ABSTRACT
Background: In December 2020, the COVID-19 disease was confirmed in 1,665,775 patients and caused 45,784 deaths in Spain. At that time, health decision support systems were identified as crucial against the pandemic. Methods: This study applies Deep Learning techniques for mortality prediction of COVID-19 patients. Two datasets with clinical information were used. They included 2,307 and 3,870 COVID-19 infected patients admitted to two Spanish hospitals. Firstly, we built a sequence of temporal events gathering all the clinical information for each patient, comparing different data representation methods. Next, we used the sequences to train a Recurrent Neural Network (RNN) model with an attention mechanism exploring interpretability. We conducted an extensive hyperparameter search and cross-validation. Finally, we ensembled the resulting RNNs to enhance sensitivity. Results: We assessed the performance of our models by averaging the performance across all the days in the sequences. Additionally, we evaluated day-by-day predictions starting from both the hospital admission day and the outcome day. We compared our models with two strong baselines, Support Vector Classifier and Random Forest, and in all cases our models were superior. Furthermore, we implemented an ensemble model that substantially increased the system's sensitivity while producing more stable predictions. Conclusions: We have shown the feasibility of our approach to predicting the clinical outcome of patients. The result is an RNN-based model that can support decision-making in healthcare systems aiming at interpretability. The system is robust enough to deal with real-world data and can overcome the problems derived from the sparsity and heterogeneity of data.
ABSTRACT
The fast growth in the amount of connected devices with computing capabilities in the past years has enabled the emergence of a new computing layer at the Edge. Despite being resource-constrained if compared with cloud servers, they offer lower latencies than those achievable by Cloud computing. The combination of both Cloud and Edge computing paradigms can provide a suitable infrastructure for complex applications' quality of service requirements that cannot easily be achieved with either of these paradigms alone. These requirements can be very different for each application, from achieving time sensitivity or assuring data privacy to storing and processing large amounts of data. Therefore, orchestrating these applications in the Cloud-Edge computing raises new challenges that need to be solved in order to fully take advantage of this layered infrastructure. This paper proposes an architecture that enables the dynamic orchestration of applications in the Cloud-Edge continuum. It focuses on the application's quality of service by providing the scheduler with input that is commonly used by modern scheduling algorithms. The architecture uses a distributed scheduling approach that can be customized in a per-application basis, which ensures that it can scale properly even in setups with high number of nodes and complex scheduling algorithms. This architecture has been implemented on top of Kubernetes and evaluated in order to asses its viability to enable more complex scheduling algorithms that take into account the quality of service of applications.
Subject(s)
Algorithms , Cloud ComputingABSTRACT
eHealth systems have adopted recent advances on sensing technologies together with advances in information and communication technologies (ICT) in order to provide people-centered services that improve the quality of life of an increasingly elderly population. As these eHealth services are founded on the acquisition and processing of sensitive data (e.g., personal details, diagnosis, treatments and medical history), any security threat would damage the public's confidence in them. This paper proposes a solution for the design and runtime management of indoor eHealth applications with security requirements. The proposal allows applications definition customized to patient particularities, including the early detection of health deterioration and suitable reaction (events) as well as security needs. At runtime, security support is twofold. A secured component-based platform supervises applications execution and provides events management, whilst the security of the communications among application components is also guaranteed. Additionally, the proposed event management scheme adopts the fog computing paradigm to enable local event related data storage and processing, thus saving communication bandwidth when communicating with the cloud. As a proof of concept, this proposal has been validated through the monitoring of the health status in diabetic patients at a nursing home.
