A hybrid learning network with progressive resizing and PCA for diagnosis of cervical cancer on WSI slides

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

Опубликована: Апрель 14, 2025

Язык: Английский

---

Enhancing Lung Cancer Classification Effectiveness Through Hyperparameter-Tuned Support Vector Machine

Journal of Computing Theories and Applications, Год журнала: 2024, Номер 1(4), С. 396 - 406

Опубликована: Март 25, 2024

This research aims to improve the effectiveness of lung cancer classification performance using Support Vector Machines (SVM) with hyperparameter tuning. Using Radial Basis Function (RBF) kernels in SVM helps deal non-linear problems. At same time, tuning is done through Random Grid Search find best combination parameters. Where parameter settings are C = 10, Gamma Probability True. Test results show that tuned improves accuracy, precision, specificity, and F1 score significantly. However, there was a slight decrease recall, namely 0.02. Even though recall one most important measuring tools disease classification, especially imbalanced datasets, specificity also plays vital role avoiding misidentifying negative cases. Without tuning, so poor considering both becomes very important. Overall, obtained by proposed method 0.99 for 1.00 0.98 f1-score, specificity. confirms potential SVMs addressing complex data challenges offers insights medical diagnostic applications.

Язык: Английский

---

VTCNet: A Feature Fusion DL Model Based on CNN and ViT for the Classification of Cervical Cells

International Journal of Imaging Systems and Technology, Год журнала: 2024, Номер 34(5)

Опубликована: Авг. 27, 2024

ABSTRACT Cervical cancer is a common malignancy worldwide with high incidence and mortality rates in underdeveloped countries. The Pap smear test, widely used for early detection of cervical cancer, aims to minimize missed diagnoses, which sometimes results higher false‐positive rates. To enhance manual screening practices, computer‐aided diagnosis (CAD) systems based on machine learning (ML) deep (DL) classifying cells have been extensively researched. In our study, we introduced DL‐based method named VTCNet the task cell classification. Our approach combines CNN‐SPPF ViT components, integrating modules like Focus SeparableC3, capture more potential information, extract local global features, merge them classification performance. We evaluated public SIPaKMeD dataset, achieving accuracies, precision, recall, F1 scores 97.16%, 97.22%, 97.19%, 97.18%, respectively. also conducted additional experiments Herlev where outperformed previous methods. achieved accuracy than traditional ML or shallow DL models through this integration. Related codes: https://github.com/Camellia‐0892/VTCNet/tree/main .

Язык: Английский

---

Improving Detection and Prediction of Traffic Congestion in VANETs: An Examination of Machine Learning

International Journal of Computing and Digital Systems, Год журнала: 2024, Номер 15(1), С. 947 - 960

Опубликована: Фев. 5, 2024

Traffic congestion remains a pressing challenge in urban areas, causing significant economic and environmental repercussions.To address this issue, accurate detection prediction of traffic are imperative for effective management planning.This research study investigates the efficacy Support Vector Machines (SVM) various other machine learning algorithms augmenting Vehicular Ad hoc Networks (VANETs).Leveraging historical patterns, we train evaluate performance algorithms.Our results demonstrate potential SVM, coupled with advanced feature engineering techniques, to outperform methods accurately identifying forecasting congestion.The SVM classifier achieved an impressive classification accuracy 0.99, showcasing its effectiveness handling diverse scenarios.Additionally, K-Nearest Neighbors (KNN) Ensemble Learning classifiers also yielded commendable accuracies 0.99.Notably, Decision Tree (DT) attained perfect score 1.00, indicating robustness patterns.The proposed approach not only achieves high but exhibits remarkable scalability, enabling application across scenarios.These findings contribute significantly development intelligent systems, providing valuable insights into optimizing transportation networks.Ultimately, implementing our holds alleviate congestion, enhance travel efficiency, foster sustainability.

Язык: Английский

---

Cervical Cancer Classification: Optimizing Accuracy, Precision, and Recall using SMOTE Preprocessing and t-SNE Feature Extraction

Опубликована: Янв. 24, 2024

Machine Learning(ML) algorithms are used in cervical cancer categorization to determine whether a person has based on pertinent information from medical records. This procedure is critical healthcare for prior treatment. For this kind of problem, many ML approaches such as RF(Random Forest), SVM(Support Vector Machine), and LR(Logistic Regression) can be used. We employed techniques increase the accuracy, precision, recall diagnosis study. SMOTE preprocessing was resolve data imbalance by producing synthetic samples minority class. Furthermore, t-SNE feature extraction capture complex structures data. SVM consistently outperforms RF Logistic Regression when combined with extraction, demonstrating improved rates. research highlights efficacy these strategies enhancing categorization, indicating possibilities precise reliable forecasting investigations. The proposed gave accuracy 92.60%, precision 0.91 0.90 respectively. tool execution python.

Язык: Английский

---

Comprehensive analysis of artificial intelligence techniques for gynaecological cancer: symptoms identification, prognosis and prediction

Artificial Intelligence Review, Год журнала: 2024, Номер 57(8)

Опубликована: Июль 29, 2024

Abstract Gynaecological cancers encompass a spectrum of malignancies affecting the female reproductive system, comprising cervix, uterus, ovaries, vulva, vagina, and fallopian tubes. The significant health threat posed by these worldwide highlight crucial need for techniques early detection prediction gynaecological cancers. Preferred reporting items systematic reviews Meta-Analysis guidelines are used to select articles published from 2013 up 2023 on Web Science, Scopus, Google Scholar, PubMed, Excerpta Medical Database, AI technique Based study different cancer, results also compared using various quality parameters such as rate, accuracy, sensitivity, specificity, area under curve precision, recall, F1-score. This work highlights impact cancer women belonging age groups regions world. A detailed categorization traditional like physical-radiological, bio-physical bio-chemical detect organizations is presented in study. Besides, this explores methodology researchers which plays role identifying symptoms at earlier stages. paper investigates pivotal years, highlighting periods when highest number research published. challenges faced while performing AI-based highlighted work. features representations Magnetic Resonance Imaging (MRI), ultrasound, pap smear, pathological, etc., proficient algorithms explored. comprehensive review contributes understanding improving prognosis cancers, provides insights future directions clinical applications. has potential substantially reduce mortality rates linked enabling identification, individualised risk assessment, improved treatment techniques. would ultimately improve patient outcomes raise standard healthcare all individuals.

Язык: Английский

---

Leveraging Machine Learning for Sophisticated Rental Value Predictions: A Case Study from Munich, Germany

Applied Sciences, Год журнала: 2024, Номер 14(20), С. 9528 - 9528

Опубликована: Окт. 18, 2024

There has been very limited research conducted to predict rental prices in the German real estate market using an AI-based approach. From a general perspective, conventional approaches struggle handle large amounts of data and fail consider numerous elements that affect prices. The absence sophisticated, data-driven analytical tools further complicates this situation, impeding stakeholders, such as tenants, landlords, agents, government, from obtaining accurate insights necessary for making well-informed decisions area. This paper applies novel machine learning (ML) approaches, including ensemble techniques, neural networks, linear regression (LR), tree-based algorithms, specifically designed forecasting Munich. To ensure accuracy reliability, performance these models is evaluated R2 score root mean squared error (RMSE). study provides two feature sets model comparison, selected by particle swarm optimisation (PSO) CatBoost. These selection methods identify significant variables based on different mechanisms, seeking optimal solution with objective function converting categorical features into target statistics (TSs) address high-dimensional issues. are ideal dataset, which contains 49 features. Testing 10 ML algorithms helps validate robustness efficacy approach utilising PyTorch framework. findings illustrate combined PyTorch-based networks (PNNs) demonstrate high compared standalone models, regardless changes. improved indicates framework predictive tasks advantageous, evidenced statistical significance test terms both RMSE (p-values < 0.001). integration results display outstanding accuracy, averaging 90% across sets. Particularly, XGB model, exhibited lowest among all sets, significantly 0.8903 0.9097 set 1 0.8717 0.9022 2 after being PNN. showcase framework, enhancing precision reliability predicting dynamic market. Given demonstrates consistent varying characteristics, methodology may be applied other locations. By offering projections, it aids investors, renters, property managers, regulators facilitating better decision-making sector.

Язык: Английский

---

Enhancing real-time health monitoring with hybrid recurrent long short-term tyrannosaurus search for menstrual cups

Biomedical Signal Processing and Control, Год журнала: 2024, Номер 100, С. 107065 - 107065

Опубликована: Окт. 23, 2024

Язык: Английский

---

Therapeutic Decision Making in Prevascular Mediastinal Tumors Using CT Radiomics and Clinical Features: Upfront Surgery or Pretreatment Needle Biopsy?

Cancers, Год журнала: 2024, Номер 16(4), С. 773 - 773

Опубликована: Фев. 13, 2024

The study aimed to develop machine learning (ML) classification models for differentiating patients who needed direct surgery from core needle biopsy among with prevascular mediastinal tumor (PMT). Patients PMT received a contrast-enhanced computed tomography (CECT) scan and initial management between January 2010 December 2020 were included in this retrospective study. Fourteen ML algorithms used construct candidate via the voting ensemble approach, based on preoperative clinical data radiomic features extracted CECT. accuracy of diagnosis was 86.1%. first model built by randomly choosing seven set fourteen had 88.0% (95% CI = 85.8 90.3%). second combination five models, including NeuralNetFastAI, NeuralNetTorch, RandomForest Entropy, Gini, XGBoost, 90.4% 87.9 93.0%), which significantly outperformed (p < 0.05). Due superior performance, clinical–radiomic may be as decision support system facilitate selection PMT.

Язык: Английский

---

Machine Learning Approach with Harmonized Multinational Datasets for Enhanced Prediction of Hypothyroidism in Patients with Type 2 Diabetes

Diagnostics, Год журнала: 2024, Номер 14(11), С. 1152 - 1152

Опубликована: Май 31, 2024

Type 2 diabetes (T2D) is a global health concern with increasing prevalence. Comorbid hypothyroidism (HT) exacerbates kidney, cardiac, neurological and other complications of T2D; these risks can be mitigated pharmacologically upon detecting HT. The current HT standard care (SOC) screening in T2D infrequent, delaying diagnosis treatment. We present first-to-date machine learning algorithm (MLA) clinical decision tool to classify patients as low vs. high risk for developing comorbid the MLA was developed using readily available patient data from harmonized multinational datasets. trained on NIH All US (AoU) UK Biobank (UKBB) (Combined dataset) achieved negative predictive value (NPV) 0.989 an AUROC 0.762 Combined dataset, exceeding AUROCs models AoU or UKBB alone (0.666 0.622, respectively), indicating that dataset diversity training improves performance. This high-NPV automated supplement SOC rule out risk, allowing prioritization lab-based testing at-risk patients. Conversely, output designates at allows tailored management thereby promotes improved outcomes.

Язык: Английский

---