The definition of anomaly detection is the identification of an unexpected event. Real-time detection of extreme events such as wildfires, cyclones, or floods using satellite data has become crucial for disaster management. Although several earth-observing satellites provide information about disasters, satellites in the geostationary orbit provide data at intervals as frequent as every minute, effectively creating a video from space. There are many techniques that have been proposed to identify anomalies in surveillance videos; however, the available datasets do not have dynamic behavior, so we discuss an anomaly framework that can work on very high-frequency datasets to find very fast-moving anomalies. In this work, we present a diffusion model which does not need any motion component to capture the fast-moving anomalies and outperforms the other baseline methods.
Purpose: To develop an algorithm for automated medical imaging order selection based on provider-input signs and symptoms using natural language processing and machine learning. The aim is to reduce the frequency of inappropriate physician imaging orders, which currently accounts for 25.7% of cases, and thereby mitigate potential patient health concerns. Materials and Methods: The study was conducted retrospectively with a four-step analysis process. The data used for training in the study consisted of anonymized imaging records and associated provider-input symptoms for CT and MRI orders in 40,667 patients from a tertiary children’s hospital. First, the data were normalized using keyword filtering and lemmatization. Second, an entity-embedding ML model converted the symptoms to high-dimensional numerical vectors suitable for model comprehension, which we used to balance the dataset through k-nearest-neighbor-based synthetic sampling. Third, a Support Vector Classifier (ML model) was trained and hyperparameter-tuned using the embedded symptoms to predict modality (CT/MRI), contrast (with/without), and anatomical region (head, neck, etc.) for the imaging orders. Finally, a web application was developed to package the model, which analyzes user-input symptoms and outputs the predicted order. Results: The model was found to have a final overall accuracy of 93.2% on a 4,704-case test set ( p < 0.001). The AUCs for the eight classes ranged from 96% to 100%, and the average F1-score was 0.92. Conclusion: This algorithm looks to act as a clinical decision support tool to help augment the present physician imaging order selection accuracy and improve patient health.