accuracy, precision, recall, F1-score, and ROC-AUC metrics. Results: The Random Forest model outperformed others with 92.0% accuracy, 91.0% precision, 89.4% recall, and an AUC of 0.932. Key predictive features included analyte type, control level, reagent lot, and operator ID. The model correctly predicted 68% of future out-of-control events within a 24-hour window, demonstrating potential for preventive action. Feature importance analysis enhanced model interpretability. Conclusion: Machine learning, particularly Random Forest, effectively augments IQC by enabling predictive monitoring. Altair RapidMiner offers a user-friendly platform, making advanced analytics accessible even without programming skills. This approach aligns with Quality 4.0 initiatives, promoting data-driven, real-time decision-making in laboratory quality assurance.