Triple Point Pooled Sera (TriPPS) QC for Laboratory Analyte Error Detection: A Machine Learning based Quality Control in Laboratory.

Background: Reliable internal quality control (IQC) is vital for ensuring analytical accuracy in clinical laboratories. Conventional rule-based QC systems, such as Westgard and Levey-Jennings, often exhibit retrospective detection and limited sensitivity to small but clinically meaningful shifts. This study introduces the Triple-Point Pooled Sera (TriPPS) Quality Control system, a novel, machine-learning-based framework integrating in-house pooled sera with adaptive algorithms for enhanced error detection. Methods: Residual patient sera were pooled to create stable, matrix-relevant IQC material for 60 consecutive analytical days. Sodium and potassium were used as representative analytes. Three complementary machine learning models were applied: k-Nearest Neighbour (k-NN) for trend detection, Isolation Forest (IF) for random error identification, and Gaussian Process Regression (GPR) for systematic bias modeling. Controlled ±1% daily biases and stochastic random errors were introduced to simulate analytical drift. Detection lag, sensitivity, and anomaly classification were evaluated. Results: The k-NN algorithm effectively identified trend errors within 0-2 days of bias onset, while IF accurately detected random fluctuations with minimal false positives. GPR modeled nonlinear systematic drift with high fidelity, capturing bias progression that is overlooked by linear methods. The integration of pooled sera enhanced the system's stability, reproducibility, and cost efficiency across all error types. Conclusion: The TriPPS system demonstrates a scalable, data-driven approach to laboratory quality control by combining pooled sera with machine learning algorithms. This framework enhances analytical vigilance, facilitates proactive error identification, and provides a practical, resource-efficient solution for real-time QC monitoring in clinical chemistry laboratories.