Feature Machine Learning Research Articles

Comparison of machine learning algorithms and feature extraction techniques for the automatic detection of surface EMG activation timing

This paper presents a methodology for automatically detecting muscular activity by denoising, extracting features, and classifying surface electromyography (sEMG) signals. The proposed methodology utilizes the Discrete Wavelet Transform (DWT) and Willison’s Amplitude Algorithm (WAMP) for feature extraction. Five classification methods, including Neural Networks (NN), Classification Vector, XGBoost, Light Gradient Boosting Machine (LGBM), and ExtraTree, were evaluated using F-Measure, Precision, and Recall as performance metrics. Through k-fold cross-validation, the XGBoost algorithm, when combined with the Eigen values feature, achieved the highest training performance with an F1-Score of 98.71 %. For the test group, the LGBM classifier using WAMP, and NN with both WAMP and Eigen values as features, demonstrated the best average performance with F1-Scores of 96.52 ± 3.45 % and 96.52 ± 3.07 %, respectively. These results highlight the precision and performance of the proposed approach in detecting EMG signals. Moreover, the framework has the potential to support clinicians in diagnosing neuromuscular disorders and developing human–machine interfaces.

Flattening the curve-How to get better results with small deep-mutational-scanning datasets.

Proteins are used in various biotechnological applications, often requiring the optimization of protein properties by introducing specific amino-acid exchanges. Deep mutational scanning (DMS) is an effective high-throughput method for evaluating the effects of these exchanges on protein function. DMS data can then inform the training of a neural network to predict the impact of mutations. Most approaches use some representation of the protein sequence for training and prediction. As proteins are characterized by complex structures and intricate residue interaction networks, directly providing structural information as input reduces the need to learn these features from the data. We introduce a method for encoding protein structures as stacked 2D contact maps, which capture residue interactions, their evolutionary conservation, and mutation-induced interaction changes. Furthermore, we explored techniques to augment neural network training performance on smaller DMS datasets. To validate our approach, we trained three neural network architectures originally used for image analysis on three DMS datasets, and we compared their performances with networks trained solely on protein sequences. The results confirm the effectiveness of the protein structure encoding in machine learning efforts on DMS data. Using structural representations as direct input to the networks, along with data augmentation and pretraining, significantly reduced demands on training data size and improved prediction performance, especially on smaller datasets, while performance on large datasets was on par with state-of-the-art sequence convolutional neural networks. The methods presented here have the potential to provide the same workflow as DMS without the experimental and financial burden of testing thousands of mutants. Additionally, we present an open-source, user-friendly software tool to make these data analysis techniques accessible, particularly to biotechnology and protein engineering researchers who wish to apply them to their mutagenesis data.

Using protein language models for protein interaction hot spot prediction with limited data.

Protein language models, inspired by the success of large language models in deciphering human language, have emerged as powerful tools for unraveling the intricate code of life inscribed within protein sequences. They have gained significant attention for their promising applications across various areas, including the sequence-based prediction of secondary and tertiary protein structure, the discovery of new functional protein sequences/folds, and the assessment of mutational impact on protein fitness. However, their utility in learning to predict protein residue properties based on scant datasets, such as protein-protein interaction (PPI)-hotspots whose mutations significantly impair PPIs, remained unclear. Here, we explore the feasibility of using protein language-learned representations as features for machine learning to predict PPI-hotspots using a dataset containing 414 experimentally confirmed PPI-hotspots and 504 PPI-nonhot spots. Our findings showcase the capacity of unsupervised learning with protein language models in capturing critical functional attributes of protein residues derived from the evolutionary information encoded within amino acid sequences. We show that methods relying on protein language models can compete with methods employing sequence and structure-based features to predict PPI-hotspots from the free protein structure. We observed an optimal number of features for model precision, suggesting a balance between information and overfitting. This study underscores the potential of transformer-based protein language models to extract critical knowledge from sparse datasets, exemplified here by the challenging realm of predicting PPI-hotspots. These models offer a cost-effective and time-efficient alternative to traditional experimental methods for predicting certain residue properties. However, the challenge of explaining why specific features are important for determining certain residue properties remains.

Refined landslide susceptibility mapping in township area using ensemble machine learning method under dataset replenishment strategy

Landslide susceptibility mapping (LSM) warrants considerable attention as a prerequisite for risk assessment and prevention in response to the increasing global occurrences of landslides. However, refined LSM at the township scale still faces the problem of the limited number of landslides and the oversight of relevant factors associated with different landslide types. This work is dedicated to propose a novel dataset replenishment strategy designed to address landslide sample scarcity, and to identify dominant factors in high landslide susceptibility zones to refine LSM at township scale for decision makers. Herein, the Wuling County in the Three Gorges Reservoir area of China was taken as taken as the research case. Initially, field investigation and calibration of 21 landslides yielded accumulation and rock areas, and the corresponding impact factors were therefore proposed to construct input features for machine learning facilitated by factor diagnosis using Pearson correlation coefficient (PCC) and multicollinearity test. Subsequently, a training and validation dataset split of 8: 2 was randomly generated, enabling the application of ensemble machine learning model for LSM prediction. The study culminated in the feasibility of dataset replenishment strategy, which was employed to explore uncertainties and identify governing factors in LSM. The results indicate that the bagging-ANN model exhibits superior performance in both accumulation and rock areas, with Areas Under Curve (AUC) of 0.987 and 0.994. Furthermore, the dataset replenishment strategy emerges as instrumental in enhancing the reliability of LSM, as evidenced by an 8.97 % reduction in average value (AVG) and a 11.00 % increase in standard deviation (STD). The governing factors for landslides in accumulation and rock areas are designated as river erosion and height, respectively. Together with the excellent performance, this study is expected to provide a promising reference for LSM in other townships worldwide.

Trend-attribute forecasting of hourly PM2.5 trends in fifteen cities of Central England applying optimized machine learning feature selection

Recorded particulate matter (PM2.5) hourly trends are compared for fifteen urban recording sites distributed across central England for the period 2018 to 2022. They include 10 urban-background and five urban-traffic (roadside) sites with some located within the same urban area. The sites all show consistent background and peak distributions with mean annual values and standard deviations higher for 2018 and 2019 than for 2020 to 2022. The objective of this study is to demonstrate that trend attributes extracted from hourly recorded univariate PM2.5 trends at these sites can be used to provide reliable short-term hourly predictions and provide valuable insight into the regional variations in the recorded trends. Fifteen trend attributes extracted from the prior 12 h (t-1 to t-12) of recorded PM2.5 data were compiled and used as input to four supervised machine learning models (SML) to forecast PM2.5 concentrations up to 13 h ahead (t0 to t+12). All recording sites delivered forecasts with similar ranges of error levels for specific hours ahead which are consistent with their PM2.5 recorded ranges. Forecasting results for four representative sites are presented in detail using models trained and cross-validated with 2020 and 2021 hourly data to forecast 2021 and 2022 hourly data, respectively. A novel optimized feature selection procedure using a suite of five optimizers is used to improve the efficiency of the forecasting models. The LASSO and support vector regression models generate the best and most generalizable hourly PM2.5 forecasts from trained and validated SML models with mean average error (MAE) of between ∼1 and ∼3 μg/m3 for t0 to t+3 h ahead. A novel overfitting indicator, exploiting the cross-validation mean values, demonstrates that these two models are not affected by overfitting. Forecasts for t+6 to t+12 h forward generate higher MAE values between ∼3 and ∼4 μg/m3 due to their tendency to underestimate some of the extreme PM2.5 peaks. These findings indicate that further model refinements are required to generate more reliable short-term predictions for the t+6 to t+24 h ahead.

Total effects with constrained features

Recent studies have emphasized the connection between machine learning feature importance measures and total order sensitivity indices (total effects, henceforth). Feature correlations and the need to avoid unrestricted permutations make the estimation of these indices challenging. Additionally, there is no established theory or approach for non-Cartesian domains. We propose four alternative strategies for computing total effects that account for both dependent and constrained features. Our first approach involves a generalized winding stairs design combined with the Knothe-Rosenblatt transformation. This approach, while applicable to a wide family of input dependencies, becomes impractical when inputs are physically constrained. Our second approach is a U-statistic that combines the Jansen estimator with a weighting factor. The U-statistic framework allows the derivation of a central limit theorem for this estimator. However, this design is computationally intensive. Then, our third approach uses derangements to significantly reduce computational burden. We prove consistency and central limit theorems for these estimators as well. Our fourth approach is based on a nearest-neighbour intuition and it further reduces computational burden. We test these estimators through a series of increasingly complex computational experiments with features constrained on compact and connected domains (circle, simplex), non-compact and non-connected domains (Sierpinski gaskets), we provide comparisons with machine learning approaches and conclude with an application to a realistic simulator.

Enhancing Classification Performance through FeatureBoostThyro: A Comparative Study of Machine Learning Algorithms and Feature Selection

Early-stage prediction of a disease is an important and challenging task. The application of machine learning techniques is playing an important role in this era. Thyroid is one of the chronic endocrine diseases, and approximately 42 million people in India are affected by this disease. This paper presents a comprehensive investigation into the enhancement of classification performance through the novel ‘FeatureBoostThyro’ (FBT) model. The study evaluates various machine learning algorithms, including stochastic gradient descent (SGD), K nearest neighbor (KNN), logistic regression (LR), naive bayes (NB), and support vector machine (SVM), in conjunction with diverse feature selection methods. The research systematically explores the impact of feature selection techniques such as information gain, relief F, chi-square, gini index, forward selection, backward selection, recursive feature elimination, and LASSO on model performance across the chosen algorithms. The analysis reveals notable variations in performance metrics, including accuracy, precision, recall, and F1-score, providing valuable insights into the interplay between algorithm and feature selection. One main contribution of this research is the introduction of the FBT model, which consistently outperforms other models across various feature selection methods, making it a promising tool for addressing complex classification tasks. The findings contribute to a broader understanding of model selection and optimization in machine learning applications. The proposed model undergoes evaluation using two distinct datasets: the primary dataset acquired from Lata Mangeshkar Hospital in Nagpur and the secondary dataset obtained from the UCI dataset.

Development and validation of a machine learning-based predictive model for assessing the 90-day prognostic outcome of patients with spontaneous intracerebral hemorrhage

BackgroundSpontaneous intracerebral hemorrhage (sICH) is associated with significant mortality and morbidity. Predicting the prognosis of patients with sICH remains an important issue, which significantly affects treatment decisions. Utilizing readily available clinical parameters to anticipate the unfavorable prognosis of sICH patients holds notable clinical significance. This study employs five machine learning algorithms to establish a practical platform for the prediction of short-term prognostic outcomes in individuals afflicted with sICH.MethodsWithin the framework of this retrospective analysis, the model underwent training utilizing data gleaned from 413 cases from the training center, with subsequent validation employing data from external validation center. Comprehensive clinical information, laboratory analysis results, and imaging features pertaining to sICH patients were harnessed as training features for machine learning. We developed and validated the model efficacy using all the selected features of the patients using five models: Support Vector Machine (SVM), Logistic Regression (LR), Random Forest (RF), XGboost and LightGBM, respectively. The process of Recursive Feature Elimination (RFE) was executed for optimal feature screening. An internal five-fold cross-validation was employed to pinpoint the most suitable hyperparameters for the model, while an external five-fold cross-validation was implemented to discern the machine learning model demonstrating the superior average performance. Finally, the machine learning model with the best average performance is selected as our final model while using it for external validation. Evaluation of the machine learning model’s performance was comprehensively conducted through the utilization of the ROC curve, accuracy, and other relevant indicators. The SHAP diagram was utilized to elucidate the variable importance within the model, culminating in the amalgamation of the above metrics to discern the most succinct features and establish a practical prognostic prediction platform.ResultsA total of 413 patients with sICH patients were collected in the training center, of which 180 were patients with poor prognosis. A total of 74 patients with sICH were collected in the external validation center, of which 26 were patients with poor prognosis. Within the training set, the test set AUC values for SVM, LR, RF, XGBoost, and LightGBM models were recorded as 0.87, 0.896, 0.916, 0.885, and 0.912, respectively. The best average performance of the machine learning models in the training set was the RF model (average AUC: 0.906 ± 0.029, P < 0.01). The model still maintains a good performance in the external validation center, with an AUC of 0.817 (95% CI 0.705–0.928). Pertaining to feature importance for short-term prognostic attributes of sICH patients, the NIHSS score reigned supreme, succeeded by AST, Age, white blood cell, and hematoma volume, among others. In culmination, guided by the RF model’s variable importance weight and the model's ROC curve insights, the NIHSS score, AST, Age, white blood cell, and hematoma volume were integrated to forge a short-term prognostic prediction platform tailored for sICH patients.ConclusionWe constructed a prediction model based on the results of the RF model incorporating five clinically accessible predictors with reliable predictive efficacy for the short-term prognosis of sICH patients. Meanwhile, the performance of the external validation set was also more stable, which can be used for accurate prediction of short-term prognosis of sICH patients.

Speakers—As a sensor for detecting acoustic loads with Artificial Intelligent (AI)

The potential of using speakers as a sensor to detect ear canal conditions was demonstrated previously. This research contains our ongoing and continuous effort to utilize a single speaker as a sensor by measuring electrical impedance varying acoustic loads. Electrical impedance data (magnitude and phase) from six different acoustic load conditions were collected as features for machine learning (ML) model training. To enhance the learning performance, the data were pre-processed and augmented with normalization and level-shifting techniques, respectively. The raw data were converted to images to optimize the learning performance to classify acoustic loads from the impedance measurement. Several forms of images were experimented such as magnitude only, overlapped magnitude and phase, and rectangular form. A total of 2100 data (350 each) were used with CNN-based State of the Art (SOTA) models such as AlexNet, ResNet, and DenseNet. Both binary and multiclass classifications were performed, showing 0.9716 average accuracy and 0.907 accuracy, respectively. This innovative single-speaker approach using impedance as ML features is poised to revolutionize traditional acoustic sensing research by harnessing the limitless power of AI.

Fault Detection in Electrical Grids: Harnessing Machine Learning for Enhanced Reliability

This study focuses on enhancing the reliability and redundancy of electrical systems through machine learning-based fault detection. This project’s objective is to create an inexpensive system for early detection of faults in electrical systems. This study proposed a methodology combining machine learning techniques and feature engineering. A dataset containing 12,001 sensor readings each consisting of six values was examined and analyzed by machine-learning techniques. Results showed the system's success in detecting faults with an accuracy of 99% from only six readings. In economically underserved regions, a common trend is the presence of less sophisticated electrical infrastructures. These systems translate into more frequent and longer power outages, a serious concern when essential institutions like hospitals rely on them to function. This research contributes to the field of fault detection by offering a practical and effective way to improve fault detection in underserved regions.

Comparative Analysis of Machine Learning Models for Prediction of Acute Liver Injury in Sepsis Patients.

Acute liver injury (ALI) is a common complication of sepsis and is associated with adverse clinical outcomes. We aimed to develop a model to predict the risk of ALI in patients with sepsis after hospitalization. Medical records of 3196 septic patients treated at the Lishui Central Hospital in Zhejiang Province from January 2015 to May 2023 were selected. Cohort 1 was divided into ALI and non-ALI groups for model training and internal validation. The initial laboratory test results of the study subjects were used as features for machine learning (ML), and models built using nine different ML algorithms were compared to select the best algorithm and model. The predictive performance of model stacking methods was then explored. The best model was externally validated in Cohort 2. In Cohort 1, LightGBM demonstrated good stability and predictive performance with an area under the curve (AUC) of 0.841. The top five most important variables in the model were diabetes, congestive heart failure, prothrombin time, heart rate, and platelet count. The LightGBM model showed stable and good ALI risk prediction ability in the external validation of Cohort 2 with an AUC of 0.815. Furthermore, an online prediction website was developed to assist healthcare professionals in applying this model more effectively. The Light GBM model can predict the risk of ALI in patients with sepsis after hospitalization.

Analysis of Machine Learning in Classifying Bank Profitability with Corruption Factor

Corruption, which is generally defined as the abuse of authority for personal benefit, is not a recent phenomenon and has become a major issue in almost all countries throughout the world. Indeed, a high level of corruption increases bank non-performing loans which in turn reduces profitability and intensifies the fragility of the banking industry. Given the adverse impacts, corruption has been used as one of the factors in bank performance evaluation. As research on corruption-bank performance with machine learning techniques is rarely reported in the literature, this paper presents the empirical comparison of different machine learning algorithms for classifying bank profitability. Besides machine learning performance comparisons, this paper presents the analysis of machine learning features importance to justify the effect of corruption factor in the different machine learning algorithms for classifying bank profitability. The results indicated that all the tested machine learning algorithms present a good ability of classifying bank profitability at accuracy percentages above 70% but corruption index has contributed very minimal effect to the machine learning performances. The framework of this research is highly reproducible to be extended with a more in-depth analysis, particularly on the bank profitability factors as well as on the machine learning algorithms.

Robust human locomotion and localization activity recognition over multisensory.

Human activity recognition (HAR) plays a pivotal role in various domains, including healthcare, sports, robotics, and security. With the growing popularity of wearable devices, particularly Inertial Measurement Units (IMUs) and Ambient sensors, researchers and engineers have sought to take advantage of these advances to accurately and efficiently detect and classify human activities. This research paper presents an advanced methodology for human activity and localization recognition, utilizing smartphone IMU, Ambient, GPS, and Audio sensor data from two public benchmark datasets: the Opportunity dataset and the Extrasensory dataset. The Opportunity dataset was collected from 12 subjects participating in a range of daily activities, and it captures data from various body-worn and object-associated sensors. The Extrasensory dataset features data from 60 participants, including thousands of data samples from smartphone and smartwatch sensors, labeled with a wide array of human activities. Our study incorporates novel feature extraction techniques for signal, GPS, and audio sensor data. Specifically, for localization, GPS, audio, and IMU sensors are utilized, while IMU and Ambient sensors are employed for locomotion activity recognition. To achieve accurate activity classification, state-of-the-art deep learning techniques, such as convolutional neural networks (CNN) and long short-term memory (LSTM), have been explored. For indoor/outdoor activities, CNNs are applied, while LSTMs are utilized for locomotion activity recognition. The proposed system has been evaluated using the k-fold cross-validation method, achieving accuracy rates of 97% and 89% for locomotion activity over the Opportunity and Extrasensory datasets, respectively, and 96% for indoor/outdoor activity over the Extrasensory dataset. These results highlight the efficiency of our methodology in accurately detecting various human activities, showing its potential for real-world applications. Moreover, the research paper introduces a hybrid system that combines machine learning and deep learning features, enhancing activity recognition performance by leveraging the strengths of both approaches.

Predicting the Fishery Ground of Jumbo Flying Squid (Dosidicus gigas) off Peru by Extracting Features of the Ocean Environment

We introduce a novel method that combines satellite data, advanced clustering techniques, machine learning feature extraction, and statistical models to enhance fishery forecasting accuracy. Focusing on jumbo flying squid in the southeast Pacific Ocean near Peru, we utilize MODIS-Aqua and MODIS-Terra satellite data on sea surface temperature (SST) to construct a deep convolutional embedded clustering (DCEC) model and extract the monthly SST features (FM) based on an optimized number of clusters determined by the Davies–Bouldi index (DBI). We use the extracted FM to construct a series of Generalized Additive Models (GAM) to forecast the catch per unit effort (CPUE) of jumbo flying squid within a spatial resolution of 0.5° × 0.5°. Our results demonstrate the following findings: (1) The SST feature clusters obtained through the DCEC model could capture the SST monthly variations; (2) The GAM models with FM outperform the models with the traditional monthly average SST in terms of predictive accuracy; (3) Using both FM and average SST together can further improve model performance. This study demonstrates the effectiveness of the DCEC combined with DBI in extracting marine environmental features and highlights the ocean environment feature extraction method to enhance the precision and reliability of fishery forecasting models.

AntiPhishStack: LSTM-Based Stacked Generalization Model for Optimized Phishing URL Detection

The escalating reliance on revolutionary online web services has introduced heightened security risks, with persistent challenges posed by phishing despite extensive security measures. Traditional phishing systems, reliant on machine learning and manual features, struggle with evolving tactics. Recent advances in deep learning offer promising avenues for tackling novel phishing challenges and malicious URLs. This paper introduces a two-phase stack generalized model named AntiPhishStack, designed to detect phishing sites. The model leverages the learning of URLs and character-level TF-IDF features symmetrically, enhancing its ability to combat emerging phishing threats. In Phase I, features are trained on a base machine learning classifier, employing K-fold cross-validation for robust mean prediction. Phase II employs a two-layered stacked-based LSTM network with five adaptive optimizers for dynamic compilation, ensuring premier prediction on these features. Additionally, the symmetrical predictions from both phases are optimized and integrated to train a meta-XGBoost classifier, contributing to a final robust prediction. The significance of this work lies in advancing phishing detection with AntiPhishStack, operating without prior phishing-specific feature knowledge. Experimental validation on two benchmark datasets, comprising benign and phishing or malicious URLs, demonstrates the model’s exceptional performance, achieving a notable 96.04% accuracy compared to existing studies. This research adds value to the ongoing discourse on symmetry and asymmetry in information security and provides a forward-thinking solution for enhancing network security in the face of evolving cyber threats.

Predicting superconducting transition temperature through advanced machine learning and innovative feature engineering

Superconductivity is a remarkable phenomenon in condensed matter physics, which comprises a fascinating array of properties expected to revolutionize energy-related technologies and pertinent fundamental research. However, the field faces the challenge of achieving superconductivity at room temperature. In recent years, Artificial Intelligence (AI) approaches have emerged as a promising tool for predicting such properties as transition temperature (Tc) to enable the rapid screening of large databases to discover new superconducting materials. This study employs the SuperCon dataset as the largest superconducting materials dataset. Then, we perform various data pre-processing steps to derive the clean DataG dataset, containing 13,022 compounds. In another stage of the study, we apply the novel CatBoost algorithm to predict the transition temperatures of novel superconducting materials. In addition, we developed a package called Jabir, which generates 322 atomic descriptors. We also designed an innovative hybrid method called the Soraya package to select the most critical features from the feature space. These yield R2 and RMSE values (0.952 and 6.45 K, respectively) superior to those previously reported in the literature. Finally, as a novel contribution to the field, a web application was designed for predicting and determining the Tc values of superconducting materials.

Optuna Based Hyperparameter Tuning for Improving the Performance Prediction Mortality and Hospital Length of Stay for Stroke Patients

Cardiovascular disease (CVD) stands as the foremost contributor to worldwide mortality, with strokes as part of significant CVD. Research on potential mortality risks and hospitalizations for stroke patients became crucial as a basis for evaluation to improve the quality and control of stroke patient services. Although machine learning technology has been widely used in health data analysis, understanding the relative performance and characteristics of machine learning (ML) models is still limited. Therefore, the study aims to broaden this understanding by comparing five ML models, namely XGBoost, Random Forest, Decision Trees, CatBoost, and Extra Trees, using stroke patient data from RSUD Banyumas Neural Poliklinik Indonesia. The model performance improvement process is the main focus, involving adjustments using the Optuna tuning library. Through this tuning approach, the key parameters of each ML model are optimally adjusted to improve their performance in predicting mortality risk and the duration of hospitalization for stroke patients. As a result, the XGBoost algorithm proved superior in predicting mortality (accuracy 86%, AUC 0.87) and the duration of hospitalization (accuracy 82%, AUC 0.79). This research has great potential to help hospitals identify high-risk stroke patients and plan more efficient treatment. This approach allows hospitals to use their resources better, improve medical services, and reduce unnecessary treatment costs

Development of an innovative approach using portable eye tracking to assist ADHD screening: a machine learning study.

Attention-deficit/hyperactivity disorder (ADHD) affects a significant proportion of the pediatric population, making early detection crucial for effective intervention. Eye movements are controlled by brain regions associated with neuropsychological functions, such as selective attention, response inhibition, and working memory, and their deficits are related to the core characteristics of ADHD. Herein, we aimed to develop a screening model for ADHD using machine learning (ML) and eye-tracking features from tasks that reflect neuropsychological deficits in ADHD. Fifty-six children (mean age 8.38 ± 1.58, 45 males) diagnosed with ADHD based on the Diagnostic and Statistical Manual of Mental Disorders, fifth edition were recruited along with seventy-nine typically developing children (TDC) (mean age 8.80 ± 1.82, 33 males). Eye-tracking data were collected using a digital device during the performance of five behavioral tasks measuring selective attention, working memory, and response inhibition (pro-saccade task, anti-saccade task, memory-guided saccade task, change detection task, and Stroop task). ML was employed to select relevant eye-tracking features for ADHD, and to subsequently construct an optimal model classifying ADHD from TDC. We identified 33 eye-tracking features in the five tasks with the potential to distinguish children with ADHD from TDC. Participants with ADHD showed increased saccade latency and degree, and shorter fixation time in eye-tracking tasks. A soft voting model integrating extra tree and random forest classifiers demonstrated high accuracy (76.3%) at identifying ADHD using eye-tracking features alone. A comparison of the model using only eye-tracking features with models using the Advanced Test of Attention or Stroop test showed no significant difference in the area under the curve (AUC) (p = 0.419 and p=0.235, respectively). Combining demographic, behavioral, and clinical data with eye-tracking features improved accuracy, but did not significantly alter the AUC (p=0.208). Our study suggests that eye-tracking features hold promise as ADHD screening tools, even when obtained using a simple digital device. The current findings emphasize that eye-tracking features could be reliable indicators of impaired neurobiological functioning in individuals with ADHD. To enhance utility as a screening tool, future research should be conducted with a larger sample of participants with a more balanced gender ratio.

Keyboard Data Protection Technique Using GAN in Password-Based User Authentication: Based on C/D Bit Vulnerability.

In computer systems, user authentication technology is required to identify users who use computers. In modern times, various user authentication technologies, including strong security features based on ownership, such as certificates and security cards, have been introduced. Nevertheless, password-based authentication technology is currently mainly used due to its convenience of use and ease of implementation. However, according to Verizon's "2022 Data Breach Investigations Report", among all security incidents, security incidents caused by password exposures accounted for 82%. Hence, the security of password authentication technology is important. Consequently, this article analyzes prior research on keyboard data attacks and defense techniques to draw the fundamental reasons for keyboard data attacks and derive countermeasures. The first prior research is about stealing keyboard data, an attack that uses machine learning to steal keyboard data to overcome the limitations of a C/D bit attack. The second prior research is an attack technique that steals keyboard data more efficiently by expanding the features of machine learning used in the first prior research. In this article, based on previous research findings, we proposed a keyboard data protection technique using GAN, a Generative Adversarial Network, and verified its feasibility. To summarize the results of performance evaluation with previous research, the machine learning-based keyboard data attack based on the prior research exhibited a 96.7% attack success rate, while the study's proposed method significantly decreased the attack success rate by approximately 13%. Notably, in all experiments, the average decrease in the keyboard data classification performance ranged from a minimum of -29% to a maximum of 52%. When evaluating performance based on maximum performance, all performance indicators were found to decrease by more than 50%.

MicroRNA-guided drug discovery for mitigating persistent pulmonary complications in critical COVID-19 survivors: A longitudinal pilot study.

The post-acute sequelae of SARS-CoV-2 infection pose a significant global challenge, with nearly 50% of critical COVID-19 survivors manifesting persistent lung abnormalities. The lack of understanding about the molecular mechanisms and effective treatments hampers their management. Here, we employed microRNA (miRNA) profiling to decipher the systemic molecular underpinnings of the persistent pulmonary complications. We conducted a longitudinal investigation including 119 critical COVID-19 survivors. A comprehensive pulmonary evaluation was performed in the short-term (median = 94.0days after hospital discharge) and long-term (median = 358 days after hospital discharge). Plasma miRNAs were quantified at the short-term evaluation using the gold-standard technique, RT-qPCR. The analyses combined machine learning feature selection techniques with bioinformatic investigations. Two additional datasets were incorporated for validation. In the short-term, 84% of the survivors exhibited impaired lung diffusion (DLCO < 80% of predicted). One year post-discharge, 54.4% of this patient subgroup still presented abnormal DLCO . Four feature selection methods identified two specific miRNAs, miR-9-5p and miR-486-5p, linked to persistent lung dysfunction. The downstream experimentally validated targetome included 1473 genes, with heterogeneous enriched pathways associated with inflammation, angiogenesis and cell senescence. Validation studies using RNA-sequencing and proteomic datasets emphasized the pivotal roles of cell migration and tissue repair in persistent lung dysfunction. The repositioning potential of the miRNA targets was limited. Our study reveals early mechanistic pathways contributing to persistent lung dysfunction in critical COVID-19 survivors, offering a promising approach for the development of targeted disease-modifying agents.