Automatic Segmentation of the Left Atrium from Computed Tomography Angiography Images.

The left atrial appendage (LAA) causes 91% of thrombi in atrial fibrillation patients, a potential harbinger of stroke. Leveraging computed tomography angiography (CTA) images, radiologists interpret the left atrium (LA) and LAA geometries to stratify stroke risk. Nevertheless, accurate LA segmentation remains a time-consuming task with high inter-observer variability. Binary masks of the LA and their corresponding CTA images were used to train and test a 3D U-Net to automate LA segmentation. One model was trained using the entire unified-image-volume while a second model was trained on regional patch-volumes which were run for inference and then assimilated back into the full volume. The unified-image-volume U-Net achieved median DSCs of 0.92 and 0.88 for the train and test sets, respectively; the patch-volume U-Net achieved median DSCs of 0.90 and 0.89 for the train and test sets, respectively. This indicates that the unified-image-volume and patch-volume U-Net models captured up to 88 and 89% of the LA/LAA boundary's regional complexity, respectively. Additionally, the results indicate that the LA/LAA were fully captured in most of the predicted segmentations. By automating the segmentation process, our deep learning model can expedite LA/LAA shape, informing stratification of stroke risk.

Modeling Cases and Deaths per Million using Daily-Aggregated Facebook COVID-19 Symptom Survey Data.

We develop a novel analytic approach to modeling future COVID-19 risk using COVID-19 Symptom Survey data aggregated daily by US state, joined with daily time-series data on confirmed cases and deaths. Specifically, we model N-day forward-looking estimates for per-US-state-per-day change in deaths per million (DPM) and cases per million (CPM) using a multivariate regression model to below baseline error (65% and 38% mean absolute percentage error for DPM/CPM, respectively). Additionally, we model future changes in the curvature of CPM/DPM as "increasing" or "decreasing" using a random forest classifier to above 72% accuracy. In sum, we develop and characterize models to establish a relationship between behaviors and beliefs of individuals captured via the Facebook COVID-19 Symptom Surveys and the trajectory of COVID-19 outbreaks evidenced in terms of CPM and DPM. Such information can be helpful in assessing collective risks of infection and death during a pandemic as well as in determining the effectiveness of appropriate risk mitigation strategies based on behaviors evidenced through survey responses.