Advanced bioimpedance analysis for infectious disease risk assessment via neural network classifiers

Area of Science:

• Biomedical Engineering
• Computational Biology
• Infectious Disease Research

Background:

• Assessing infectious disease risk and complications requires accurate diagnostic tools.
• Traditional methods like X-ray have limitations in sensitivity and specificity.

Purpose of the Study:

• To develop a novel neural network classification model for analyzing biomaterial impedance in living systems.
• To evaluate the effectiveness of this model in assessing infectious disease risk, specifically viral pneumonia.

Main Methods:

• Developed a neural network model using multidimensional bioimpedance measurements.
• Employed a modified Voigt model to represent bioimpedance and extracted descriptors.
• Utilized iterative algorithms to generate Voigt models and trained neural network classifiers.
• Conducted experimental validation on patients with viral pneumonia, comparing bioimpedance analysis with X-ray.

Main Results:

• The hybrid bioimpedance classifiers demonstrated effectiveness in assessing infectious disease risk.
• Bioimpedance analysis achieved 79% accuracy in pneumonia detection, outperforming X-ray at 77% accuracy.
• Cole plots were generated from bioimpedance data for analysis.

Conclusions:

• The developed bioimpedance analysis approach shows significant potential for revolutionizing infectious disease diagnosis and treatment.
• This method can improve the accuracy of infection risk assessment and mitigate complications.
• The bioimpedance method offers a more accurate alternative to X-ray for pneumonia diagnosis.