Optimization of the Preanalytical Phase by Estimating Serum Indices Using an Automated Classifier

Area of Science:

• Clinical Chemistry
• Laboratory Medicine
• Data Science in Healthcare

Background:

• Preanalytical errors significantly impact laboratory testing accuracy.
• Hemolysis, lipemia, and icterus are common preanalytical interferences.
• Accurate identification of these interferences is crucial for reliable results.

Purpose of the Study:

• To evaluate the RSD classifier for serum interference estimation.
• To identify additional biochemical parameters affected by preanalytical interferences.
• To determine the economic impact of implementing this methodology.

Main Methods:

• Collected serum indices from 65,529 requests using the RSD system and Cobas platforms.
• Employed data mining to find associations between serum indices and laboratory tests.
• Evaluated interference impact on 91 biochemistry, immunoassay, and coagulation tests.

Main Results:

• The RSD model demonstrated 94.4% accuracy in compatibility with analytical methods.
• Implementation of RSD-based serum indices estimation could save €10,561 annually by avoiding unnecessary tests.
• Identified iron, transferrin, fibrinogen, and alkaline phosphatase as tests affected by hemolysis.

Conclusions:

• The RSD classifier is an effective method for estimating serum indices and bilirubin values.
• Implementing this methodology can save over €15,000 annually without increasing turnaround times.
• Iron, alkaline phosphatase, transferrin, and fibrinogen should be considered in interference evaluation protocols.