Library For Support Vector Machines Research Articles

Identification of Coal and Gangue Using Visible-Near Infrared (Vis-NIR) Hyperspectral Imaging

Accurate and rapid identification of coal and gangue is a crucial issue for efficient resource utilization and environmental protection. In this study, visible and near infrared (Vis-NIR) hyperspectral imaging was used for the classification of coal and gangue. Principal component analysis was adopted to generate score scatter plots for differentiating specific grouping of samples. Competitive adaptive reweighted sampling and random frog (RF) were employed to screen out the effective wavelengths. Three models, partial least squares, a library for support vector machines, and back-propagation (BP) neural network were developed to identify coal and gangue using effective wavelengths and full spectra as inputs. The results showed that all models provided satisfactory identification rates of over 91.6%. The best performances emerged from BP neural network using effective wavelengths extracted by RF with identification accuracy of 100% in validation sets. It can be concluded that Vis-NIR hyperspectral imaging is a noninvasive analytical method with promising application prospects for coal and gangue identification.

The Prediction of Flow Stress in the Hot Compression of a Ni-Cr-Mo Steel Using Machine Learning Algorithms

The constitutive model refers to the mapping relationship between the stress and deformation conditions (such as strain, strain rate, and temperature) after being loaded. In this work, the hot deformation behavior of a Ni-Cr-Mo steel was investigated by conducting isothermal compression tests using a Gleeble-3800 thermal simulator with deformation temperatures ranging from 800 °C to 1200 °C, strain rates ranging from 0.01 s−1 to 10 s−1, and deformations of 55%. To analyze the constitutive relation of the Ni-Cr-Mo steel at high temperatures, five machine learning algorithms were employed to predict the flow stress, namely, back-propagation artificial neural network (BP-ANN), Random Committee, Bagging, k-nearest neighbor (k-NN), and a library for support vector machines (libSVM). A comparative study between the experimental and the predicted results was performed. The results show that correlation coefficient (R), root mean square error (RMSE), mean absolute value error (MAE), mean square error (MSE), and average absolute relative error (AARE) obtained from the Random Committee on the testing set are 0.98897, 8.00808 MPa, 5.54244 MPa, 64.12927 MPa2 and 5.67135%, respectively, whereas the metrics obtained via other algorithms are all inferior to the Random Committee. It suggests that the Random Committee can predict the flow stress of the steel more effectively.

Soluble solids content monitoring and shelf life analysis of winter jujube at different maturity stages by Vis-NIR hyperspectral imaging

Visible and near infrared (Vis-NIR) hyperspectral imaging was used for soluble solids content (SSC) monitoring and shelf life analysis of winter jujube at different maturity stages. Relationships between SSC of mid-ripe and ripe fruit and their spectral data were investigated using support vector regression (SVR) and partial least squares regression (PLSR) models. The best determination coefficient and residual predictive deviation of the external validation set were 0.837 and 2.47 for mid-ripe winter jujube, and 0.806 and 2.28 for ripe winter jujube, respectively. This implied that the more effective prediction performances for SSC emerged from SVR model using the effective wavelengths screened by successive projections algorithm (SPA). A significant correlation between the spatial distribution of SSC and maturity and shelf life was found in the prediction maps. Furthermore, shelf life was analyzed using a library for support vector machines (LIBSVM) with an accuracy of 89% and 91% for mid-ripe and ripe fruit in the external validation set, respectively. These results indicate the great potential of hyperspectral imaging for quality monitoring and shelf life analysis of postharvest winter jujube.

Research on a Multimodel Fusion Diagnosis Method for Evaporation Ducts in the East China Sea

Evaporation ducts are abnormal states of the atmosphere in the air–sea boundary layer that directly affect the propagation trajectory of electromagnetic (EM) waves. Therefore, an accurate diagnosis of the evaporation duct height (EDH) is important for studying the propagation trajectory of EM waves in evaporation ducts. Most evaporation duct models (EDMs) based on the Monin–Obukhov similarity theory are empirical methods. Different EDMs have different levels of environmental adaptability. Evaporation duct diagnosis methods based on machine learning methods only consider the mathematical relationship between data and do not explore the physical mechanism of evaporation ducts. To solve the above problems, this study observed the meteorological and hydrological parameters of the five layers of the low-altitude atmosphere in the East China Sea on board the research vessel Xiangyanghong 18 in April 2021 and obtained the atmospheric refractivity profile. An evaporation duct multimodel fusion diagnosis method (MMF) based on a library for support vector machines (LIBSVM) is proposed. First, based on the observed meteorological and hydrological data, the differences between the EDH diagnosis results of different EDMs and MMF were analyzed. When ASTD ≥ 0, the average errors of the diagnostic results of BYC, NPS, NWA, NRL, LKB, and MMF are 2.57 m, 2.92 m, 2.67 m, 3.27 m, 2.57 m, and 0.24 m, respectively. When ASTD < 0, the average errors are 2.95 m, 2.94 m, 2.98 m, 2.99 m, 2.97 m, and 0.41 m, respectively. Then, the EM wave path loss accuracy analysis was performed on the EDH diagnosis results of the NPS model and the MMF. When ASTD ≥ 0, the average path loss errors of the NPS model and MMF are 5.44 dB and 2.74 dB, respectively. When ASTD < 0, the average errors are 5.21 dB and 3.46 dB, respectively. The results show that the MMF is suitable for EDH diagnosis, and the diagnosis accuracy is higher than other models.

An approach to recognizing the working cycle stage with small sample data for energy conservation

Energy conservation is an important strategy for low-carbon development. With utilizing stage recognition and control strategies, engineering machinery can achieve energy savings, and reduce carbon emissions. The traditional machine learning approach requires a large amount of data to train models. In addition to many cases where the available data is limited, even if the data volume is sufficient, it often means that a large amount of testing is required, which is difficult for users who are not professionals and may cause further waste, it is necessary to develop methods that can achieve high accuracy even with limited data. This paper proposes an intelligent recognition method based on library for support vector machine (LIBSVM), which uses the relationship between working stages and pressure waveforms of working cycle. Initially, the pressure waveforms are carefully extracted and subjected to a robust filtering process to effectively eliminate noise artifacts. Subsequently, relevant features are extracted from the waveforms, which include important characteristics such as the amplitude and frequency of the pressure waves, then the extracted features are normalized using appropriate techniques. The LIBSVM model is carefully trained using the aforementioned extracted features and label data. Finally, to further enhance the accuracy of the proposed method, an intelligent calibration system is utilized. This system plays a crucial role in ensuring that the results obtained from the machine learning algorithm are effectively calibrated, thereby improving the overall accuracy of the method. A series of experiments have verified the accuracy of the proposed identification method, and the effectiveness of using this method for energy-saving design has been verified through real machine validation.

Detection of small yellow croaker freshness by hyperspectral imaging

Rapid discrimination of fresh and repeatedly thawed small yellow croaker is of great significance for monitoring quality and ensuring consumer safety. In this study, we selected 160 fish samples of similar size and divided them equally into four groups (fresh, freeze-thaw once, freeze-thaw twice and freeze-thaw three times). Hyperspectral images were acquired from fresh and repeatedly frozen-thawed fish samples. The grey scale co-occurrence matrix (GLCM) was then applied to extract texture information from the first three principal component (PC) images, and a library for support vector machines (LIBSVM) were employed to discriminate fresh and repeatedly frozen-thawed fish samples using spectral characteristics, texture features and their fusion, respectively. The results indicated that LIBSVM model using the fused data showed the highest classification accuracy of 96.88%, and freshness degradation of fish samples after three freeze-thaw cycles was observed in the second PC image. In the freshness level for validation model, PLSR model achieved good performance withRV2= 0.90 and RPD= 3.17. The current findings demonstrate that hyperspectral imaging(HSI) is feasible for non-destructive determination of small yellow croaker freshness, providing technical guidance for the storage and marketing of aquatic products.

Research on Fault Diagnosis of PST Electro-Hydraulic Control System of Heavy Tractor Based on Support Vector Machine

Due to the harsh working environment of the tractor, the transmission can often be faulty. In order to ensure the reliability of its operation, it must be monitored and the fault discovered. In this paper, the support vector machine (SVM) method is used. The eigenvector conversion of the original data uses the following eigenvectors: Three fault modes (leakage fault of shift clutch hydraulic cylinder, blockage fault of oil passage, and blockage fault of proportional valve spool) are identified in matrix and laboratory (MATLAB) with the help of the library for support vector machines (LibSVM) toolkit, and the classification accuracy of test samples is 90%. The normal mode of the PST electro-hydraulic system and the three kinds of fault modes mentioned above are discriminated against, and the correct rate of fault diagnosis reaches 95%, which meets the needs of practical engineering. Analysis of the fault recording data of the power shifting transmission shift solenoid valve shows that the difference between fault pressure data and normal data is small, and the value of traffic data is greater. This method can realize the fault mode online recognition based on controller area network (CAN) communication, and the research results provide a theoretical basis for the fault diagnosis of the PST electro-hydraulic control system.

GSVMA: A Genetic Support Vector Machine ANOVA Method for CAD Diagnosis.

BackgroundCoronary artery disease (CAD) is one of the crucial reasons for cardiovascular mortality in middle-aged people worldwide. The most typical tool is angiography for diagnosing CAD. The challenges of CAD diagnosis using angiography are costly and have side effects. One of the alternative solutions is the use of machine learning-based patterns for CAD diagnosis.MethodsHence, this paper provides a new hybrid machine learning model called genetic support vector machine and analysis of variance (GSVMA). The analysis of variance (ANOVA) is known as the kernel function for the SVM algorithm. The proposed model is performed based on the Z-Alizadeh Sani dataset so that a genetic optimization algorithm is used to select crucial features. In addition, SVM with ANOVA, linear SVM (LSVM), and library for support vector machine (LIBSVM) with radial basis function (RBF) methods were applied to classify the dataset.ResultsAs a result, the GSVMA hybrid method performs better than other methods. This proposed method has the highest accuracy of 89.45% through a 10-fold crossvalidation technique with 31 selected features on the Z-Alizadeh Sani dataset.ConclusionWe demonstrated that SVM combined with genetic optimization algorithm could be lead to more accuracy. Therefore, our study confirms that the GSVMA method outperforms other methods so that it can facilitate CAD diagnosis.

Longitudinal clustering analysis and prediction of Parkinson's disease progression using radiomics and hybrid machine learning.

We employed machine learning approaches to (I) determine distinct progression trajectories in Parkinson's disease (PD) (unsupervised clustering task), and (II) predict progression trajectories (supervised prediction task), from early (years 0 and 1) data, making use of clinical and imaging features. We studied PD-subjects derived from longitudinal datasets (years 0, 1, 2 & 4; Parkinson's Progressive Marker Initiative). We extracted and analyzed 981 features, including motor, non-motor, and radiomics features extracted for each region-of-interest (ROIs: left/right caudate and putamen) using our standardized standardized environment for radiomics analysis (SERA) radiomics software. Segmentation of ROIs on dopamine transposer - single photon emission computed tomography (DAT SPECT) images were performed via magnetic resonance images (MRI). After performing cross-sectional clustering on 885 subjects (original dataset) to identify disease subtypes, we identified optimal longitudinal trajectories using hybrid machine learning systems (HMLS), including principal component analysis (PCA) + K-Means algorithms (KMA) followed by Bayesian information criterion (BIC), Calinski-Harabatz criterion (CHC), and elbow criterion (EC). Subsequently, prediction of the identified trajectories from early year data was performed using multiple HMLSs including 16 Dimension Reduction Algorithms (DRA) and 10 classification algorithms. We identified 3 distinct progression trajectories. Hotelling's t squared test (HTST) showed that the identified trajectories were distinct. The trajectories included those with (I, II) disease escalation (2 trajectories, 27% and 38% of patients) and (III) stable disease (1 trajectory, 35% of patients). For trajectory prediction from early year data, HMLSs including the stochastic neighbor embedding algorithm (SNEA, as a DRA) as well as locally linear embedding algorithm (LLEA, as a DRA), linked with the new probabilistic neural network classifier (NPNNC, as a classifier), resulted in accuracies of 78.4% and 79.2% respectively, while other HMLSs such as SNEA + Lib_SVM (library for support vector machines) and t_SNE (t-distributed stochastic neighbor embedding) + NPNNC resulted in 76.5% and 76.1% respectively. This study moves beyond cross-sectional PD subtyping to clustering of longitudinal disease trajectories. We conclude that combining medical information with SPECT-based radiomics features, and optimal utilization of HMLSs, can identify distinct disease trajectories in PD patients, and enable effective prediction of disease trajectories from early year data.

Decreased Nucleus Accumbens Connectivity at Rest in Medication-Free Patients with Obsessive-Compulsive Disorder.

Background Patients with obsessive-compulsive disorder (OCD) experience deficiencies in reward processing. The investigation of the reward circuit and its essential connectivity may further clarify the pathogenesis of OCD. Methods The current research was designed to analyze the nucleus accumbens (NAc) functional connectivity at rest in medicine-free patients with OCD. Forty medication-free patients and 38 gender-, education-, and age-matched healthy controls (HCs) were recruited for resting-state functional magnetic resonance imaging. Seed-based functional connectivity (FC) was used to analyze the data. LIBSVM (library for support vector machines) was designed to identify whether altered FC could be applied to differentiate OCD. Results Patients with OCD showed remarkably decreased FC values between the left NAc and the bilateral orbitofrontal cortex (OFC) and bilateral medial prefrontal cortex (MPFC) and between the right NAc and the left OFC at rest in the reward circuit. Moreover, decreased left NAc-bilateral MPFC connectivity can be deemed as a potential biomarker to differentiate OCD from HCs with a sensitivity of 80.00% and a specificity of 76.32%. Conclusion The current results emphasize the importance of the reward circuit in the pathogenesis of OCD.

Maturity determination at harvest and spatial assessment of moisture content in okra using Vis-NIR hyperspectral imaging

Maturity determination of fresh okra fruit is a crucial issue for farmers to optimize harvest date for good taste and economic return. In this study, visible and near infrared (Vis-NIR) hyperspectral imaging was employed to evaluate the maturity stage and moisture content of fresh okra fruit precisely. Immature, mature and post-mature okra samples were identified by measuring the shear force, and physicochemical analysis indicated the negative correlation between maturity and moisture content. A library for support vector machines (LIBSVM) model was developed using effective wavelengths, texture features and their fusion, respectively. The LIBSVM model using the fused dataset obtained the highest total maturity classification accuracy reaching 91.7 % for cross-validation. Further investigation was conducted to predict moisture content in okra with different maturity by multiple linear regression (MLR) model, and the determination coefficient, the root mean square error and residual predictive deviation of cross-validation was Rcv2 = 0.816, RMSECV = 1.348 %, RPDcv = 2.333, respectively. After that spatial distribution map was generated to visualize the moisture content in okra fruit. These results demonstrated the potential of hyperspectral imaging for maturity determination and moisture content changes during growth, providing the support for the development of field monitoring sensor to determine the optimum harvest date of okra fruit.

An extended Kalman filter based data-driven method for state of charge estimation of Li-ion batteries

Accurate state-of-charge (SOC) estimation, which can effectively prevent battery overcharge and over-discharge, provide accurate driving range and extend battery life, is challenging due to complicated battery dynamics and ever-changing ambient conditions. In this paper, an extended Kalman filter (EKF) based data-driven method for SOC estimation of Li-ion Batteries is proposed. First, the model characteristic parameters are dynamically tracked through the recursive least squares method. Then the filtered output result of the EKF algorithm reflecting the battery's dynamic characteristics is used as the training data of the extreme gradient boosting (XGBoost) model. Benefit from the XGBoost's excellent machine learning and predictive capabilities, which can realize the battery's SOC high-precision prediction based on the EKF algorithm. The simulation results show that the proposed XGBoost algorithm, especially in the low SOC range, has good convergence and robustness, compared with the EKF algorithm and the library for support vector machines (LIBSVM) algorithm. The algorithm is validated under different driving conditions and achieves accurate SOC estimation with a maximum absolute error of less than 2%. It can be easily concluded that the proposed method can achieve accurate and stable SOC estimation, effectively avoiding the open-loop risk of data-driven algorithms.

Cancer Classification with a Cost-Sensitive Naive Bayes Stacking Ensemble.

Ensemble learning combines multiple learners to perform combinatorial learning, which has advantages of good flexibility and higher generalization performance. To achieve higher quality cancer classification, in this study, the fast correlation-based feature selection (FCBF) method was used to preprocess the data to eliminate irrelevant and redundant features. Then, the classification was carried out in the stacking ensemble learner. A library for support vector machine (LIBSVM), K-nearest neighbor (KNN), decision tree C4.5 (C4.5), and random forest (RF) were used as the primary learners of the stacking ensemble. Given the imbalanced characteristics of cancer gene expression data, the embedding cost-sensitive naive Bayes was used as the metalearner of the stacking ensemble, which was represented as CSNB stacking. The proposed CSNB stacking method was applied to nine cancer datasets to further verify the classification performance of the model. Compared with other classification methods, such as single classifier algorithms and ensemble algorithms, the experimental results showed the effectiveness and robustness of the proposed method in processing different types of cancer data. This method may therefore help guide cancer diagnosis and research.

Effective Estimation of Hourly Global Solar Radiation Using Machine Learning Algorithms

The precise estimation of solar radiation is of great importance in solar energy applications with respect to installation and capacity. In estimate modelling on selected target locations, various computer-based and experimental methods and techniques are employed. In the present study, the Multilayer Feed-Forward Neural Network (MFFNN), K -Nearest Neighbors ( K -NN), a Library for Support Vector Machines (LibSVM), and M5 rules algorithms, which are among the Machine Learning (ML) algorithms, were used to estimate the hourly average solar radiation of two geographic locations on the same latitude. The input variables that had the most impact on solar radiation were identified and grouped as a result of 29 different applications that were developed by using 6 different feature selection methods with Waikato Environment for Knowledge Analysis (WEKA) software. Estimation models were developed by using the selected data groups and all input variables for each target location. The results show that the estimations developed with the feature selection method were more successful for target locations, and the radiation potentials were similar. The performance of the estimation models was evaluated by comparing each model with different statistical indicators and with previous studies. According to the RMSE, MAE, R 2 , and SMAPE statistical scales, the results of the most successful estimation models that were developed with MFFNN were 0.0508-0.0536, 0.0341-0.0352, 0.9488-0.9656, and 7.77%-7.79%, respectively.

Different sEMG and EEG Features Analysis for Gait phase Recognition.

This research focuses on the gait phase recognition using different sEMG and EEG features. Seven healthy volunteers, 23-26 years old, were enrolled in this experiment. Seven phases of gait were divided by three-dimensional trajectory of lower limbs during treadmill walking and classified by Library for Support Vector Machines (LIBSVM). These gait phases include loading response, mid-stance, terminal Stance, pre-swing, initial swing, mid-swing, and terminal swing. Different sEMG and EEG features were assessed in this study. Gait phases of three kinds of walking speed were analyzed. Results showed that the slope sign change (SSC) and mean power frequency (MPF) of sEMG signals and SSC of EEG signals achieved higher accuracy of gait phase recognition than other features, and the accuracy are 95.58% (1.4 km/h), 97.63% (2.0 km/h) and 98.10% (2.6 km/h) respectively. Furthermore, the accuracy of gait phase recognition in the speed of 2.6 km/h is better than other walking speeds.

Detection and Segmentation of Cement Concrete Pavement Pothole Based on Image Processing Technology

Potholes are the most common form of distress on cement concrete pavements, which can compromise pavement safety and ridability. Thus, timely and accurate pothole detection is an important task in developing proper maintenance strategies and ensuring driving safety. This paper proposes a method of integrating the processing of grayscale and texture features. This method mainly combines industrial camera to realize rapid and accurate detection of pothole. Image processing techniques including texture filters, image grayscale, morphology, and extraction of the maximum connected domain are used synergistically to extract useful features from digital images. A machine learning model based on the library for support vector machine (LIBSVM) is constructed to distinguish potholes from longitudinal cracks, transverse cracks, and complex cracks. The method is validated using data collected from agricultural and pastoral areas of Inner Mongolia, China. The comprehensive experiments for recognition of potholes show that the recall, precision, and F1-Score achieved are 100%, 97.4%, and 98.7%, respectively. In addition, the overlap rate between the extracted pothole region and the original image is estimated. Images with an overlap rate greater than 90% accounted for 76.8% of the total image, and images with an overlap rate greater than 80% accounted for 94% of the total image. A comparison discloses that the proposed approach is superior to the existing method not only from the perspective of the accuracy of pothole detection but also from the perspective of the segmentation effect and processing efficiency.

Machine Learning in Prediction of Second Primary Cancer and Recurrence in Colorectal Cancer.

Background: Colorectal cancer (CRC) is the third commonly diagnosed cancer worldwide. Recurrence of CRC (Re) and onset of a second primary malignancy (SPM) are important indicators in treating CRC, but it is often difficult to predict the onset of a SPM. Therefore, we used mechanical learning to identify risk factors that affect Re and SPM.Patient and Methods: CRC patients with cancer registry database at three medical centers were identified. All patients were classified based on Re or no recurrence (NRe) as well as SPM or no SPM (NSPM). Two classifiers, namely A Library for Support Vector Machines (LIBSVM) and Reduced Error Pruning Tree (REPTree), were applied to analyze the relationship between clinical features and Re and/or SPM category by constructing optimized models.Results: When Re and SPM were evaluated separately, the accuracy of LIBSVM was 0.878 and that of REPTree was 0.622. When Re and SPM were evaluated in combination, the precision of models for SPM+Re, NSPM+Re, SPM+NRe, and NSPM+NRe was 0.878, 0.662, 0.774, and 0.778, respectively.Conclusions: Machine learning can be used to rank factors affecting tumor Re and SPM. In clinical practice, routine checkups are necessary to ensure early detection of new tumors. The success of prediction and early detection may be enhanced in the future by applying “big data” analysis methods such as machine learning.

Intelligent identification for working-cycle stages of excavator based on main pump pressure

Abstract An identification of the working stages of the excavator is used as the basis to realize a staged energy-saving control and reduce the fuel consumption. However, existing methods based on computer vision and multi-sensor information fusion technologies have limitations in practical applications. In this paper, an intelligent identification method for the working-cycle stages of an excavator is proposed based on the relationship between the working stages and the main pump pressure waveform. Three machine learning algorithms, a support vector machine (SVM), back propagation neural network (BPNN), and logistic regression (LR), were used to establish the intelligent recognition model. In addition, an intelligent calibration system was established to reduce the effects of human-induced factors and pressure fluctuations on the accuracy when using the original main pump pressure signals. The relationships among the identification accuracy, time window width, and signal sampling frequency were also studied. The results show that the library for support vector machine (LIBSVM) model outperforms the BPNN and LR, recording an accuracy of 94.64% when using the test set and an accuracy of 93.82% with an intelligent calibration system when using the original main pump pressure signals. The optimal time window width and signal sampling frequency are 0.5 s and 20 Hz, respectively. The proposed method is a practical approach to realize the real-time supervision of the working stage of an excavator, providing theoretical support for a staged energy-saving control.

Increased Nucleus Accumbens Connectivity in Resting-State Patients With Drug-Naive, First-Episode Somatization Disorder.

The nucleus accumbens (NAc) plays an important role in the reward circuit, and abnormal regional activities of the reward circuit have been reported in various psychiatric disorders including somatization disorder (SD). However, few researches are designed to analyze the NAc connectivity in SD. This study was designed to explore the NAc connectivity in first-episode, drug-naive patients with SD using the bilateral NAc as seeds. Twenty-five first-episode, drug-naive patients with SD and 28 healthy controls were recruited. Functional connectivity (FC) was designed to analyze the images. LIBSVM (a library for support vector machines) was used to identify whether abnormal FC could be utilized to discriminate the patients from the controls. The patients showed significantly increased FC between the left NAc and the right gyrus rectus and left medial prefrontal cortex/anterior cingulate cortex (MPFC/ACC), and between the right NAc and the left gyrus rectus and left MPFC/ACC compared with the controls. The patients could be separated from the controls through increased FC between the left NAc and the right gyrus rectus with a sensitivity of 88.00% and a specificity of 82.14%. The findings reveal that patients with SD have increased NAc connectivity with the frontal regions of the reward circuit. Increased left NAc-right gyrus rectus connectivity can be used as a potential marker to discriminate patients with SD from healthy controls. The study thus highlights the importance of the reward circuit in the neuropathology of SD.

Microcontroller Implementation of Support Vector Machine for Detecting Blood Glucose Levels Using Breath Volatile Organic Compounds.

This paper presents an embedded system-based solution for sensor arrays to estimate blood glucose levels from volatile organic compounds (VOCs) in a patient’s breath. Support vector machine (SVM) was trained on a general-purpose computer using an existing SVM library. A training model, optimized to achieve the most accurate results, was implemented in a microcontroller with an ATMega microprocessor. Training and testing was conducted using artificial breath that mimics known VOC footprints of high and low blood glucose levels. The embedded solution was able to correctly categorize the corresponding glucose levels of the artificial breath samples with 97.1% accuracy. The presented results make a significant contribution toward the development of a portable device for detecting blood glucose levels from a patient’s breath.