Noninvasive Optical Flow Analysis of White Blood Cell Dynamics for Enhanced COVID-19 Screening

The COVID-19 pandemic has underscored the urgent need for enhanced first-line screening methods to complement traditional temperature checks and interviews prior to confirmatory polymerase chain reaction or rapid antigen testing. This study investigates the utility of white blood cell (WBC) counts as a predictive biomarker for COVID-19 identification and explores a novel, noninvasive approach for estimating WBC counts. Two key experiments were conducted. First, the predictive power of WBC counts was evaluated for COVID-19 detection using machine learning algorithms on a publicly available dataset. Second, a noninvasive optical flow analysis technique was proposed to estimate WBC counts from capillary blood flow images. The findings revealed that WBC was consistently selected as a significant feature across various feature selection methods. A LinearModel_BAG_L1 algorithm implemented with AutoGluon achieved an area under the receiver operating characteristic curve of 0.81 for COVID-19 prediction. Furthermore, the optical flow analysis method demonstrated a strong positive correlation ( r =0.66) with conventional blood tests in estimating WBC counts. Although WBC counts alone may not provide sufficient diagnostic accuracy, the results highlight their value as a supplementary biomarker for preliminary COVID-19 screening. Additionally, the feasibility of noninvasive WBC estimation suggests promising applications in enhancing current testing frameworks and increasing accessibility to early detection tools.