An in vivo bioelectronic nose utilizing support vector machine techniques for quantitative assessment of the odor detection performance

The biological olfactory system-based in vivo bioelectronic nose (BEN) exhibits significant advantages in terms of sensitivity, selectivity, and response time. This study presents a methodology that integrates principal component analysis (PCA) with support vector machine (SVM) techniques to facilitate the classification and prediction of odorants within in vivo BEN systems. A bioengineered in vivo BEN system was created through the overexpression of the olfactory receptor OR3, aimed at enhancing the specificity of odor detection in targeted environments. The performance of two distinct in vivo BEN systems was quantitatively assessed and compared utilizing SVM. The developed system demonstrated robust multi-class discrimination capabilities, achieving accurate classification of single odorants across varying concentrations (10⁻³ to 10⁻¹³ M) with strong linear correlation (R² = 0.97) between predicted and actual concentrations. The BEN system incorporating novel sensitive elements exhibited 7 % higher classification accuracy for 6 odors, along with stable odor discrimination for 40 days post-implantation. The PCA-SVM framework outperformed conventional classifiers in cross-validation tests and achieved 70 % classification accuracy under background interference. These advancements establish a scalable platform for real-world odorant analysis. Future validation in complex environmental/clinical settings is warranted to confirm its efficacy in applied scenarios such as environmental monitoring and precision diagnostics.