Machine Learning-based Classification Research Articles

Estimating afforestation related forest cover change using data fusion and machine learning

Abstract Climate change threatens natural ecosystems, including biodiversity, and has emerged as the greatest challenge to human well-being. To support sustainable forest use, it is important to accurately map the distribution of afforestation and how the forests have changed over time. Climate change is affected by land use and land cover (LULC) since it alters biogeochemical cycles. Pakistan's declining forest cover has motivated policymakers to undertake the Green Growth Initiative (GGI). In accordance with its commitment to the Bonn Challenge, Pakistan executed the Billion Tree Afforestation Project (BTAP) to restore forests covering approxi-mately 0.35 million Ha. To estimate forest cover mapping accuracy, we developed a novel method to combine Sentinel-1, Sentinel-2 data, and vegetation indices to assess forest cover change. The methods and models were used to create forest maps for the period of 2016 to 2022. The optimal mapping approach was chosen after classifying the land cover using two machine learning-based classifiers: random forests (RF) and support vector machines (SVM). Finally, the spatiotemporal distribution of the changes induced by afforestation was generated. Based on the classifications of land cover, forest and non-forest identification was carried out. The results show that RF produces more accurate results, with kappa coefficient of 94-97% & 0.93-0.96 respectively. The classification accuracy and Kappa coefficient of the SVM model ranges from 92-94% & 0.91-0.95. Overall, forested areas in the study area increased by 3.7%. The techniques used in this study are cost-effective for accurately monitoring changes in forest cover.

Machine learning-based classification of valvular heart disease using cardiovascular risk factors

Valvular Heart Disease (VHD) is a globally significant cause of mortality, particularly among aging populations. Despite advancements in percutaneous and surgical interventions, there are still uncertainties that remain regarding the risk factors that significantly contribute to this condition within the domain of cardiovascular disease. This study investigates these uncertainties and the role of machine learning in categorizing VHD based on cardiovascular risk factors. It follows a two-part investigation comprising feature extraction and classification phases. Feature extraction is initially performed using a wrapping approach and refined further with binary logistic regression. The second phase employs five classifiers: Artificial Neural Network (ANN), XGBoost, Random Forest (RF), Naïve Bayes, and Support Vector Machine (SVM), along with advanced methods such as SVM combined with Principal Component Analysis (PCA) and a majority-voting ensemble method (MV5). Data on VHD cases were collected from DHQ Hospital Faisalabad using simple random sampling. Various statistical measures, such as the ROC curve, F-measure, sensitivity, specificity, accuracy, MCC, and Kappa are applied to assess the results. The findings reveal that the combination of SVM with PCA achieves the highest overall performance while the MV5 ensemble method also demonstrates high accuracy and balance in sensitivity and specificity. The variation in VHD prevalence linked to specific risk factors highlights the importance of a comprehensive approach to reduce this disease’s burden. The Exceptional performance of SVM + PCA and MV5 highlights their significance in diagnosing VHD and advancing knowledge in biomedicine.

Automated Classification of Coronary Plaque on Intravascular Ultrasound by Deep Classifier Cascades.

Intravascular ultrasound (IVUS) is the gold standard modality for in vivo visualization of coronary arteries and atherosclerotic plaques. Classification of coronary plaques helps to characterize heterogeneous components and evaluate the risk of plaque rupture. Manual classification is time-consuming and labor-intensive. Several machine learning-based classification approaches have been proposed and evaluated in recent years. In the current study, we develop a novel pipeline composed of serial classifiers for distinguishing IVUS images into five categories: normal, calcified plaque, attenuated plaque, fibrous plaque, and echolucent plaque. The cascades comprise densely connected classification models and machine learning classifiers at different stages. Over 100,000 IVUS frames of five different lesion types were collected and labeled from 471 patients for model training and evaluation. The overall accuracy of the proposed classifier is 0.877, indicating that the proposed framework has the capacity to identify the nature and category of coronary plaques in IVUS images. Further, it may provide real-time assistance on plaque identification and facilitate clinical decision-making in routine practice.

Explainable olive grove and grapevine pest forecasting through machine learning-based classification and regression

Pests significantly impact agricultural productivity, making early detection crucial for maximizing yields. This paper explores the use of machine learning models to predict olive fly and red spider mite infestations in Andalusia. Four datasets on crop phenology, pest populations, and damage levels were used, with models developed using the Python package H2O, which focuses on interpretability through SHAP values and ICE plots. The results showed high precision in predicting pest outbreaks, particularly for the olive fly, with minimal differences between models using feature selection. In the vineyard dataset, the selection of characteristics improved the performance of the model by reducing the MAE and increasing R2. Explainability techniques identified solar radiation and wind direction as key factors in olive fly predictions, while past pest occurrences and wind velocity were influential for red spider mites, providing farmers with actionable insights for timely pest control.

Machine Learning-Based Classification System for Tuberculosis Detection Using Locally Collected Radiographs

Machine Learning-Based Classification System for Tuberculosis Detection Using Locally Collected Radiographs

Machine learning-based classification of varicocoele grading: A promising approach for diagnosis and treatment optimization.

Varicocoele is a correctable cause of male infertility. Although physical examination is still being used in diagnosis and grading, it gives conflicting results when compared to ultrasonography-based varicocoele grading. We aimed to develop a multi-class machine learning model for the grading of varicocoeles based on ultrasonographic measurements. Between January and May 2024, we enrolled unilateral varicocoele patients at an infertility clinic, assessing their varicocoele stages using the Dubin and Amelar system. We measured vascular diameter and reflux time at the testicular apex and the subinguinal region ultrasonography in both the supine and standing positions. Using these measurements, we developed four multi-class machine learning models, evaluating their performance metrics and determining which patient position and projection were most influential in varicocoele grading. We included 248 patients with unilateral varicocoele in the study, their average age was 26.61±4.95 years old. Of these, 212 had left-sided and 36 had right-sided varicocoeles. According to the Dubin and Amelar system, there were 66 grade I, 96 grade II, and 86 grade III varicocoeles. Among the models we created, the random forest (RF) model performed best, with an overall accuracy of 0.81±0.06, an F1 score of 0.79±0.02, a sensitivity of 0.69±0.02, and a specificity of 0.8±0.03. Vascular diameter measurement at the testicular apex in the supine position had the most impact on grading across all models. In support vector machine and multi-layer perceptron models, reflux time measurements from the subinguinal projection in the standing position contributed the most, while in RF and k-nearest neighbors models, measurements from the subinguinal projection in the supine position were the most influential. Machine learning methods have demonstrated superior accuracy in predicting disease compared to traditional statistical regressions and nomograms. These advancements hold promise for clinically automated prediction of varicocoele grades in patients. Tailored varicocoele grading for individuals has the potential to enhance treatment effectiveness and overall quality of life.

Primary Hyperhidrosis and Sensitive Skin: Exploring the Link with Predictive Machine Learning-Based Classification Models.

Primary hyperhidrosis (PHH) is a disorder of excessive sweating caused by aberrant cholinergic signaling. Sensitive skin (SS) is a condition of subjective cutaneous hyperreactivity to innocuous stimuli, impacting 40% to 70% of the population. SS is exacerbated by sweat, stress, and heat, suggesting that cholinergic stimulation may contribute to SS flares. To survey PHH sufferers to assess hyperhidrosis (HH) and SS symptom burden. An International Review Board (IRB)-exempt survey was disseminated by the International Hyperhidrosis Society. A predictive classification model for SS was built using random forest machine learning algorithms. Of the 637 respondents with PHH, 89% reported SS; and there was a significant association between HH and SS severity scores. Importantly, SS occurred on body sites affected and unaffected by HH. Predictive modeling designated Sensitive Scale-10 (SS-10), a validated questionnaire to gauge SS severity, to be the most helpful in predicting SS in this cohort. Self-reported data. These data are the first to propose and support a relationship between SS and HH. SS occurred with greatest frequency at HH-afflicted body sites, but also occurred on unaffected sites, suggesting that sweat is not the sole causative link. Future work can explore cholinergic signaling as a potential link between these conditions. Screening HH patients for SS may be warranted. J Drugs Dermatol. 2024;23(10):882-888. doi:10.36849/JDD.8461.

A novel framework for concealed weapons detection using passive terahertz imaging

ABSTRACT In this work, a novel framework for non-destructive, non-intrusive and completely automatic concealed weapon detection under human clothing for entry control and security check applications using passive terahertz (THz) images has been given. The technique models the problem of concealed weapon detection as that of binary segmentation, where the pixels corresponding to concealed weapons are foreground while the remaining are background. The core idea of the proposed framework is to first oversegment the THz image into superpixels and subsequently multiple handcrafted features are extracted from each superpixel. Finally, a machine learning-based classifier is used to perform binary classification thereby classifying each superpixel as foreground or background such that each pixel in a foreground superpixel is a foreground pixel and vice-versa for background. It is worth mentioning that this technique does not require a perfect segmentation to extract features from, rather an oversegmentation suffices. Furthermore, both global features like those based on image saliency and local features like those based on intensity are extracted considering the image properties. The detailed experiments and comparative analysis confirm that the proposed technique efficiently detects concealed weapons, exhibiting superior performance compared to state-of-the-art binary segmentation methods for the purpose.

Machine Learning-Based Classification of Body Imbalance and Its Intensity Using Electromyogram and Ground Reaction Force in Immersive Environments

Machine Learning-Based Classification of Body Imbalance and Its Intensity Using Electromyogram and Ground Reaction Force in Immersive Environments

Machine learning-based classification of Parkinson's disease using acoustic features: Insights from multilingual speech tasks

This study advances the automation of Parkinson's disease (PD) diagnosis by analyzing speech characteristics, leveraging a comprehensive approach that integrates a voting-based machine learning model. Given the growing prevalence of PD, especially among the elderly population, continuous and efficient diagnosis is of paramount importance. Conventional monitoring methods suffer from limitations related to time, cost, and accessibility, underscoring the need for the development of automated diagnostic tools. In this paper, we present a robust model for classifying speech patterns in Korean PD patients, addressing a significant research gap. Our model employs straightforward preprocessing techniques and a voting-based machine learning approach, demonstrating superior performance, particularly when training data is limited. Furthermore, we emphasize the effectiveness of the eGeMAPSv2 feature set in PD analysis and introduce new features that substantially enhance classification accuracy. The proposed model, achieving an accuracy of 84.73 % and an area under the ROC (AUC) score of 92.18 % on a dataset comprising 100 Korean PD patients and 100 healthy controls, offers a practical solution for automated diagnosis applications, such as smartphone apps. Future research endeavors will concentrate on enhancing the model's performance and delving deeper into the relationship between high-importance features and PD.

A stacked machine learning-based classification model for endometriosis and adenomyosis: a retrospective cohort study utilizing peripheral blood and coagulation markers.

Endometriosis (EMs) and adenomyosis (AD) are common gynecological diseases that impact women's health, and they share symptoms such as dysmenorrhea, chronic pain, and infertility, which adversely affect women's quality of life. Current diagnostic approaches for EMs and AD involve invasive surgical procedures, and thus, methods of noninvasive differentiation between EMs and AD are needed. This retrospective cohort study introduces a novel, noninvasive classification methodology employing a stacked ensemble machine learning (ML) model that utilizes peripheral blood and coagulation markers to distinguish between EMs and AD. The study included a total of 558 patients (329 with EMs and 229 with AD), in whom key hematological and coagulation markers were analyzed to identify distinctive profiles. Feature selection was conducted through ML (logistic regression, support vector machine, and K-nearest neighbors) to determine significant hematological markers. Red cell distribution width, mean corpuscular hemoglobin concentration, activated partial thromboplastin time, international normalized ratio, and antithrombin III were proved to be the key distinguishing indexes for disease differentiation. Among all the ML classification models developed, the stacked ensemble model demonstrated superior performance (area under the curve = 0.803, 95% credibility interval = 0.701-0.904). Our findings demonstrate the effectiveness of the stacked ensemble ML model for classifying EMs and AD. Integrating biomarkers into this multi-algorithm framework offers a novel approach to noninvasive diagnosis. These results advocate for the application of stacked ensemble ML utilizing cost-effective and readily available peripheral blood and coagulation indicators for the early, rapid, and noninvasive differential diagnosis of EMs and AD, offering a potentially transformative approach for clinical decision-making and personalized treatment strategies.

AI drives the assessment of lung cancer microenvironment composition

PurposeThe abundance and distribution of tumor-infiltrating lymphocytes (TILs) as well as that of other components of the tumor microenvironment is of particular importance for predicting response to immunotherapy in lung cancer (LC). We describe here a pilot study employing artificial intelligence (AI) in the assessment of TILs and other cell populations, intending to reduce the inter- or intra-observer variability that commonly characterizes this evaluation. DesignWe developed a machine learning-based classifier to detect tumor, immune, and stromal cells on hematoxylin and eosin-stained sections, using the open-source framework QuPath. We evaluated the quantity of the aforementioned three cell populations among 37 LC whole slide images regions of interest, comparing the assessments made by five pathologists, both before and after using graphical predictions made by AI, for a total of 1110 quantitative measurements. ResultsOur findings indicate noteworthy variations in score distribution among pathologists and between individual pathologists and AI. The AI-guided pathologist's evaluations resulted in reduction of significant discrepancies across pathologists: three comparisons showed a loss of significance (p > 0.05),whereas other four showed a reduction in significance (p > 0.01). ConclusionsWe show that employing a machine learning approach in cell population quantification reduces inter- and intra-observer variability, improving reproducibility and facilitating its use in further validation studies.

Review on spiking neural network-based ECG classification methods for low-power environments.

This paper reviews arrhythmia classification studies using electrocardiogram (ECG) signals. Research on automatically diagnosing arrhythmia in daily life has been actively underway for early detection and treatment of heart disease. Development of automatic arrhythmia classification using ECG signal began based on handcrafted morphological feature extraction and machine learning-based classification methods. As deep neural networks (DNN) show excellent performance in the signal processing field, studies using various types of DNN are also being conducted in ECG classification. However, these DNN-based studies have extremely high computational complexity, making it challenging to perform real-time classification, and are unsuitable for low-power environments such as wearable devices due to high power consumption. Currently, research based on spiking neural network (SNN), which mimics the low-power operation of the human nervous system, is attracting attention as a method that can dramatically reduce complexity and power consumption. The classification accuracy of the SNN-based ECG classification studies is close to that of the DNN-based studies. When combined with neuromorphic hardware, it shows ultra-low-power performance, suggesting the possibility of use in lightweight devices. In this paper, the SNN-based ECG classification studies for low-power environments are mainly reviewed, and prior to this, conventional and DNN-based ECG classification studies are also reviewed. We hope that this review will be helpful to researchers and engineers interested in the field of ECG classification.

From Stability to Variability: Classification of Healthy Individuals, Prediabetes, and Type 2 Diabetes Using Glycemic Variability Indices from Continuous Glucose Monitoring Data.

Objective: This study aims to investigate the continuum of glucose control from normoglycemia to dysglycemia (HbA1c ≥ 5.7%/39 mmol/mol) using metrics derived from continuous glucose monitoring (CGM). In addition, we aim to develop a machine learning-based classification model to classify dysglycemia based on observed patterns. Methods: Data from five distinct studies, each featuring at least two days of CGM, were pooled. Participants included individuals classified as healthy, with prediabetes, or with type 2 diabetes mellitus (T2DM). Various CGM indices were extracted and compared across groups. The data set was split 70/30 for training and testing two classification models (XGBoost/Logistic Regression) to differentiate between prediabetes or dysglycemia and the healthy group. Results: The analysis included 836 participants (healthy: n = 282; prediabetes: n = 133; T2DM: n = 432). Across all CGM indices, a progressive shift was observed from the healthy group to those with diabetes (P < 0.001). Statistically significant differences (P < 0.01) were noted in mean glucose, time below range, time above 140 mg/dl, mobility, multiscale complexity index, and glycemic risk index when transitioning from health to prediabetes. The XGBoost models achieved the highest receiver operating characteristic area under the curve values on the test data set ranging from 0.91 [confidence interval (CI): 0.87-0.95] (prediabetes identification) to 0.97 [CI: 0.95-0.98] (dysglycemia identification). Conclusion: Our findings demonstrate a gradual deterioration of glucose homeostasis and increased glycemic variability across the spectrum from normo- to dysglycemia, as evidenced by CGM metrics. The performance of CGM-based indices in classifying healthy individuals and those with prediabetes and diabetes is promising.

EEG classification of resting state and arithmetic cognitive workload using functional connectivity of different frequency bands and machine learning techniques

ABSTRACT This study aimed to perform a comparative study of the functional connectivity of different frequency bands for the identification of resting and arithmetic cognitive workload EEG using machine learning techniques. Functional connectivity was calculated from preprocessed EEGs for both rest and task states in 5 EEG sub-bands: alpha (8–13 Hz), theta (4–8 Hz), delta (1–4 Hz), gamma (30–45 Hz), and beta (13–30 Hz). This was done through Weighted Phase Lag Index (WPLI). After that, PCA was applied to the calculated feature vectors to decrease the dimensionality of the feature space. Eventually, the normalized chosen features were used as input for different machine learning-based classification models, and the performance was assessed through the leave-one-subject cross-validation (LOSOCV) algorithm. Experimental results showed that the classification results on the basis of the connectivity features of delta, theta, alpha, beta, and gamma frequency bands were 90.27%, 77.78%, 62.50%, 62.50%, and 76.39%, respectively. The obtained results showed that used machine learning models and functional connectivity technique are successfully applied to detect mental workload from rest- and task-EEG. In summary, EEG functional connectivity in the delta frequency is a potent tool for comprehending the neural basis of mental workload and has significant applications in various fields.

Machine learning-based classification of quality grades for concrete vibration behaviour

The vibration behaviour of the vibrating rods is one of the key factors for the quality of concrete vibration that is essential for the long-term safety of concrete structures. Although many vibration operation regulations have been widely applied, evaluating method for concrete vibration samples is relatively rare. To fill this gap, this paper proposes a concrete vibration data acquisition system and attempts to perform a quality analysis of samples employing machine learning that amalgamates various parameters. The experimental results demonstrate that the data collection system to identify the vibration state achieves an accuracy of 93.75% and an accuracy of 90.3% in classifying vibration quality levels. The proposed method can classify and evaluate concrete quality levels, which strongly supports the visualization of concrete vibration process and the implementation of autonomous robot vibration. In the future, improving the reliability of the system and enhancing the accuracy of algorithms will be the focused.

A simple classification and clustering of poverty in rural areas using machine learning

This study aimed to determine the socio-economic poverty status of those living in rural areas using data surveys obtained from household expenditure and income. Machine learning-based classification and clustering models were proven to provide an overview of efforts to determine similarities in poverty characteristics. Efforts to address poverty classification and clustering typically involve comprehensive strategies that aim to improve socio-economic conditions in the affected areas. This research focuses on the combined application of machine learning classification and clustering techniques to analyze poverty. It aims to investigate whether the integration of classification and clustering algorithms can enhance the accuracy of poverty analysis by identifying distinct poverty classes or clusters based on multidimensional indicators. The results showed the superiority of machine learning in mapping poverty in rural areas; therefore, it can be adopted in the private sector and government domains. It is important to have access to relevant and reliable data to apply these machine learning techniques effectively. Data sources may include household surveys, census data, administrative records, satellite imagery, and other socioeconomic indicators. Machine learning classification and clustering analyses are used as a decision support tool to gain an understanding of poverty data from each village. These strategies are also used to describe the profile of poverty clusters in the community in terms of significant socio-economic indicators present in the data. Village clusters based on an analysis of existing poverty indicators are grouped into high, moderate, and low poverty levels. Machine learning can be a valuable tool for analyzing and understanding poverty by classifying individuals or households into different poverty categories and identifying patterns and clusters of poverty. These insights can inform targeted interventions, policy decisions, and resource allocation for poverty reduction programs.

Exploring emerald global geochemical provenance through fingerprinting and machine learning methods

Emeralds – the green colored variety of beryl – occur as gem-quality specimens in over fifty deposits globally. While digital traceability methods for emerald have limitations, sample-based approaches offer robust alternatives, particularly for determining the geographic origin of emerald. Three factors make emerald suitable for provenance studies and hence for developing models for origin determination. First, the diverse elemental chemistry of emerald at minor (<1 wt%) and trace levels (<1 to 100’s ppmw) exhibits unique inter-element fractionations between global deposits. Second, minimally destructive techniques, including laser ablation inductively coupled plasma mass spectrometry (LA-ICP-MS), enable measurement of these diagnostic elemental signatures. Third, when applied to extensive datasets, machine learning (ML) techniques enable the creation of predictive models and statistical discrimination with adequate characterization of the deposits. This study employs a carefully selected dataset comprising more than 1000 LA-ICP-MS analyses of gem-quality emeralds, enriched with new analyses. This dataset represents the largest available for global emerald deposits. We conducted unsupervised exploratory analysis using Principal Component Analysis (PCA). For machine learning-based classification, we employed Support Vector Machine Classification (SVM-C), achieving an initial accuracy rate of 79%. This was enhanced to 96.8% through the use of hierarchical SVM-C with PCA filters as our modeling approach. The ML models were trained using the concentrations of eight statistically significant elements (Li, V, Cr, Fe, Sc, Ga, Rb, Cs). By leveraging high-quality LA-ICP-MS data and ML techniques, accurate identification of the geographical origin of emerald becomes possible. These models are important for accurate provenance of emerald, and from a geochemical perspective, for understanding the formation environments of beryl-bearing pegmatites and shales.

Embedded FPGA developments in 130 nm and 28 nm CMOS for machine learning in particle detector readout

Embedded field programmable gate array (eFPGA) technology allows the implementation of reconfigurable logic within the design of an application-specific integrated circuit (ASIC). This approach offers the low power and efficiency of an ASIC along with the ease of FPGA configuration, particularly beneficial for the use case of machine learning in the data pipeline of next-generation collider experiments. An open-source framework called “FABulous” was used to design eFPGAs using 130 nm and 28 nm CMOS technology nodes, which were subsequently fabricated and verified through testing. The capability of an eFPGA to act as a front-end readout chip was assessed using simulation of high energy particles passing through a silicon pixel sensor. A machine learning-based classifier, designed for reduction of sensor data at the source, was synthesized and configured onto the eFPGA. A successful proof-of-concept was demonstrated through reproduction of the expected algorithm result on the eFPGA with perfect accuracy. Further development of the eFPGA technology and its application to collider detector readout is discussed.

Improving Task-Agnostic Energy Shaping Control of Powered Exoskeletons with Task/Gait Classification.

Emerging task-agnostic control methods offer a promising avenue for versatile assistance in powered exoskeletons without explicit task detection, but typically come with a performance trade-off for specific tasks and/or users. One such approach employs data-driven optimization of an energy shaping controller to provide naturalistic assistance across essential daily tasks with passivity/stability guarantees. This study introduces a novel control method that merges energy shaping with a machine learning-based classifier to deliver optimal support accommodating diverse individual tasks and users. The classifier detects transitions between multiple tasks and gait patterns in order to employ a more optimal, task-agnostic controller based on the weighted sum of multiple optimized energy-shaping controllers. To demonstrate the efficacy of this integrated control strategy, an in-silico assessment is conducted over a range of gait patterns and tasks, including incline walking, stairs ascent/descent, and stand-to-sit transitions. The proposed method surpasses benchmark approaches in 5-fold cross-validation ( ), yielding 93.17 ± 7.39% cosine similarity and 77.92 ± 19.76% variance-accounted-for across tasks and users. These findings highlight the control approach's adaptability in aligning with human joint moments across various tasks.