Abstract
Heart failure (HF) is a multifactorial chronic syndrome with progressive increasing incidence causing a huge financial burden worldwide. Remote monitoring should, in theory, improve HF management, but given increasing morbidity and mortality, a question remains: are we monitoring it properly? Device-based home monitoring enables objective and continuous measurement of vital variables and non-invasive devices should be first choice for elderly patients. There is no shortage of literature on the subject, however, most studies were designed to monitor a single variable or class of variables that were not properly assembled and, to the best of our knowledge, there are no large randomized studies about their impact on HF patient management. To overcome this problem, we carefully selected the most critical possible HF decompensating factors to design MONITORIA, a non-invasive device for comprehensive HF home monitoring. MONITORIA stands for MOnitoring Non-Invasively To Overcome mortality Rates of heart Insufficiency on Ambulatory, and in this paper, which is part I of a series of three articles, we discuss the theoretical basis for its design. MONITORIA and its inherent follow-up strategy will optimize HF patient care as it is a promising device, which will essentially adapt innovation not to the disease but rather to the patients. (C) 2020 Sociedade Portuguesa de Cardiologia. Published by Elsevier Espana, S.L.U.

Abstract
Introduction: Heart failure (HF) represents a huge financial and economic burden worldwide. Some authors advocate that remote monitoring should be implemented to improve HF management, but given its increasing incidence, as well as its morbidity and mortality, a question still remains: are we monitoring it properly? There is no shortage of literature on home monitoring devices, however, most of them are designed to monitor an unsuitable array of variables and, to the best of our knowledge, there are no large randomized studies about their impact on morbidity/mortality of HF patients. Objective: Description of a novel monitoring device. Methods: As a solution, we designed MONITORIA (MOnitoring NonInvasively To Overcome mortality Rates of heart Insufficiency on Ambulatory). Results: This is a multimodal device that will provide real time monitoring of vital, electrophysiological, hemodynamic and chemical signs, transthoracic impedance, and physical activity levels. The device is meant to perform continuous analysis and transmission of all data. Significant alterations in a patient's variable will alert the attending physician and, in case of potentially life-threatening situations, the national emergency medical system. The MONITORIA device will, also, have a function that sends shocks or functions as a pacemaker to treat certain arrhythmias/blockades. This function can be activated the very first time the patient utilizes it, based on their risk of sudden cardiac death. Discussion/Conclusions: MONITORIA is a promising device mostly because it is included in a follow-up program that takes into account a multi-perspective feature of HF development and is based on the real world patient, adapting innovations not to the disease but rather to the patients. (C) 2021 Sociedade Portuguesa de Cardiologia. Published by Elsevier Espana, S.L.U.

Abstract
Ambient assisted living (AAL) environments represent a key concept for dealing with the inevitable problem of population-ageing. Until recently, the use of computational intensive techniques, like Machine Learning (ML) or Computer Vision (CV), were not suitable for IoT end-nodes due to their limited resources. However, recent advances in edge intelligence have somehow changed this landscape for smart environments. This paper presents an AAL scenario where the use of ML is tested in kitchen appliances recognition using CV. The goal is to help users working with those appliances through Augmented Reality (AR) on a mobile device. Seven types of kitchen appliances were selected: blender, coffee machine, fridge, water kettle, microwave, stove, and toaster. A dataset with nearly 4900 images was organized. Three different deep learning (DL) models from the literature were selected, each with a total number of parameters and architecture compatibles with their execution on mobile devices. The results obtained in the training of these models reveal precision in the test set above 95% for the model with better results. The combination of edge AI and ML opens the application of CV in smart homes and AAL without compromising mandatory requirements as system privacy or latency. © 2021, The Author(s), under exclusive license to Springer Nature Switzerland AG.

Abstract
Psoriasis is a chronic inflammatory skin disease with a high worldwide incidence that in worst cases reaches 4.6%. This dermatosis can be associated with other comorbidities and has a significant negative impact on labor productivity and the quality of life of affected people. During day-to-day lives, psoriasis patients come across several practical clinical difficulties, e.g. to i) easily register a time evolution of affected skin areas (for later analysis by health carers); ii) daily evaluate the size of each affected skin area, to be able to iii) calculate the amount of medication to be applied on those affected body areas. In such a context, this paper proposes the Follow-App mobile system aiming to support people with psoriasis, by alleviating and managing their daily life with the disease. More precisely, the goals of the system are: to allow individual photographic registration of body parts affected by psoriasis; in addition, cataloging each image according to its body segment location and sampling date; then, on those photos, automatically detect and segment the affected skin surface, to posteriorly be able to calculate the area of the lesions; finally, based on the area and prescribed medicine, dynamically accounting the amount of topical medicine to use. These were the requirements addressed by the proposed system prototype. The evaluation tests on the ability to detect and quantify the area of the skin lesions were performed on a data-set with 22 images. The proposed segmentation algorithm for detecting the area of redness lesions reached an IoU rate over 81%. Therefore, the proposed Follow-App mobile system may become an important asset for people with psoriasis since the extent and redness of affected areas are major evaluation factors for the disease severity. © 2021, The Author(s), under exclusive license to Springer Nature Switzerland AG.

Abstract
This project proposes a machine learning (ML) pipeline for inferring office employee's well-being, from heterogeneous sources of contextual data (cf. physiological, social and workplace environment), which brings several demanding issues. In this paper we focus specifically in raw data collection problems and pre-processing challenges. To start with, context data was collected in real environments, during weeks, in several office organizations and involving employees along theirs daily working routines. Moreover, data collection resort to a wide range of sources (e.g. sensors, questionnaires, apps, etc.) that were subject to potential interferences and noisy conditions. Given the influence of data quality in ML algorithms results and considering the number of instruments used, it was essential to implement a pre-processing stage to automate and improve the quality of collected data. Hence, the usefulness of the proposed DQVA tool, which computes several common statistical measures and provides also graphical and tabular visual insights about the data. For example, it allows to: i) compare data sources from different participants and organizations, on a per sensor/data source basis (through data tables, data distribution histograms, and visualizations); iii) check and pinspot the existence of outliers; iv) visually spot signal gaps; etc. Therefore, we argue that the proposed DQVA tool allows to evaluate, per sensor and per individual, raw data quality, on the integration stage of our classification pipeline. It proved to be an agile, useful and simple to re-use tool for detecting raw data irregularities, thus increasing data quality assurances for the next steps of our classification pipeline.

Abstract
Resistive switching behaviour has been demonstrated to be a common characteristic to many materials. In this regard, research teams to date have produced a plethora of different devices exhibiting diverse behaviour, but when system design is considered, finding a 'one-model-fits-all' solution can be quite difficult, or even impossible. However, it is in the interest of the community to achieve more general modelling tools for design that allows a quick model update as devices evolve. Laying the grounds with such a principle, this paper presents an artificial neural network learning approach to resistive switching modelling. The efficacy of the method is demonstrated firstly with two simulated devices and secondly with a 4 mu m(2) amorphous IGZO device. For the amorphous IGZO device, a normalized root-mean-squared error (NRMSE) of 5.66 x 10(-3) is achieved with a [2, 50,50 ,1] network structure, representing a good balance between model complexity and accuracy. A brief study on the number of hidden layers and neurons and its effect on network performance is also conducted with the best NRMSE reported at 4.63 x 10(-3). The low error rate achieved in both simulated and real-world devices is a good indicator that the presented approach is flexible and can suit multiple device types.