Development of a Data Processing Algorithm for the Electronic Nose Based on Piezoelectric Sensors in Blood Sample Analysis without Sample Preparation: A Pilot Study

Abstract

The study presents the first results of blood analysis conducted without sample preparation using a portable “electronic nose” based on piezoelectric sensors in hospital departments of various specialization. Over 6 months, clinical laboratory personnel at a regional hospital performed parallel blood analyses using both traditional and sensor-based methods. The study evaluates the impact of environmental conditions—including room temperature, measurement repetition frequency, and the nature of piezoelectric sensor electrode modifiers—on the signal reproducibility within the sensor array. The authors propose effective approaches and algorithms for the processing of the data of a multidimensional piezoelectric sensor array in the detection of the volatile organic compound (VOC) profile of small-volume blood samples (≤0.5 mL). For blood analysis without sample preparation, second-grade distilled water serves as an effective internal standard under laboratory conditions. The study includes blood samples from 250 patients, with the sensor array reliably differentiating cases of pronounced inflammatory pathologies, oncology, severe renal dysfunction, and extreme stress conditions (e.g., surgical procedures, traffic accidents with fatal injuries, or burns). The proposed parameter also identifies other pathological conditions; however, its magnitude varies based on individual patient characteristics, comorbidities, disease compensation, and the severity of pathological processes on admission (e.g., type 2 diabetes mellitus). A component-by-component analysis of VOC profiles for samples with significant pathologies will be addressed in a subsequent report. The study optimizes VOC detection methods, including measurement mode and repetition frequency, and introduces simple yet effective algorithms for the processing of sensor array data.