AI-Driven Heart Disease Prediction Using Machine Learning and Deep Learning Techniques

International Journal of Computational and Experimental Science and Engineering, Journal Year: 2025, Volume and Issue: 11(2)

Published: April 10, 2025

Heart disease remains a leading cause of mortality worldwide, necessitating early detection and prevention strategies. This study explores machine learning (ML) approaches for predicting heart using patient datasets. Various ML algorithms, including Logistic Regression, Naive Bayes, Support Vector Machine (SVM), K-Nearest Neighbors (KNN), Decision Tree, Random Forest, XGBoost, an Artificial Neural Network (ANN), were implemented to classify presence. The Forest model achieved the highest accuracy 95%. findings demonstrate that can significantly enhance prediction, aiding diagnosis treatment.

Language: Английский

---

Deep Learning Based Automated Detection of Arcus Senilis and Its Clinical Risks in Ocular Health

International Journal of Computational and Experimental Science and Engineering, Journal Year: 2025, Volume and Issue: 11(2)

Published: April 10, 2025

Arcus Senilis is a clinical indicator of lipid deposition in the cornea, commonly observed aging individuals. This study aims to develop an automated deep learning-based pipeline for detecting and estimating cholesterol levels from ocular images. We implemented image-based classification system using EfficientNetB0, state-of-the-art convolutional neural network (CNN). The dataset was pre-processed Contrast Limited Adaptive Histogram Equalization (CLAHE) enhance contrast. model trained transfer learning, incorporating global average pooling fully connected layers classify presence estimate levels. Additionally, patient metadata, including age levels, integrated prediction accuracy. on labelled dataset, with multi-task learning approach handling both (Arcus detection) regression (cholesterol level estimation). Performance evaluated Mean Absolute Error (MAE), R² Score, Accuracy, Confusion Matrices. proposed achieved accuracy 92.5% detection (MAE) 8.4 mg/dL estimation. effectively distinguished normal eyes provided clinically relevant estimations. Evaluation metrics, precision, recall, F1-score, demonstrated its reliability compared traditional machine approaches such as SVM + HOG Features, ResNet50, VGG16. provides non-invasive, accurate, solution findings suggest potential applications ophthalmic diagnostics metabolism assessment.

Language: Английский

---

An Efficient Hybrid Improved Feature Vector Manifold Clustering with Neighbour Search Optimization

International Journal of Computational and Experimental Science and Engineering, Journal Year: 2025, Volume and Issue: 11(2)

Published: April 29, 2025

In this paper, the IFMCNSO algorithm a novel hybrid Improved Feature Vector Manifold clustering with Neighbour search optimization —is presented. Many methods for linear or nonlinear manifold have been developed recently. While in many cases they proven to perform better than classic algorithms, majority of these approaches high complexity. order overcome problem, particularly high-dimensional datasets, work provides an effective method called IFMCNSO. By using strategy, domain which feature vector learning and Neighbor techniques can be used is greatly expanded, enabling parameterization real-world data sets. A good nearly optimal solution found acceptable amount time. comprehensive comparison proposed state-of-the-art namely DCNaN, RDMN, HFMST, HFMST-PSO, reveals that achieves higher Rand Index (RI) Adjusted (ARI) scores, underscoring its exceptional performance accuracy

Language: Английский

---