Deep convolutional neural networks with genetic algorithm-based synthetic minority over-sampling technique for improved imbalanced data classification

Applied Soft Computing, Год журнала: 2024, Номер 156, С. 111491 - 111491

Опубликована: Март 11, 2024

Язык: Английский

---

Entropy-based genetic feature engineering and multi-classifier fusion for anomaly detection in vehicle controller area networks

Future Generation Computer Systems, Год журнала: 2025, Номер unknown, С. 107779 - 107779

Опубликована: Фев. 1, 2025

Язык: Английский

---

Prediction of tablet disintegration time based on formulations properties via artificial intelligence by comparing machine learning models and validation

Scientific Reports, Год журнала: 2025, Номер 15(1)

Опубликована: Апрель 21, 2025

This research assesses multiple predictive models aimed at estimating disintegration time for pharmaceutical oral formulations, based on a dataset comprising nearly 2,000 data points that include molecular, physical, compositional, and formulation attributes. Drug properties were considered as the inputs to estimate output which is tablet time. Advanced machine learning methods, including Bayesian Ridge Regression (BRR), Relevance Vector Machine (RVM), Sparse Learning (SBL) utilized after comprehensive preprocessing involving outlier detection, normalization, feature selection. Grey Wolf Optimization (GWO) was model optimization obtain optimal combinations of hyper-parameters. Among models, SBL stood out its superior performance, achieving highest R² scores lowest Root Mean Square Error (RMSE) Absolute Percentage (MAPE) error rates in both training testing phases. It also demonstrated robustness effectively avoided overfitting. SHapley Additive exPlanations (SHAP) analysis provided valuable insights into contributions, highlighting wetting sodium saccharin key factors influencing

Язык: Английский

---

Developing a Model to Predict Self-Reported Student Performance during Online Education Based on the Acoustic Environment

Sustainability, Год журнала: 2024, Номер 16(11), С. 4411 - 4411

Опубликована: Май 23, 2024

In recent years, great developments in online university education have been observed, favored by advances ICT. There are numerous studies on the perception of academic performance classes, influenced aspects a very diverse nature; however, acoustic environment students at home, which can certainly affect activities, has barely evaluated. This study assesses influence home students’ self-reported performance. assessment is performed calculating prediction models using Recursive Feature Elimination method with 40 initial features and following classifiers: Random Forest, Gradient Boosting, Support Vector Machine. The optimal number predictors their relative importance were also was assessed metrics such as accuracy area under receiver operating characteristic curve (ROC_AUC-score). model smallest (with 14 predictors, 9 them about perceived environment) best achieves an 0.7794; furthermore, maximum difference for same algorithm between 33 0.03. Consequently, simplicity ease interpretation, reduced variables preferred.

Язык: Английский

---

Prediction of Pt, Ir, Ru, and Rh complexes light absorption in the therapeutic window for phototherapy using machine learning

Journal of Cheminformatics, Год журнала: 2025, Номер 17(1)

Опубликована: Янв. 5, 2025

Effective light-based cancer treatments, such as photodynamic therapy (PDT) and photoactivated chemotherapy (PACT), rely on compounds that are activated by light efficiently, absorb within the therapeutic window (600–850 nm). Traditional prediction methods for these absorption properties, including Time-Dependent Density Functional Theory (TDDFT), often computationally intensive time-consuming. In this study, we explore a machine learning (ML) approach to predict in region of platinum, iridium, ruthenium, rhodium complexes, aiming at streamlining screening potential photoactivatable prodrugs. By compiling dataset 9775 complexes from Reaxys database, trained six classification models, random forests, support vector machines, neural networks, utilizing various molecular descriptors. Our findings indicate Extreme Gradient Boosting Classifier (XGBC) paired with AtomPairs2D descriptors delivers highest predictive accuracy robustness. This ML-based method significantly accelerates identification suitable compounds, providing valuable tool early-stage design development phototherapy drugs. The also allows change relevant structural characteristics base molecule using information supervised approach. Scientific Contribution: proposed predicts ability transition metal-based UV–vis window, key trait phototherapeutic agents. While ML models have been used properties organic molecules, applying metal is novel. model efficient, fast, resource-light, decision tree-based algorithms provide interpretable results. interpretability helps understand rules facilitates targeted modifications convert inactive into potentially active ones.

Язык: Английский

---

A citizen science platform to sample beehive sounds for monitoring ANSP

Journal of Environmental Management, Год журнала: 2025, Номер 375, С. 124247 - 124247

Опубликована: Янв. 25, 2025

Язык: Английский

---

Psychological Impact on First Responders Dispatched to Out-of-Hospital Cardiac Arrest via Smartphone Alerting System: A Longitudinal Survey-Based Study

Resuscitation Plus, Год журнала: 2025, Номер unknown, С. 100941 - 100941

Опубликована: Март 1, 2025

Язык: Английский

---

Automated Detection of Shot Events in Game Phases Using GNSS Data from a Single Team

Communications in computer and information science, Год журнала: 2025, Номер unknown, С. 95 - 105

Опубликована: Янв. 1, 2025

Язык: Английский

---

Prediction of quantitative function of artificially-designed protein from structural information

bioRxiv (Cold Spring Harbor Laboratory), Год журнала: 2025, Номер unknown

Опубликована: Апрель 16, 2025

Abstract Artificially designed proteins are widely used in applications such as optogenetics and biosensing. While experimental optimization of these is effective, it also costly labor-intensive. To address this challenge, computational approaches have been developed, primarily relying on sequence-based features. However, protein function inherently tied to its three-dimensional (3D) structure, incorporating structural information could enable more accurate predictions provide deeper biological interpretability. Here, we proposed a structure-based analysis framework called ‘Foldinsight’ for predicting functionalities. In our framework, first predict structures from sequences using AlphaFold2 then utilize properties. Since vary the number atoms lack direct atomic correspondence, applied molecular field mapping, which captures energy states surrounding converts them into fixed-length numerical vectors. This transformation enables application machine learning, allowing properties be predicted structure-derived Applying channelrhodopsin mutants, achieved predictive performance comparable models. Additionally, successfully identified key regions contributing functional differences, highlighting advantage data modeling.

Язык: Английский

---

Feature Extraction and Classification of Social Media Data Using Deep Learning Techniques for Depression Detection

Lecture notes in networks and systems, Год журнала: 2025, Номер unknown, С. 551 - 561

Опубликована: Янв. 1, 2025

Язык: Английский

---