ASTRO: Reducing COVID-19 Exposure through Contact Prediction and Avoidance

The fight against the COVID-19 pandemic has highlighted the importance and benefits of recommending paths that reduce the exposure to and the spread of the SARS-CoV-2 coronavirus by avoiding crowded indoor or outdoor areas. Existing path discovery techniques are inadequate for coping with such dynamic and heterogeneous (indoor and outdoor) environments—they typically find an optimal path assuming a homogeneous and/or static graph, and hence they cannot be used to support contact avoidance. In this article, we pose the need for Mobile Contact Avoidance Navigation and propose ASTRO ( A ccessible S patio- T emporal R oute O ptimization), a novel graph-based path discovering algorithm that can reduce the risk of COVID-19 exposure by taking into consideration the congestion in indoor spaces. ASTRO operates in an A * manner to find the most promising path for safe movement within and across multiple buildings without constructing the full graph. For its path finding, ASTRO requires predicting congestion in corridors and hallways. Consequently, we propose a new grid-based partitioning scheme combined with a hash-based two-level structure to store congestion models, called CM-Structure , which enables on-the-fly forecasting of congestion in corridors and hallways. We demonstrate the effectiveness of ASTRO and the accuracy of CM-Structure ’s congestion models empirically with realistic datasets, showing up to one order of magnitude reduction in COVID-19 exposure.

TB-NET: A Two-Branch Neural Network for Direction of Arrival Estimation under Model Imperfections

For direction of arrival (DoA) estimation, the data-driven deep-learning method has an advantage over the model-based methods since it is more robust against model imperfections. Conventionally, networks are based singly on regression or classification and may lead to unstable training and limited resolution. Alternatively, this paper proposes a two-branch neural network (TB-Net) that combines classification and regression in parallel. The grid-based classification branch is optimized by binary cross-entropy (BCE) loss and provides a mask that indicates the existence of the DoAs at predefined grids. The regression branch refines the DoA estimates by predicting the deviations from the grids. At the output layer, the outputs of the two branches are combined to obtain final DoA estimates. To achieve a lightweight model, only convolutional layers are used in the proposed TB-Net. The simulation results demonstrated that compared with the model-based and existing deep-learning methods, the proposed method can achieve higher DoA estimation accuracy in the presence of model imperfections and only has a size of 1.8 MB.

Data Checkers: A Grid-Based UI for Managing Patient-Generated Data Sharing to Support Collaborative Self-Care

Chronic health conditions are becoming increasingly prevalent. As part of chronic care, sharing patient-generated health data (PGHD) is likely to play a prominent role. Sharing PGHD is increasingly recognized as potentially useful for not only monitoring health conditions but for informing and supporting collaboration with caregivers and healthcare providers. In this paper, we describe a new design for the fine-grained control over sharing one's PGHD to support collaborative self-care, one that centers on giving people with health conditions control over their own data. The system, Data Checkers (DC), uses a grid-based interface and a preview feature to provide users with the ability to control data access and dissemination. DC is of particular use in the case of severe chronic conditions, such as spinal cord injuries and disorders (SCI/D), that require not just intermittent involvement of healthcare providers but daily support and assistance from caregivers. In this paper, after providing relevant background information, we articulate our steps for developing this innovative system for sharing PGHD including (a) use of a co-design process; (b) identification of design requirements; and (c) creation of the DC System. We then present a qualitative evaluation of DC to show how DC satisfied these design requirements in a way that provided advantages for care. Our work extends existing research in the areas of Human-Computer Interaction (HCI), Computer-Supported Cooperative Work (CSCW), Ubiquitous Computing (Ubicomp), and Health Informatics about sharing data and PGHD.

Pseudo-prospective testing of 5-year earthquake forecasts for California using inlabru

Probabilistic earthquake forecasts estimate the likelihood of future earthquakes within a specified time-space-magnitude window and are important because they inform planning of hazard mitigation activities on different timescales. The spatial component of such forecasts, expressed as seismicity models, generally rely upon some combination of past event locations and underlying factors which might affect spatial intensity, such as strain rate, fault location and slip rate or past seismicity. For the first time, we extend previously reported spatial seismicity models, generated using the open source inlabru package, to time-independent earthquake forecasts using California as a case study. The inlabru approach allows the rapid evaluation of point process models which integrate different spatial datasets. We explore how well various candidate forecasts perform compared to observed activity over three contiguous five year time periods using the same training window for the seismicity data. In each case we compare models constructed from both full and declustered earthquake catalogues. In doing this, we compare the use of synthetic catalogue forecasts to the more widely-used grid-based approach of previous forecast testing experiments. The simulated-catalogue approach uses the full model posteriors to create Bayesian earthquake forecasts. We show that simulated-catalogue based forecasts perform better than the grid-based equivalents due to (a) their ability to capture more uncertainty in the model components and (b) the associated relaxation of the Poisson assumption in testing. We demonstrate that the inlabru models perform well overall over various time periods, and hence that independent data such as fault slip rates can improve forecasting power on the time scales examined. Together, these findings represent a significant improvement in earthquake forecasting is possible, though this has yet to be tested and proven in true prospective mode.

How deep is your model? Network topology selection from a model validation perspective

Deep learning is a powerful tool, which is becoming increasingly popular in financial modeling. However, model validation requirements such as SR 11-7 pose a significant obstacle to the deployment of neural networks in a bank’s production system. Their typically high number of (hyper-)parameters poses a particular challenge to model selection, benchmarking and documentation. We present a simple grid based method together with an open source implementation and show how this pragmatically satisfies model validation requirements. We illustrate the method by learning the option pricing formula in the Black–Scholes and the Heston model.