Application of data mining for monitoring turnaround time in clinical laboratories: A workflow-based model using Orange visual programming

INTRODUCTION: Turnaround time (TAT) is a key quality indicator in laboratory medicine, but routine department-level mon-itoring is limited by fragmented data and scarce analytical resources. We aimed to develop and evaluate a modular, reproducible data analytics workflow—built in the open-source Orange visual programming platform and integrated with hospital information management system (HIMS) data—to monitor laboratory TAT, localize delays, and support quality improvement interventions.
METHODS: Approximately 3.75 million timestamped records from a 1,875-bed tertiary care hospital were extracted over three consecutive months. A six-phase Orange pipeline performed data import, concatenation, filtering, unit-level stratification, One-Class Support Vector Machine outlier exclusion, and visualization. Monthly reports were submitted to the laboratory quality committee and prompted targeted interventions. Unit-level monthly means were compared using the Friedman and Wilcoxon signed-rank tests.
RESULTS: Order-to-phlebotomy time decreased significantly across the three months (Friedman p=0.006; −17.5% with Wilcoxon p=0.005 after excluding one administrative unit). Sample reception time varied significantly (p=0.004) owing to a February peak, with no sustained net change by March (p=0.93). Total analytical TAT fell by 19.1% overall (im-munoassay: −23.7%; clinical chemistry: −22.3%)—a consistent trend that did not reach statistical significance at the test-group level (p=0.074).
DISCUSSION AND CONCLUSION: A low-code, visual Orange workflow enabled reproducible, near-real-time monthly TAT monitoring and helped target interventions, offering an accessible alternative to resource-intensive approaches.

Keywords: Clinical laboratory services, data mining, quality improvement, quality indicators, health care, time factors, workflow