A hybrid ResNet-ViT approach to bridge the global and local features for myocardial infarction detection.

Myocardial infarction (MI) remains a significant contributor to global mortality and morbidity, necessitating accurate and timely diagnosis. Current diagnostic methods encounter challenges in capturing intricate patterns, urging the need for advanced automated approaches to enhance MI detection. In this study, we strive to advance MI detection by proposing a hybrid approach that combines the strengths of ResNet and Vision Transformer (ViT) models, leveraging global and local features for improved accuracy. We introduce a slim-model ViT design with multibranch networks and channel attention mechanisms to enhance patch embedding extraction, addressing ViT's limitations. By training data through both ResNet and modified ViT models, we incorporate a dual-pathway feature extraction strategy. The fusion of global and local features addresses the challenge of robust feature vector creation. Our approach showcases enhanced learning capabilities through modified ViT architecture and ResNet architecture. The dual-pathway training enriches feature extraction, culminating in a comprehensive feature vector. Preliminary results demonstrate significant potential for accurate detection of MI. Our study introduces a hybrid ResNet-ViT model for advanced MI detection, highlighting the synergy between global and local feature extraction. This approach holds promise for elevating MI classification accuracy, with implications for improved patient care. Further validation and clinical applicability exploration are warranted.

Topic2features: a novel framework to classify noisy and sparse textual data using LDA topic distributions.

In supervised machine learning, specifically in classification tasks, selecting and analyzing the feature vector to achieve better results is one of the most important tasks. Traditional methods such as comparing the features' cosine similarity and exploring the datasets manually to check which feature vector is suitable is relatively time consuming. Many classification tasks failed to achieve better classification results because of poor feature vector selection and sparseness of data. In this paper, we proposed a novel framework, topic2features (T2F), to deal with short and sparse data using the topic distributions of hidden topics gathered from dataset and converting into feature vectors to build supervised classifier. For this we leveraged the unsupervised topic modelling LDA (latent dirichlet allocation) approach to retrieve the topic distributions employed in supervised learning algorithms. We made use of labelled data and topic distributions of hidden topics that were generated from that data. We explored how the representation based on topics affect the classification performance by applying supervised classification algorithms. Additionally, we did careful evaluation on two types of datasets and compared them with baseline approaches without topic distributions and other comparable methods. The results show that our framework performs significantly better in terms of classification performance compared to the baseline(without T2F) approaches and also yields improvement in terms of F1 score compared to other compared approaches.