deep learning

Medical postgraduate trainees are given increasing levels of responsibility during training in the apprenticeship-model of training [1-3]. Responsibility is said to be a key driver of deep learning and understanding [4-7]. Trainees with greater levels of responsibility for decision making have higher levels of motivation to learn compared with trainees who self-assess as having less autonomy [5]. The Royal College of Pathologists (RCPath) indicate that ‘graded responsibility’ is part of training histopathologists and provided a framework for implementation with increasingly complex specimens suitable for reporting by more senior trainees [8,9]. 

Clinical applications of Artificial Intelligence (AI) in healthcare are relatively rare. The high expectations in relation to data analysis influencing general healthcare have not materialized, with few exceptions, and then predominantly in the field of rare diseases, oncology and pathology, and interpretation of laboratory results. While electronic health records, introduced over the last decade or so in the UK have increased access to medical and treatment histories of patients, diagnoses, medications, treatment plans, immunization dates, allergies, radiology images, laboratory and test results, these have potential for evidence-based tools that providers can use to make decisions about a patient’s care, as well as streamline workflow. In the following text, we review the advances achieved using machine learning and deep learning technology, as well as robot use and telemedicine in the healthcare of older people. Key points: 1. Artificial Intelligence use is extensively explored in prevention, diagnosis, novel drug designs and after-care. 2. AI studies on older adults include a small number of patients and lack reproducibility needed for their wider clinical use in different clinical settings and larger populations. 3. Telemedicine and robot assisted technology are well received by older service users. 4. Ethical concerns need to be resolved prior to wider AI use in routine clinical setting.

The SARS-CoV-2 (COVID-19) pandemic outbreak has led to some lockdowns and changed human mobility and lifestyle in this country. Mashhad, one of the most polluted cities in Iran has experienced critical air pollution conditions in recent years. In the present study, the potential relationships between air quality conditions (such as popular index and criteria air pollutant concentration) and COVID-19 cases and deaths were investigated in Mashhad, Iran. To do that, the Long Short-Term Memory (LSTM) based hybrid deep learning architecture was implemented on AQI, meteorological data (such as temperature, sea level pressure, dew points, and wind speed), traffic index and impact number of death, and active cases COVID-19 from March 2019 to March 2022 in Mashhad. The results reveal the LSTM model could predict the AQI accurately. The lower error between the real and predicted AQI, including MSE, MSLE, and MAE is 0.0153, 0.0058, and 0.1043, respectively. Also, the cosine similarity between predicted AQI and real amounts of it is 1. Moreover, in the first peak of the pandemic (Aug 2021), we have the minimum amount of AQI. Meanwhile, by increasing the number of active cases and death and by starting lockdown, because the traffic is decreased, the air quality is good and the amount of AQI related to PM2.5 is 54.68. Furthermore, the decrease the active cases and death in pandemic causes a significant increase in AQI, which is 123.52 in Nov 2021, due to a decline in lockdowns, resumption of human activities, and probable temperature inversions. 

Cancer prognosis reflects a complex biological process measured by multiple types of omics data. Deep learning frameworks have been proposed to integrate multi-omics data and predict patient outcomes in different cancer types, potentially revolutionizing cancer prognosis with superior performance. This minireview summarizes the advances in the strategies for multi-omics data integration and the performance of different deep learning models in prognosis prediction of diverse cancer types using multi-omics data published in the past 18 months. The challenges and limitations of deep learning models for predicting cancer outcomes based on multi-omics data are discussed.

Coronavirus disease (COVID-19) is a viral pneumonia that is found in China and has spread globally. Early diagnosis is important for effective and timely treatment. Thus, many ongoing studies attempt to solve key COVID-19 problems such as workload classification, detection, and differentiation from other pneumonia and healthy lungs using different imaging modalities. Researchers have identified some limitations in the deployment of deep learning methods to detect COVID-19, but there are still unmet challenges to be addressed. The use of binary classifiers or building classifiers based on only a few classes is some of the limitations that most of the existing research on the COVID-19 classification problem suffers from. Additionally, most prior studies have focused on model or ensemble models that depend on a flat single-feature imaging modality without using any clinical information or benefiting from the hierarchical structure of pneumonia, which leads to clinical challenges, and evaluated their systems using a small public dataset. Additionally, reliance on diagnostic processes based on CT as the main imaging modality, ignoring chest X-rays. Radiologists, computer scientists, and physicians all need to come to an understanding of these interdisciplinary issues. This article first highlights the challenges of deep learning deployment for COVID-19 detection using a literature review and document analysis. Second, it provides six key recommendations that could assist future researchers in this field in improving the diagnostic process for COVID-19. However, there is a need for a collective effort from all of them to consider the provided recommendations to effectively solve these issues.

Microbiome-gut-brain axis represents a complex, bidirectional communication network connecting the gastrointestinal tract and its microbial populations with the central nervous system (CNS). This complex system is important for maintaining physiological homeostasis and has significant implications for mental health. The human gut has trillions of microorganisms, collectively termed gut microbiota, which play important roles in digestion, immune function, and production of various metabolites. Some current research shows that these microorganisms strongly influence the brain function and behaviour of individuals, forming the basis of the microbiome-gut-brain axis. The communication between gut microbiota and the brain occurs via multiple pathways: neural pathway (e.g., vagus nerve), endocrine pathway (e.g., hormone production), immune pathway (e.g., inflammation modulation), and metabolic pathway (e.g., production of short-chain fatty acids). Dysbiosis, or imbalance of gut microbiota, has been linked to mental health disorders such as anxiety, depression, multiple sclerosis, autism spectrum disorders, etc, offering new perspectives on their etiology and potential therapeutic interventions. Artificial Intelligence (AI) has emerged as a powerful tool in interpreting the complexities of the microbiome-gut-brain axis. AI techniques, such as machine learning and deep learning, enable the integration and analysis of large, multifaceted datasets, uncovering patterns and correlations that can be avoided by traditional methods. These techniques enable predictive modeling, biomarker discovery, and understanding of underlying biological mechanisms, enhancing research efficiency and covering ways for personalized therapeutic approaches. The application of AI in microbiome research has provided valuable insights into mental health conditions. AI models have identified specific gut bacteria linked to disease, offered predictive models, and discovered distinct microbiome signatures associated with specific diseases. Integrating AI with microbiome research holds promise for revolutionizing mental health care, offering new diagnostic tools and targeted therapies. Challenges remain, but the potential benefits of AI-driven insights into microbiome-gut-brain interactions are immense and offer hope for innovative treatments and preventative measures to improve mental health outcomes.

Introduction: Pneumothorax is a life-threatening condition that requires prompt recognition and therapy to prevent deterioration. Radiologist workload often precludes rapid assessment of the usual diagnostic modality, the chest radiograph, particularly after hours. The aim was to develop a deep learning model using a segmentation-based Deep Convolutional Neural Network (DCNN) to detect pneumothorax on chest radiographs to provide rapid and accurate pneumothorax diagnosis.Methods: This is a retrospective study of spontaneous pneumothorax at a single center, containing 130 positive and 70 negative radiographs. Subsequent manual contour mapping was performed to draw a mask of the pneumothorax. These image pairs were used to train a DCNN model (a modified AlexNet) after pretraining on the ImageNet dataset.Results: The DCNN achieved an accuracy of 0.83, with sensitivity of 98.1%, and specificity of 68.5%.Conclusion: This segmentation-based DCNN accuracy is comparable to previous categorization-based CDNN models, despite using a smaller sample size for training, while including the benefits of visual representation for clinician feedback. Segmentation-based DCNNs show promise in the development of accurate and clinically useful models for medical imaging.

A significant obstacle to agricultural productivity that jeopardizes the availability of food is crop diseases and farmer livelihoods by reducing crop yields. Traditional visual assessment methods for disease diagnosis are effective but complex, often requiring expert observers. Recent advancements in deep learning indicate the potential for increasing accuracy and automating disease identification. Developing accessible diagnostic tools, such as web applications leveraging CNNs, can provide farmers with efficient and accurate disease identification, especially in regions with limited access to advanced diagnostic technologies. The main goal is to develop a productive system that can recognize tomato plant diseases. The model was trained on a collection of images of healthy and damaged tomato leaves from PlantVillage using transfer learning techniques. The images from the dataset were cleansed by resizing them from 256 × 256 to 224 × 224 to match the dimensions used in pre-trained models using min-max normalization. An evaluation of VGG16, VGG19, and DenseNet121 models based on performance accuracy and loss value for 7 categories of tomatoes guided the selection of the most effective model for practical application. VGG16 achieved 84.54% accuracy, VGG19 achieved 84.62%, and DenseNet121 achieved 98.28%, making DenseNet121 the chosen model due to its highest performance accuracy. The web application development based on the DenseNet121 architecture was integrated using the Django web framework, which is built on Python. This enables real-time disease diagnosis for uploaded images of tomato leaves. The proposed system allows early detection and diagnosis of tomato plant diseases, helping to mitigate crop losses. This supports sustainable farming practices and increases agricultural productivity.

Cardiovascular Diseases (CVDs) remain a major global health concern, necessitating accurate and comprehensive diagnostic techniques. Traditional medical imaging modalities, such as CT angiography, PET, MRI, and ultrasound, provide crucial but limited information when used independently. Image fusion techniques integrate complementary modalities, enhance visualization, and improve diagnostic accuracy. This paper presents a theoretical study of advanced image fusion methods applied to cardiovascular imaging. We explore wavelet-based, Principal Component Analysis (PCA), and deep learning-driven fusion models, emphasizing their theoretical underpinnings, mathematical formulation, and potential clinical applications. The proposed framework enables improved coronary artery visualization, cardiac function assessment, and real-time hemodynamic analysis, offering a non-invasive and highly effective approach to cardiovascular diagnostics.MSC Codes: 68U10,94A08,92C55,65T60,62H25,68T07.

The integration of deep learning and genetic analysis has transformed the assessment of elite sports performance, particularly in competitive swimming. This study examines the fusion of deep learning techniques with DNA markers, physiological biometrics, and performance analytics to enhance the prediction and optimization of swimmer performance. A structured dataset comprising genetic sequences, physiological parameters, and biomechanical attributes was utilized to train a neural network model capable of categorizing swimmers based on genetic predisposition and athletic potential. The model achieved high classification accuracy, demonstrating a strong link between genetic markers, physiological traits, and competitive swimming outcomes. The findings emphasize the potential of AI-driven analytics in talent identification, customized training adaptations, and injury prevention. Furthermore, the study highlights the effectiveness of deep learning in analyzing complex genomic and physiological data to generate meaningful insights for performance enhancement. While the results validate the feasibility of using genetic and AI-based models for performance prediction, further studies are needed to broaden dataset diversity, integrate epigenetic influences, and test the model across varied athlete populations. This research contributes to the expanding field of AI-driven sports science and provides a solid foundation for incorporating genomics with deep learning to enhance elite athletic performance.

This paper shows the developments and directions in feature processing. We begin by revisiting conventional feature processing methods, then focus on deep feature extraction techniques and the application of feature processing. The article also analyzes the current research challenges and outlines future development directions, providing valuable insights in related fields.