Development of a Portable Dielectric Biosensor for Rapid Detection of Viscosity Variations and Its In Vitro Evaluations Using Saliva Samples of COPD Patients and Healthy Control

Abstract

Chronic Obstructive Pulmonary Disease (COPD) is a life-threatening lung disease affecting millions of people worldwide. Although the majority of patients with objective COPD go undiagnosed until the late stages of their disease, recent studies suggest that the regular screening of sputum viscosity could provide important information on the disease detection. Since the viscosity of sputum is mainly defined by its mucin–protein and water contents, dielectric biosensors can be used for detection of viscosity variations by screening changes in sputum’s contents. Therefore, the objective of this work was to develop a portable dielectric biosensor for rapid detection of viscosity changes and to evaluate its clinical performance in characterizing viscosity differences of saliva samples collected from COPD patients and Healthy Control (HC). For this purpose, a portable dielectric biosensor, capable of providing real-time measurements, was developed. The sensor performance for dielectric characterization of mediums with high water content, such as saliva, was evaluated using isopropanol–water mixtures. Subsequently, saliva samples, collected from COPD patients and HC, were investigated for clinical assessments. The radio frequency biosensor provided high repeatability of 1.1% throughout experiments. High repeatability, ease of cleaning, low-cost, and portability of the biosensor made it a suitable technology for point-of-care applications.

In-Vitro Classification of Saliva Samples of COPD Patients and Healthy Controls Using Machine Learning Tools

“…As shown in Fig. 1, a previously developed permittivity biosensor (IHP Microelectronics, Frankfurt Oder, Germany) was used for the dielectric characterization of saliva samples [14], [15]. The output of the biosensor was extracted into an Excel file using a user-friendly data acquisition (PLX-DAQ) interface, as demonstrated in Fig.…”

“…Consequently, only 80 samples out of the available 239 were dielectrically characterized in this study. As extensively reported in the previous work, the developed biosensor is capable of measuring both real and imaginary parts of the permittivity of a material-under-test [14], [15]. From a physical point of view, the real part of the permittivity represents a material's energy absorption (dielectric properties) in an interaction with an electromagnetic field; while the imaginary part of it is an indicator of a material's energy loss (conductivity properties).…”

“…All measurements were conducted in a lab environment with a controlled room temperature following a primary cleaning procedure using ethanol and compressed air for the removal of extraneous particles from the sensor surface. It is noteworthy that temperature fluctuations can possibly impair the biosensor performance in realworld applications as part of measurement uncertainties associated with this system [14], [15]. As a solution, the ambient temperature can be introduced as an input variable into the ML model [26].…”

“…Upon the entry of water into mucin's gels matrix, there is a considerable amount of proton release resulting from cations exchange (particularly Ca and Na), which drastically changes the dielectric properties of mucin samples [13]. In other words, sputum samples collected from COPD patients are expected to have different permittivity characteristics compared to samples of Healthy Controls (HC), which could be used as a biomarker for the assessment of the disease in a Point-of-Care (PoC) environment [14], [15]. However, due to complexities of obtaining sputum samples non-invasively on a daily-basis, saliva could be an alternative with better patient compliance for PoC applications [16].…”

“…Although the dielectric characterization of saliva samples can potentially shine a spotlight onto the detection of COPD in a PoC setting, the comprehensive diagnosis of the disease requires a sophisticated algorithm by concurrent consideration of all essential parameters related to a patient's personal and medical backgrounds [15]. These demographic param-eters include, but are not limited to, age, gender, weight, cytokine level, pathogen load, and the smoking background of subjects [15]. Therefore, without analytical insight, information obtained on one specific parameter has a low clinical value for the disease diagnosis [17].…”