Patient-based feedback control for erythroid variables obtained using in-house automated hematology analyzers in veterinary medicine.

The research article discusses the application of analytic algorithms for assessing the performance and mitigation of biases in automated hematology analyzers used in veterinary medicine, in particular those analyzing samples from dogs, cats, and horses.

Objective of the Study

• The primary objective of this research is to test and then apply a method of analyzing weighted moving averages – a statistical technique commonly used in quality control – to monitor the performance of hematology analyzers using animal patient samples. The study aims to use these findings to adjust for biases created by these instruments when analyzing batches of samples.

Methods Used in the Research

• For the purpose of the study, an array of algorithms were developed and applied to sequential batches of 20 samples at a time.
• Repeated samples within a batch or large populations of samples with similar abnormalities were purposely excluded to prevent skewing of results.
• Six hematologic variables, namely red blood cell (RBC) count, hematocrit (HCT), hemoglobin (HGB) concentration, mean corpuscular volume (MCV), mean corpuscular hemoglobin (MCH), and mean corpuscular hemoglobin concentration (MCHC), were grouped into batches of weighted moving averages for analysis.
• The data was then analyzed using control chart rules, a gradient descent algorithm, and fuzzy logic to define and adjust for biases.

Results and Findings

• The study revealed biases in 102 hematology analyzers used for assessment of RBC count, HCT, HGB concentration, MCV, MCH, and MCHC.
• After the adjustment analysis, all variables except HGB concentration needed an adjustment, with only a small adjustment needed for RBC counts, and MCV requiring the greatest adjustment.
• These adjustments were then validated by comparing packed cell volumes (PCVs) with the original and adjusted HCT values.

Conclusion and Implication

• This study proposes a system that mitigates system bias for several crucial hematologic variables using a feedback control approach.
• However, it also concludes that several fundamental assumptions need to be satisfied for this approach to function correctly, indicating areas of further study and caution when applying these methods in practical scenarios.