Search

Objective Diabetes mellitus is an important chronic disease that is prevalent around the world. Different countries and diverse cultures use varying approaches to dealing with this chronic condition. Also, with the advancement of computation and automated decision-making, many tools and technologies are now available to patients suffering from this disease. In this work, the investigators attempt to analyze approaches taken towards managing this illness in India and the United States. Methods In this work, the investigators have used available literature and data to compare the use of artificial intelligence in diabetes management. Findings The article provides key insights to comparison of diabetes management in terms of the nature of the healthcare system, availability, electronic health records, cultural factors, data privacy, affordability, and other important variables. Interestingly, variables such as quality of electronic health records, and cultural factors are key impediments in implementing an efficiency-driven management system for dealing with this chronic disease. Conclusion The article adds to the body of knowledge associated with the management of this disease, establishing a critical need for using artificial intelligence in diabetes care management.

Retinal imaging is crucial for observing the retina and accurately diagnosing pathological problems. Optical Coherence Tomography (OCT) has been a transformative breakthrough for developing high-resolution cross-sectional images. It is imperative to delineate the multiple layers of the retina for a proper diagnosis. A novel segmentation-based approach is introduced in this study to identify seven distinct layers of the retina using OCT images. The developed approach presents SliceOCTNet, a customized U-shaped Convolutional Neural Network (CNN) that introduces group normalization and intricate skip connections. Paired alongside a hybrid loss function, the SliceOCTNet outperformed most state-of-the-art approaches. The introduction of Group Normalization in SliceOCTNet stabilized the model and improved layer identification even when working with small datasets. The use of skip connections also contributed to an improvement in the spatial outlook of the model. Implementing a hybrid loss function addresses the class imbalance problem in the dataset. Duke University’s spectral-domain optical coherence tomography (SD-OCT) B-scan dataset of healthy and Diabetic Macular Edema (DME) afflicted patients was utilized to train and evaluate the SliceOCTNet. The model accurately recognizes the seven layers of the retina. It can achieve a high dice coefficient value of 0.941 and refine the segmentation process to a higher level of precision.

The complexity and high dimensionality of healthcare data present substantial challenges in building machine learning (ML) models, given the large number of variables such as patient demographics and medical history. Effective feature selection is crucial to address issues such as increased computational resource, longer training times, overfitting, and reduced model interpretability, etc. This study evaluates a range of feature selection methods to identify the most impactful features for predicting dental expenditures using publicly available Medical Expenditure Panel Survey (MEPS) data. Sixteen ML models are assessed to determine the top performing model, after which state-of-the-art filter, wrapper, embedded, and hybrid feature selection techniques are applied to optimize the feature set. The highest performance, in terms of coefficient of determination ( $R^{2}$ ), is achieved using a hybrid feature selection method that combines the mutual information filter with the embedded features from the CatBoost regressor. The results indicate that the proposed system is suitable for real-time deployment even with reduced features, providing potential benefits such as minimizing the need for irrelevant and difficult-to-obtain features. Moreover, automated feature selection significantly enhances model performance, yielding a $R^{2}$ score of 0.86, compared to the score 0.73 achieved with carefully selected manual features. Additionally, to enhance the interpretability of the top-performing ML model, explanatory visualizations are employed to examine the influence of key features on predicting dental expenditures.

Fundus Fluorescein Angiography (FFA) is an important invasive diagnostic tool employed in ophthalmology for the assessment of retinal and choroidal blood flow, components situated at the rear of the eye. Identifying lesions within FFA images, indicative of anomalies in the retina and choroid visualized through the injection of a fluorescein dye, is a critical step in diagnosing and managing chronic ocular diseases. Such lesions often hint at a range of eye disorders, including but not limited to diabetic retinopathy, age-related macular degeneration, and retinal vein occlusions. In conventional practice, these images are interpreted manually by a skilled ophthalmologist. Nonetheless, due to the intricate nature of these images, the manual process is often time-consuming and subject to individual interpretation. The advent of advanced technologies has instigated the development of automated lesion detection techniques, thereby alleviating these constraints.In this study, we introduce FFA-Lens, a web-based application engineered to detect 25 prevalent lesions in FFA images, serving as a crucial resource in diagnosing and treating chronic ocular diseases. Notably, FFA-Lens has demonstrated remarkable precision and recall values exceeding 0.8 in lesion detection, emphasizing its significant potential. By enabling early detection of ocular conditions, automating lesion identification, and enhancing diagnostic efficiency, FFA-Lens can substantially improve clinical disease management strategies.

Background Dental radiology has significantly benefited from cone-beam computed tomography (CBCT) because of its compact size and low radiation exposure. Canal tracking is an important application of CBCT for determining the relationship between the inferior alveolar nerve and third molar. Usually, canal tacking is performed manually, which takes a lot of time. This study aimed to develop an artificial intelligence (AI) model to automate classification of the mandibular canal in relation to the third molar. Methods This retrospective study was conducted using 434 CBCT images. 3D slicer software was used to annotate and classify the data into lingual, buccal, and inferior categories. Two convolution neural network models, AlexNet and ResNet50, were developed to classify this relationship. The study included 262 images for training and 172 images for testing, with the model performance evaluated by sensitivity, precision, and F1 score. Results The performance of the two models was evaluated using a 3 × 3 confusion matrix, with the data categorized into 3 clases: lingual, buccal, and inferior. The mandibular canal and third molar have a close anatomical relationship, highlighting the need for precise imaging in dental and surgical settings. To accurately classify the mandibular canal in relation to the third molar, both AlexNet and ResNet50 demonstrated high accuracy, with F1 scores ranging from 0.64 to 0.92 for different classes, with accuracy of 81% and 83%, respectively, for accurately classifying the mandibular canal in relation to the third molar. Conclusion The present study developed and evaluated AI models to accurately classify and establish the relationship between the mandibular canal and third molars using CBCT images with a higher accuracy rate.