Deep Learning Framework for Optimizing Early Detection of Measles Using Transfer Learning DOI

Nouman Saleem,

Anam Ishaq,

Malaika Riaz

et al.

Indus journal of bioscience research., Journal Year: 2024, Volume and Issue: 2(2), P. 985 - 998

Published: Dec. 15, 2024

Measles is a highly infectious viral disease that can have serious health consequences. Accurate and early diagnosis crucial. This study aims to enhance automated classification detection of this disease. To address the class imbalance, we augmented dataset normal images. Spatial features were extracted using convolutional neural networks, traditional classifiers, including support vector machine, Random Forest, logistic regression, k-nearest neighbors applied these features. Initial accuracy based solely on spatial was as follows: Forest 63%, SVM KNN 60%, Logistic Regression 63%. Through 10-fold cross-validation, mean accuracies recorded 65% for RF, 62% SVM, 60% KNN, 61% LR. Despite initial results, implementation transfer learning led significant improvements. By extracting probabilistic from RF models concatenating derived features, substantially enhanced. The improved model achieved 99% LR, with reaching 98%. Cross-validation confirmed robustness models, approximately 98% minimal standard deviations 0.01. findings demonstrate combining classifiers improves efficiency lesion approach shows potential clinical applications.

Language: Английский

Novel Meta Learning Approach for Detecting Postpartum Depression Disorder Using Questionnaire Data DOI Creative Commons
Shazia Nasim, Ahmad Sami Al-Shamayleh, Nisrean Thalji

et al.

IEEE Access, Journal Year: 2024, Volume and Issue: 12, P. 101247 - 101259

Published: Jan. 1, 2024

Postpartum depression (PPD) is becoming increasingly prevalent worldwide, often manifesting in new mothers due to a complex interplay of physical, behavioral, and emotional transformations post-childbirth. The primary aim our research analyze the contributory factors leading PPD, including familial, social, other maternal health-related aspects, devise predictive model that can accurately assess risk PPD. In this research, we analyzed benchmark dataset 1,503 entries gathered from medical institution, where data was compiled through questionnaires disseminated using digital Google Forms platform. We deployed eleven advanced machine-learning algorithms for comparison. proposed novel MDKR model, meta-learner designed excel predicting Questionnaire initially processed decision tree, k-nearest classifier, random forest models. Subsequently, outputs these models are fed into multi-layer perceptron final prediction. Compared state-of-the-art studies, surfaced as most proficient, with an exemplary accuracy 99% detecting addition, have confirmed performance k-fold validation tuning hyperparameters. comparative assessment all concerning their ability predict PPD levels, emerged superior model. This meta-learning has significantly contributed identifying pivotal influencing enhancing framework within healthcare domains.

Language: Английский

Citations

2
Innovative Approach to Detecting Autism Spectrum Disorder Using Explainable Features and Smart Web Application DOI Creative Commons
Mohammad Abu Tareq Rony,

Fatama Tuz Johora,

Nisrean Thalji

et al.

Mathematics, Journal Year: 2024, Volume and Issue: 12(22), P. 3515 - 3515

Published: Nov. 11, 2024

Autism Spectrum Disorder (ASD) is a complex developmental condition marked by challenges in social interaction, communication, and behavior, often involving restricted interests repetitive actions. The diversity symptoms skill profiles across individuals creates diagnostic landscape that requires multifaceted approach for accurate understanding intervention. This study employed advanced machine-learning techniques to enhance the accuracy reliability of ASD diagnosis. We used standard dataset comprising 1054 patient samples 20 variables. research methodology involved rigorous preprocessing, including selecting key variables through data mining (DM) visualization Chi-Square tests, analysis variance, correlation analysis, along with outlier removal ensure robust model performance. proposed DM logistic regression (LR) Shapley Additive exPlanations (DMLRS) achieved highest at 99%, outperforming state-of-the-art methods. eXplainable artificial intelligence was incorporated using interpretability. compared other approaches, XGBoost, Deep Models Residual Connections Ensemble (DMRCE), fast lightweight automated machine learning systems. Each method fine-tuned, performance verified k-fold cross-validation. In addition, real-time web application developed integrates DMLRS Django framework app represents significant advancement medical informatics, offering practical, user-friendly, innovative solution early detection

Language: Английский

Citations

0
SAD: Self-assessment of depression for Bangladeshi university students using machine learning and NLP DOI Creative Commons
Md Shawmoon Azad,

Shakirul Islam Leeon,

Riasat Khan

et al.

Array, Journal Year: 2024, Volume and Issue: 25, P. 100372 - 100372

Published: Dec. 9, 2024

Language: Английский

Citations

0
Deep Learning Framework for Optimizing Early Detection of Measles Using Transfer Learning DOI

Nouman Saleem,

Anam Ishaq,

Malaika Riaz

et al.

Indus journal of bioscience research., Journal Year: 2024, Volume and Issue: 2(2), P. 985 - 998

Published: Dec. 15, 2024

Measles is a highly infectious viral disease that can have serious health consequences. Accurate and early diagnosis crucial. This study aims to enhance automated classification detection of this disease. To address the class imbalance, we augmented dataset normal images. Spatial features were extracted using convolutional neural networks, traditional classifiers, including support vector machine, Random Forest, logistic regression, k-nearest neighbors applied these features. Initial accuracy based solely on spatial was as follows: Forest 63%, SVM KNN 60%, Logistic Regression 63%. Through 10-fold cross-validation, mean accuracies recorded 65% for RF, 62% SVM, 60% KNN, 61% LR. Despite initial results, implementation transfer learning led significant improvements. By extracting probabilistic from RF models concatenating derived features, substantially enhanced. The improved model achieved 99% LR, with reaching 98%. Cross-validation confirmed robustness models, approximately 98% minimal standard deviations 0.01. findings demonstrate combining classifiers improves efficiency lesion approach shows potential clinical applications.

Language: Английский

Citations

0