Multiclass Machine Learning Research Articles

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.

Machine learning and remote sensing-based lithological mapping of the Duwi Shear-Belt area, Central Eastern Desert, Egypt

Machine learning and remote sensing techniques are widely accepted as valuable, cost-effective tools in lithological discrimination and mineralogical investigations. The current study represents an attempt to use machine learning classification along with several remote sensing techniques being applied to Landsat-8/9 satellite data to discriminate the various outcropping lithological rock units at the Duwi Shear Belt (DSB) area in the Central Eastern Desert of Egypt. Multi-class machine learning classification, multiple conventional remote sensing mapping techniques, spectral separability analysis based on the Jeffries-Matusita (J-M) distance measure, fieldwork, and petrographic investigations were integrated to enhance the lithological discrimination of the exposed rock units at DSB area. The well-recognized machine learning classifier (Support Vector Machine—SVM) was adopted in this study, with training data determined carefully based on enhancing the lithological discrimination attained from various remote sensing techniques of False Color Composites (FCC), Principal Component Analysis (PCA), and Minimum Noise Fraction (MNF), along with the fieldwork data and the previously published geologic maps. High overall accuracy of the SVM classification was obtained, however, inspection of the individual rock unit classes’ accuracies revealed lower accuracy for certain types of rock units which were also found associated with lower separability scores as well. Among the least separable rock units were; metagabbro rocks that showed high spectral similarity with the volcaniclastic metasediments rocks, and the metaultramafics of the ophiolitic mélange showed spectral attitude of high correlation to that of the Hammamat volcanosedimentary rocks. Target-oriented Color Ratio Composites (CRC) technique was implemented to better discriminate these hardly separable rock units. A final integrated geological map was obtained comprising the various discriminated Neoproterozoic basement rock units of the DSB area. The successfully mapped litho-units include; Meatiq Group (amphibolites, gneissic granitoids, and mylonitized granitoids), ophiolitic mélange (metaultramafics, metagabbro-amphibolites, and volcaniclastic metasediments), Dokhan volcanics, Hammamat sediments, and granites. An adequate description of these rock units was also given in light of the conducted intense fieldwork and petrographic investigations.

Detection of Parkinson disease using multiclass machine learning approach

Parkinson’s Disease (PD) is a prevalent neurological condition characterized by motor and cognitive impairments, typically manifesting around the age of 50 and presenting symptoms such as gait difficulties and speech impairments. Although a cure remains elusive, symptom management through medication is possible. Timely detection is pivotal for effective disease management. In this study, we leverage Machine Learning (ML) and Deep Learning (DL) techniques, specifically K-Nearest Neighbor (KNN) and Feed-forward Neural Network (FNN) models, to differentiate between individuals with PD and healthy individuals based on voice signal characteristics. Our dataset, sourced from the University of California at Irvine (UCI), comprises 195 voice recordings collected from 31 patients. To optimize model performance, we employ various strategies including Synthetic Minority Over-sampling Technique (SMOTE) for addressing class imbalance, Feature Selection to identify the most relevant features, and hyperparameter tuning using RandomizedSearchCV. Our experimentation reveals that the FNN and KSVM models, trained on an 80–20 split of the dataset for training and testing respectively, yield the most promising results. The FNN model achieves an impressive overall accuracy of 99.11%, with 98.78% recall, 99.96% precision, and a 99.23% f1-score. Similarly, the KSVM model demonstrates strong performance with an overall accuracy of 95.89%, recall of 96.88%, precision of 98.71%, and an f1-score of 97.62%. Overall, our study showcases the efficacy of ML and DL techniques in accurately identifying PD from voice signals, underscoring the potential for these approaches to contribute significantly to early diagnosis and intervention strategies for Parkinson’s Disease.

Feasibility of Principal Component Analysis for Multi-Class Earthquake Prediction Machine Learning Model Utilizing Geomagnetic Field Data

Geomagnetic field data have been found to contain earthquake (EQ) precursory signals; however, analyzing this high-resolution, imbalanced data presents challenges when implementing machine learning (ML). This study explored feasibility of principal component analyses (PCA) for reducing the dimensionality of global geomagnetic field data to improve the accuracy of EQ predictive models. Multi-class ML models capable of predicting EQ intensity in terms of the Mercalli Intensity Scale were developed. Ensemble and Support Vector Machine (SVM) models, known for their robustness and capabilities in handling complex relationships, were trained, while a Synthetic Minority Oversampling Technique (SMOTE) was employed to address the imbalanced EQ data. Both models were trained on PCA-extracted features from the balanced dataset, resulting in reasonable model performance. The ensemble model outperformed the SVM model in various aspects, including accuracy (77.50% vs. 75.88%), specificity (96.79% vs. 96.55%), F1-score (77.05% vs. 76.16%), and Matthew Correlation Coefficient (73.88% vs. 73.11%). These findings suggest the potential of a PCA-based ML model for more reliable EQ prediction.

Harnessing Metabolites as Serum Biomarkers for Liver Graft Pathology Prediction Using Machine Learning.

Graft injury affects over 50% of liver transplant (LT) recipients, but non-invasive biomarkers to diagnose and guide treatment are currently limited. We aimed to develop a biomarker of graft injury by integrating serum metabolomic profiles with clinical variables. Serum from 55 LT recipients with biopsy confirmed metabolic dysfunction-associated steatohepatitis (MASH), T-cell mediated rejection (TCMR) and biliary complications was collected and processed using a combination of LC-MS/MS assay. The metabolomic profiles were integrated with clinical information using a multi-class Machine Learning (ML) classifier. The model's efficacy was assessed through the Out-of-Bag (OOB) error estimate evaluation. Our ML model yielded an overall accuracy of 79.66% with an OOB estimate of the error rate at 19.75%. The model exhibited a maximum ability to distinguish MASH, with an OOB error estimate of 7.4% compared to 22.2% for biliary and 29.6% for TCMR. The metabolites serine and serotonin emerged as the topmost predictors. When predicting binary outcomes using three models: Biliary (biliary vs. rest), MASH (MASH vs. rest) and TCMR (TCMR vs. rest); the AUCs were 0.882, 0.972 and 0.896, respectively. Our ML tool integrating serum metabolites with clinical variables shows promise as a non-invasive, multi-class serum biomarker of graft pathology.

Multiclass Sentiment Prediction of Airport Service Online Reviews Using Aspect-Based Sentimental Analysis and Machine Learning

Airport service quality ratings found on social media such as Airline Quality and Google Maps offer invaluable insights for airport management to improve their quality of services. However, there is currently a lack of research analysing these reviews by airport services using sentimental analysis approaches. This research applies multiclass models based on Aspect-Based Sentimental Analysis to conduct a comprehensive analysis of travellers’ reviews, in which the major airport services are tagged by positive, negative, and non-existent sentiments. Seven airport services commonly utilised in previous studies are also introduced. Subsequently, various Deep Learning architectures and Machine Learning classification algorithms are developed, tested, and compared using data collected from Twitter, Google Maps, and Airline Quality, encompassing travellers’ feedback on airport service quality. The results show that the traditional Machine Learning algorithms such as the Random Forest algorithm outperform Deep Learning models in the multiclass prediction of airport service quality using travellers’ feedback. The findings of this study offer concrete justifications for utilising multiclass Machine Learning models to understand the travellers’ sentiments and therefore identify airport services required for improvement.

RAINBOW: A colorful approach to multipassband light-curve estimation

Context. Time series generated by repeatedly observing astronomical transients are generally sparse, irregularly sampled, noisy, and multidimensional (obtained through a set of broad-band filters). In order to fully exploit their scientific potential, it is necessary to use this incomplete information to estimate a continuous light-curve behavior. Traditional approaches use ad hoc functional forms to approximate the light curve in each filter independently (hereafter, the MONOCHROMATIC method). Aims. We present RAINBOW, a physically motivated framework that enables simultaneous multiband light-curve fitting. It allows the user to construct a 2D continuous surface across wavelength and time, even when the number of observations in each filter is significantly limited. Methods. Assuming the electromagnetic radiation emission from the transient can be approximated by a blackbody, we combined an expected temperature evolution and a parametric function describing its bolometric light curve. These three ingredients allow the information available in one passband to guide the reconstruction in the others, thus enabling a proper use of multisurvey data. We demonstrate the effectiveness of our method by applying it to simulated data from the Photometric LSST Astronomical Time-series Classification Challenge (PLAsTiCC) as well as to real data from the Young Supernova Experiment (YSE DR1). Results. We evaluate the quality of the estimated light curves according to three different tests: goodness of fit, peak-time prediction, and ability to transfer information to machine-learning (ML) based classifiers. The results confirm that RAINBOW leads to an equivalent goodness of fit (supernovae II) or to a goodness of fit that is better by up to 75% (supernovae Ibc) than the MONOCHROMATIC approach. Similarly, the accuracy improves for all classes in our sample when the RAINBOW best-fit values are used as a parameter space in a multiclass ML classification. Conclusions. Our approach enables a straightforward light-curve estimation for objects with observations in multiple filters and from multiple experiments. It is particularly well suited when the light-curve sampling is sparse. We demonstrate its potential for characterizing supernova-like events here, but the same approach can be used for other classes by changing the function describing the light-curve behavior and temperature representation. In the context of the upcoming large-scale sky surveys and their potential for multisurvey analysis, this represents an important milestone in the path to enable population studies of photometric transients.

Novel multiclass classification machine learning approach for the early-stage classification of systemic autoimmune rheumatic diseases

ObjectiveSystemic autoimmune rheumatic diseases (SARDs) encompass a diverse group of complex conditions with overlapping clinical features, making accurate diagnosis challenging. This study aims to develop a multiclass machine learning (ML)...

Multi-class machine learning classification of PFAS in environmental water samples: a blinded test of performance on unknowns

A multi-class method was developed to identify PFAS origin based on chemical composition, and performance of the method was evaluated in a blinded test against unknowns. The method showed great promise in its ability to recognize sample origin.

Time-Frequency Image-based Speech Emotion Recognition using Artificial Neural Network

Automatic Speech Emotion Recognition (ASER) is a state-of-the-art application in artificial intelligence. Speech recognition intelligence is employed in various applications such as digital assistance, security, and other human-machine interactive products. In the present work, three open-source acoustic datasets, namely SAVEE, RAVDESS, and EmoDB, have been utilized (Haq et al., 2008, Livingstone et al., 2005, Burkhardt et al., 2005). From these datasets, six emotions namely anger, disgust, fear, happy, neutral, and sad, are selected for automatic speech emotion recognition. Various types of algorithms are already reported for extracting emotional content from acoustic signals. This work proposes a time-frequency (t-f) image-based multiclass speech emotion classification model for the six emotions mentioned above. The proposed model extracts 472 grayscale image features from the t-f images of speech signals. The t-f image is a visual representation of the time component and frequency component at that time in the two-dimensional space, and differing colors show its amplitude. An artificial neural network-based multiclass machine learning approach is used to classify selected emotions. The experimental results show that the above-mentioned emotions' average classification accuracy (CA) of 88.6%, 85.5%, and 93.56% is achieved using SAVEE, RAVDESS, and EmoDB datasets, respectively. Also, an average CA of 83.44% has been achieved for the combination of all three datasets. The maximum reported average classification accuracy (CA) using spectrogram for SAVEE, RAVDESS, and EmoDB dataset is 87.8%, 79.5 %, and 83.4%, respectively (Wani et al., 2020, Mustaqeem and Kwon, 2019, Badshah et al., 2017). The proposed t-f image-based classification model shows improvement in average CA by 0.91%, 7.54%, and 12.18 % for SAVEE, RAVDESS, and EmoDB datasets, respectively. This study can be helpful in human-computer interface applications to detect emotions precisely from acoustic signals.

Evaluation of UN SDG-related formal learning activities in a university common core curriculum

Purpose Higher education plays an essential role in achieving the United Nations sustainable development goals (SDGs). However, there are only scattered studies on monitoring how universities promote SDGs through their curriculum. The purpose of this study is to investigate the connection of existing common core courses in a university to SDG education. In particular, this study wanted to know how common core courses can be classified by machine-learning approach according to SDGs. Design/methodology/approach In this report, the authors used machine learning techniques to tag the 166 common core courses in a university with SDGs and then analyzed the results based on visualizations. The training data set comes from the OSDG public community data set which the community had verified. Meanwhile, key descriptions of common core courses had been used for the classification. The study used the multinomial logistic regression algorithm for the classification. Descriptive analysis at course-level, theme-level and curriculum-level had been included to illustrate the proposed approach’s functions. Findings The results indicate that the machine-learning classification approach can significantly accelerate the SDG classification of courses. However, currently, it cannot replace human classification due to the complexity of the problem and the lack of relevant training data. Research limitations/implications The study can achieve a more accurate model training through adopting advanced machine learning algorithms (e.g. deep learning, multioutput multiclass machine learning algorithms); developing a more effective test data set by extracting more relevant information from syllabus and learning materials; expanding the training data set of SDGs that currently have insufficient records (e.g. SDG 12); and replacing the existing training data set from OSDG by authentic education-related documents (such as course syllabus) with SDG classifications. The performance of the algorithm should also be compared to other computer-based and human-based SDG classification approaches for cross-checking the results, with a systematic evaluation framework. Furthermore, the study can be analyzed by circulating results to students and understanding how they would interpret and use the results for choosing courses for studying. Furthermore, the study mainly focused on the classification of topics that are taught in courses but cannot measure the effectiveness of adopted pedagogies, assessment strategies and competency development strategies in courses. The study can also conduct analysis based on assessment tasks and rubrics of courses to see whether the assessment tasks can help students understand and take action on SDGs. Originality/value The proposed approach explores the possibility of using machine learning for SDG classifications in scale.

Advancing Federated Learning through Verifiable Computations and Homomorphic Encryption.

Federated learning, as one of the three main technical routes for privacy computing, has been widely studied and applied in both academia and industry. However, malicious nodes may tamper with the algorithm execution process or submit false learning results, which directly affects the performance of federated learning. In addition, learning nodes can easily obtain the global model. In practical applications, we would like to obtain the federated learning results only by the demand side. Unfortunately, no discussion on protecting the privacy of the global model is found in the existing research. As emerging cryptographic tools, the zero-knowledge virtual machine (ZKVM) and homomorphic encryption provide new ideas for the design of federated learning frameworks. We have introduced ZKVM for the first time, creating learning nodes as local computing provers. This provides execution integrity proofs for multi-class machine learning algorithms. Meanwhile, we discuss how to generate verifiable proofs for large-scale machine learning tasks under resource constraints. In addition, we implement the fully homomorphic encryption (FHE) scheme in ZKVM. We encrypt the model weights so that the federated learning nodes always collaborate in the ciphertext space. The real results can be obtained only after the demand side decrypts them using the private key. The innovativeness of this paper is demonstrated in the following aspects: 1. We introduce the ZKVM for the first time, which achieves zero-knowledge proofs (ZKP) for machine learning tasks with multiple classes and arbitrary scales. 2. We encrypt the global model, which protects the model privacy during local computation and transmission. 3. We propose and implement a new federated learning framework. We measure the verification costs under different federated learning rounds on the IRIS dataset. Despite the impact of homomorphic encryption on computational accuracy, the framework proposed in this paper achieves a satisfactory 90% model accuracy. Our framework is highly secure and is expected to further improve the overall efficiency as cryptographic tools continue to evolve.

Prediction of clinical outcomes after percutaneous coronary intervention: Machine-learning analysis of the National Inpatient Sample

BackgroundThis study aimed to develop a multiclass machine-learning (ML) model to predict all-cause mortality, ischemic and hemorrhagic events in unselected hospitalized patients undergoing percutaneous coronary intervention (PCI). MethodsThis retrospective study included 1,815,595 unselected weighted hospitalizations undergoing PCI from the National Inpatient Sample (2016–2019). Five most common ML algorithms (logistic regression, support vector machine (SVM), naive Bayes, random forest (RF), and extreme gradient boosting (XGBoost)) were trained and tested with 101 input features. The study endpoints were different combinations of all-cause mortality, ischemic cerebrovascular events (CVE) and major bleeding. An area under the curve (AUC) with 95% confidence interval (95% CI) was selected as a performance metric. ResultsThe study population was split to a training cohort of 1,186,880 PCI discharges, validation cohort (for calibration) of 296,725 hospitalizations and a test cohort of 331,990 PCI discharges. A total of 98,180 (5.4%) hospital entries included study outcomes. Logistic regression, SVM, naive Bayes, and RF model demonstrated AUCs of 0.83 (95% CI 0.82–0.84), 0.84 (95% CI 0.83–0.86), 0.81 (95% CI 0.80–0.82), and 0.83 (95% CI 0.81–0.84), retrospectively. The XGBoost classifier performed the best with an AUC of 0.86 (95% CI 0.85–0.87) with excellent calibration. We then built a web-based application that provides predictions based on the XGBoost model. ConclusionWe derived the multi-task XGBoost classifier based on 101 features to predict different combinations of all-cause death, ischemic CVE and major bleeding. Such models may be useful in benchmarking and risk prediction using routinely collected administrative data.

Impact analysis and optimization of notched high-gain Antipodal Vivaldi antenna using machine learning

Optimizing antenna parameters without knowing their impact on the gain is a time-consuming and iterative process. This article shows how the Classification and Regression Tree (CART) technique, the Random Forest (RF) algorithm, and the Principal Component Analysis (PCA) method is used to study the effect of geometrical parameters of rectangular notches of the Antipodal Vivaldi antenna on gain. Carving out the notches is a general approach for end-fire antenna design. Integrating the EM simulation with machine learning could reduce the time spent on AVA design. In this approach, the gain category of the antenna under design: Low, Medium, and High, is first classified by the three multiclass machine learning models, K-Nearest Neighbor (KNN), Artificial Neural Network (ANN), and Support Vector Machine (SVM). When the predicted category of gain belongs to the High class, in that case, the four regression models, Polynomial, SVM, ANN, and RF, predict the gain value to anticipate the antenna gain according to the notch dimensions to reduce the optimization time of the simulation-driven antenna. In this work, the band-limited approach of notches on gain is also been validated using simulation before fabrication. Machine learning models are optimized the antenna parameters to achieve a high gain within the frequency range of 20 to 40 GHz. A peak simulated gain of 19 dB is achieved at the frequency of 27 GHz. The proposed structure is fabricated utilizing a single-layer Rogers substrate (RO4003C) with a dielectric constant of 3.38 and compact dimensions of 96 mm × 174 mm × 0.508 mm, i.e., 0.32 × 0.58 × 0.0017 λ03; here the symbol λ0 denotes the wavelength of free space at the minimum operating frequency.

Reduced bacterial resistance antibiotics with improved microbiota tolerance in human intestinal: Molecular design and mechanism analysis

Antibiotic selectivity and bacterial resistance are critical global public health issues. We constructed a multi-class machine learning model to study antibiotic effects on human intestinal microbiota abundance and identified key features. Binding energies of β-lactam antibiotics with Escherichia coli PBP3 mutant protein were calculated, and a 2D-QSAR model for bacterial resistance was established. Sensitivity analysis identified key features affecting bacterial resistance. By coupling key features from the machine learning model and 2D-QSAR model, we designed ten flucloxacillin (FLU) substitutes that improved intestinal microbiota tolerance and reduced antibiotic bacterial resistance. Concurrently, the substitutes exhibited superior degradability in soil, aquatic environments, and under photolytic conditions, coupled with a reduced environmental toxicity compared to the FLU. Evaluations under combined medication revealed significant improvements in functionality and bacterial resistance for 80% of FLU substitutes, with 50% showing more than a twofold increase. Mechanistic analysis demonstrated enhanced binding to target proteins and increased biodegradability for FLU substitutes due to more concentrated surface charges. Reduced solvent hindrance and increased cell membrane permeability of FLU substitutes, mainly due to enhanced interactions with phospholipid bilayers, contributed to their functional selectivity. This study aims to address poor antibiotic selectivity and strong bacterial resistance, providing guidance for designing antibiotic substitutes.

Pharmacophore-Based Machine Learning Model To Predict Ligand Selectivity for E3 Ligase Binders.

E3 ligases are enzymes that play a critical role in ubiquitin-mediated protein degradation and are involved in various cellular processes. Pharmacophore analysis is a useful approach for predicting E3 ligase binding selectivity, which involves identifying key chemical features necessary for a ligand to interact with a specific protein target cavity. While pharmacophore analysis is not always sufficient to accurately predict ligand binding affinity, it can be a valuable tool for filtering and/or designing focused libraries for screening campaigns. In this study, we present a fast and an inexpensive approach using a pharmacophore fingerprinting scheme known as ErG, which is used in a multi-class machine learning classification model. This model can assign the correct E3 ligase binder to its known E3 ligase and predict the probability of each molecule to bind to different E3 ligases. Practical applications of this approach are demonstrated on commercial libraries such as Asinex for the rational design of E3 ligase binders. The scripts and data associated with this study can be found on GitHub at https://github.com/Fraunhofer-ITMP/E3_binder_Model.

Prediction of Chronic Stress and Protective Factors in Adults: Development of an Interpretable Prediction Model Based on XGBoost and SHAP Using National Cross-sectional DEGS1 Data.

Chronic stress is highly prevalent in the German population. It has known adverse effects on mental health, such as burnout and depression. Known long-term effects of chronic stress are cardiovascular disease, diabetes, and cancer. This study aims to derive an interpretable multiclass machine learning model for predicting chronic stress levels and factors protecting against chronic stress based on representative nationwide data from the German Health Interview and Examination Survey for Adults, which is part of the national health monitoring program. A data set from the German Health Interview and Examination Survey for Adults study including demographic, clinical, and laboratory data from 5801 participants was analyzed. A multiclass eXtreme Gradient Boosting (XGBoost) model was constructed to classify participants into 3 categories including low, middle, and high chronic stress levels. The model's performance was evaluated using the area under the receiver operating characteristic curve, precision, recall, specificity, and the F1-score. Additionally, SHapley Additive exPlanations was used to interpret the prediction XGBoost model and to identify factors protecting against chronic stress. The multiclass XGBoost model exhibited the macroaverage scores, with an area under the receiver operating characteristic curve of 81%, precision of 63%, recall of 52%, specificity of 78%, and F1-score of 54%. The most important features for low-level chronic stress were male gender, very good general health, high satisfaction with living space, and strong social support. This study presents a multiclass interpretable prediction model for chronic stress in adults in Germany. The explainable artificial intelligence technique SHapley Additive exPlanations identified relevant protective factors for chronic stress, which need to be considered when developing interventions to reduce chronic stress.

Development of a Predictive Model for Persistent Dizziness Following Vestibular Schwannoma Surgery.

In an era of vestibular schwannoma (VS) surgery where functional preservation is increasingly emphasized, persistent postoperative dizziness is a relatively understudied functional outcome. The primary objective was to develop a predictive model to identify patients at risk for developing persistent postoperative dizziness after VS resection. Retrospective review of patients who underwent VS surgery at our institution with a minimum of 12 months of postoperative follow-up. Demographic, tumor-specific, preoperative, and immediate postoperative features were collected as predictors. The primary outcome was self-reported dizziness at 3-, 6-, and 12-month follow-up. Binary and multiclass machine learning classification models were developed using these features. A total of 1,137 cases were used for modeling. The median age was 67 years, and 54% were female. Median tumor size was 2 cm, and the most common approach was suboccipital (85%). Overall, 63% of patients did not report postoperative dizziness at any timepoint; 11% at 3-month follow-up; 9% at 6-months; and 17% at 12-months. Both binary and multiclass models achieved high performance with AUCs of 0.89 and 0.86 respectively. Features important to model predictions were preoperative headache, need for physical therapy on discharge, vitamin D deficiency, and systemic comorbidities. We demonstrate the feasibility of a machine learning approach to predict persistent dizziness following vestibular schwannoma surgery with high accuracy. These models could be used to provide quantitative estimates of risk, helping counsel patients on what to expect after surgery and manage patients proactively in the postoperative setting. 4 Laryngoscope, 133:3534-3539, 2023.

Detecting suicidality on social media: Machine learning at rescue

The rise in technological advancements and Social Networking Sites (SNS) made people more engaged in their virtual lives. Research has revealed that people feel more comfortable posting their feelings, including suicidal thoughts, on SNS than discussing them through face-to-face settings due to the social stigma associated with mental health. This research study aims to develop a multi-class machine learning classifier for identifying suicidal risk levels in social media posts. The proposed Enhanced Feature Engineering Approach for Suicidal Risk Identification (EFASRI) is used to extract features from a novel dataset collected from Twitter and Reddit platforms. Three machine learning algorithms, i.e. Support Vector Machine (SVM), Random Forest (RF) and Extreme Gradient Boosting (XGB) were employed for classification. The study demonstrates significant improvements in the precision, recall, and overall accuracy compared to previous research that used classical feature extraction mechanisms. The best-performing algorithm, Extreme Gradient Boosting (XGB), achieved an overall accuracy of 96.33%. The findings imply that different features contain different levels of information, and the right combination of the features supplied to the machine learning algorithms may improve the prediction results.

Lithium‐bearing Pegmatite Exploration in Western Altun, Xinjiang, using Remote‐Sensing Technology

AbstractWestern Altun in Xinjiang is an important area, where lithium (Li)‐bearing pegmatites have been found in recent years. However, the complex terrain and harsh environment of western Altun exacerbates in prospecting for Li‐bearing pegmatites. Therefore, remote‐sensing techniques can be an effective means for prospecting Li‐bearing pegmatites. In this study, the fault information and lithologyical information in the region were obtained using the median‐resolution remote‐sensing image Landsat‐8, the radar image Sentinel‐1 and hyperspectral data GF‐5. Using Landsat‐8 data, the hydroxyl alteration information closely related to pegmatite in the region was extracted by principal component analysis, pseudo‐anomaly processing and other methods. The high spatial resolution remote‐sensing data WorldView‐2 and WorldView‐3 short‐wave infrared images were used and analyzed by principal component analysis (PCA), the band ratio method and multi‐class machine learning (ML), combined with conventional thresholds specified the algorithms used to automatically extract Li‐bearing pegmatite information. Finally, the Li‐bearing pegmatite exploration area was determined, based on a comprehensive analysis of the faults, hydroxyl alteration lithology and Li‐bearing pegmatite information. Field investigations have verified that the distribution of pegmatites in the central part of the study area is consistent with that of Li‐bearing pegmatites extracted in this study. This study provides a new technique for prospecting Li‐bearing pegmatites, which shows that remote‐sensing technology possesses great potential for identifying lithium‐bearing pegmatites, especially in areas that are not readily accessible.