Artificial intelligence in pediatric allergy research DOI Creative Commons
Daniil Lisik, Rani Basna, Duy-Tai Dinh

и другие.

European Journal of Pediatrics, Год журнала: 2024, Номер 184(1)

Опубликована: Дек. 21, 2024

Abstract Atopic dermatitis, food allergy, allergic rhinitis, and asthma are among the most common diseases in childhood. They heterogeneous diseases, can co-exist their development, manifest complex associations with other disorders environmental hereditary factors. Elucidating these intricacies by identifying clinically distinguishable groups actionable risk factors will allow for better understanding of which enhance clinical management benefit society affected individuals families. Artificial intelligence (AI) is a promising tool this context, enabling discovery meaningful patterns data. Numerous studies within pediatric allergy have continue to use AI, primarily characterize disease endotypes/phenotypes develop models predict future outcomes. However, implementations used relatively simplistic data from one source, such as questionnaires. In addition, methodological approaches reporting lacking. This review provides practical hands-on guide conducting AI-based including (1) an introduction essential AI concepts techniques, (2) blueprint structuring analysis pipelines (from selection variables interpretation results), (3) overview pitfalls remedies. Furthermore, state-of-the art implementation research, well implications perspectives discussed. Conclusion : solutions undoubtedly transform showcased findings innovative technical solutions, but fully harness potential, methodologically robust more advanced techniques on richer be needed. What Known: • Pediatric allergies common, inflicting substantial morbidity societal costs. The field artificial undergoing rapid increasing various fields medicine research. New: Promising applications been reported, largely lags behind fields, particularly regard algorithms non-tabular lacking computational hampers evidence synthesis critical appraisal. Multi-center collaborations multi-omics rich unstructured utilization deep learning likely provide impactful discoveries.

Язык: Английский

A new method for GAN-based data augmentation for classes with distinct clusters DOI
Mehmet Kuntalp, Okan Düzyel

Expert Systems with Applications, Год журнала: 2023, Номер 235, С. 121199 - 121199

Опубликована: Авг. 17, 2023

Язык: Английский

Процитировано

18
An Optimized Machine Learning and Deep Learning Framework for Facial and Masked Facial Recognition DOI Creative Commons
Putthiporn Thanathamathee, Siriporn Sawangarreerak,

Prateep Kongkla

и другие.

Emerging Science Journal, Год журнала: 2023, Номер 7(4), С. 1173 - 1187

Опубликована: Июль 12, 2023

In this study, we aimed to find an optimized approach improving facial and masked recognition using machine learning deep techniques. Prior studies only used a single model for classification did not report optimal parameter values. contrast, utilized grid search with hyperparameter tuning nested cross-validation achieve better results during the verification phase. We performed experiments on large dataset of images without masks. Our findings showed that SVM had highest accuracy compared other models, achieving 0.99912. The precision values masks were 0.99925 0.98417, respectively. tested our in real-life scenarios found it accurately identified individuals through recognition. Furthermore, study stands out from others as incorporates phase enhance model's performance, generalization, robustness while optimizing data utilization. has potential implications security systems various domains, including public safety healthcare.

Язык: Английский

Процитировано

10
Fault diagnosis for rotating machinery based on deep learning DOI
Wessam S. ElAraby, Ahmed H. Madian, Mohamed H. Saad

и другие.

Noise & Vibration Worldwide, Год журнала: 2025, Номер unknown

Опубликована: Июнь 4, 2025

Effective fault diagnosis is critical for the safe and efficient operation of rotating machinery in nuclear facilities. This paper proposes a deep learning-based approach that integrates multi-domain signal analysis transfer learning to classify rotor conditions as either healthy or faulty. Vibration signals are transformed into 2D images processed using pretrained models: ResNet50, GoogleNet, custom Deep Convolutional Neural Network (DCNN). Signal transforms, including Fast Fourier Transform (FFT), Fractional Transform, Short-Time (STFT), Continuous Wavelet (CWT), Synchrosqueezing applied enhance feature representation. ResNet50 achieved up 100% accuracy on primary dataset over 99% secondary dataset. GoogleNet DCNN also demonstrated excellent performance, achieving accuracies specific domains. Additionally, YamNet enabled effective sound-based classification vibration signals. These results show advanced processing together with can lead very accurate quick detection important safety situations.

Язык: Английский

Процитировано

0
Diagnosis of Lung Diseases from Chest X-Ray Images Using Different Fusion Techniques DOI

Fatma Mostafa,

Lamiaa A. Elrefaei, Mostafa M. Fouda

и другие.

Опубликована: Авг. 23, 2023

Lung diseases refer to a group of disorders that affect the lungs and respiratory system. Several factors, such as genetics, environmental pollution, infections, smoking can these. include coronavirus (COVID-19), pneumonia, chronic obstructive pulmonary disease (COPD), asthma. cause significant damage lung function lead failure or even death. The symptoms range from mild difficulty breathing severe ones, including chest pain, bloody coughing, shortness breath. Early detection increase chances successful treatment improve overall outcome for affected individuals. Artificial intelligence (AI) has demonstrated considerable potential detecting diagnosing through machine learning algorithms deep models. using X-rays (CXRs) is in this paper by applying feature-level fusion (FLF) decision-level (DLF) techniques. FLF involves concatenating features two models before classification process. In comparison, DLF executed after training then results make single decision. are DenseNet-169 Vision Transformer (ViT-L32). On COVID-19 Radiography database, proposed have been tested trained. data preprocessed augmentation blurring method. An 'Adam' optimizer used while compiling model. accuracy 93.3%, achieved an 94.54%, which better than without fusion.

Язык: Английский

Процитировано

1
Artificial intelligence in pediatric allergy research DOI Creative Commons
Daniil Lisik, Rani Basna, Duy-Tai Dinh

и другие.

European Journal of Pediatrics, Год журнала: 2024, Номер 184(1)

Опубликована: Дек. 21, 2024

Abstract Atopic dermatitis, food allergy, allergic rhinitis, and asthma are among the most common diseases in childhood. They heterogeneous diseases, can co-exist their development, manifest complex associations with other disorders environmental hereditary factors. Elucidating these intricacies by identifying clinically distinguishable groups actionable risk factors will allow for better understanding of which enhance clinical management benefit society affected individuals families. Artificial intelligence (AI) is a promising tool this context, enabling discovery meaningful patterns data. Numerous studies within pediatric allergy have continue to use AI, primarily characterize disease endotypes/phenotypes develop models predict future outcomes. However, implementations used relatively simplistic data from one source, such as questionnaires. In addition, methodological approaches reporting lacking. This review provides practical hands-on guide conducting AI-based including (1) an introduction essential AI concepts techniques, (2) blueprint structuring analysis pipelines (from selection variables interpretation results), (3) overview pitfalls remedies. Furthermore, state-of-the art implementation research, well implications perspectives discussed. Conclusion : solutions undoubtedly transform showcased findings innovative technical solutions, but fully harness potential, methodologically robust more advanced techniques on richer be needed. What Known: • Pediatric allergies common, inflicting substantial morbidity societal costs. The field artificial undergoing rapid increasing various fields medicine research. New: Promising applications been reported, largely lags behind fields, particularly regard algorithms non-tabular lacking computational hampers evidence synthesis critical appraisal. Multi-center collaborations multi-omics rich unstructured utilization deep learning likely provide impactful discoveries.

Язык: Английский

Процитировано

0