The heavy use of machine learning algorithms in safety-critical systems poses serious questions related to safety, security, and predictability issues, requiring novel architectural approaches to guarantee such properties. This paper presents an architecture solution that leverages heterogeneous platforms and virtualization technologies to support AI-powered applications consisting of modules with mixed criticalities and safety requirements. The hypervisor exploits the security features of the Xilinx ZCU104 MPSoCs to create two isolated execution environments: a high performance domain running deep learning algorithms under the Linux operating system and a safety-critical domain running control and monitoring functions under the freeRTOS real-time operating system. The proposed approach is validated by a use case consisting of an unmanned aerial vehicle capable of tracking moving targets using a deep neural network accelerated on the FGPA available on the platform.

Transthyretin related cardiac amyloidosis (TTR-CA) is an infiltrative cardiomyopathy that cause heart failure with preserved ejection fraction, mainly in aging people. Due to the introduction of a non invasive diagnostic algorithm, this disease, previously considered to be rare, is increasingly recognized. The natural history of TTR-CA includes two different stages: a presymptomatic and a symptomatic stage. Due to the availability of new disease-modifying therapies, the need to reach a diagnosis in the first stage has become impelling. While in variant TTR-CA an early identification of the disease may be obtained with a genetic screening in proband's relatives, in the wild-type form it represents a challenging issue. Once the diagnosis has been made, in order to identifying patients with a higher risk of cardiovascular events and death it is necessary to focus on risk stratification. Two prognostic scores have been proposed both based on biomarkers and laboratory findings. However, a multiparametric approach combining information from electrocardiogram, echocardiogram, cardiopulmonary exercise test and cardiac magnetic resonance may be warranted for a more comprehensive risk prediction. In this review, we aim at evaluating a step by step risk stratification, providing a clinical diagnostic and prognostic approach for the management of patients with TTR-CA.