Abstract
This paper presents a systematic mapping review of the literature on innovative digital solutions to detect cognitive impairment of community-dwelling older adults. Seventy-six articles were included in this mapping review. Most of the included articles (i.e., 65 articles) reported the implementation and validation of computerized versions of paper-based neuropsychological tests. In turn, 11 studies are related to the application of emerging technologies to detect cognitive decline, including serious games, virtual reality, and data analytics (e.g., algorithms to analyse data from ubiquitous daily activity and interaction sensing) approaches. From these 11 studies, four include experimental setups to determine if the developed digital solutions can discriminate cognitive impairments. Based on the mapping review findings is possible to conclude that further research is required to develop cognitive screening approaches alternative to computerized versions of paper-based neuropsychological tests.

Abstract
During the last decades, local, regional, and national governments promoted the development of smart cities, aiming the integration of traditional urban infrastructures and information technologies to provide high quality and sustainable urban services. Smart cities' implementations may change the way the individuals experience the urban spaces. Looking specifically to older adults, smart cities' applications have the potential of promoting their autonomy, independence, safety, well-being, social participation, and inclusion. This paper presents a survey of the scientific literature aiming to analyse current evidence related to smart cities' applications to support older adults and to identify issues for future research.

Abstract

Abstract
Embodied Communication Agents are being used to support older adults but there is still a lack of knowledge about their impact. Therefore, this paper presents a scoping review of the literature on user centred evaluation of the impact of Embodied Communication Agents when embedded in digital solutions to support older adults. The ten studies that were included in this scoping review reported on the implementation of digital solutions to promote physical activity, minimize loneliness, provide spiritual comfort, facilitate problem solving and knowledge acquisition, and support the diagnosis of neurodegenerative diseases. In terms of impact assessment, different outcomes were considered, and the included studies reported significant impact in terms of steps walked and acknowledge acquisition, a trend in the reduction of loneliness, and good sensitivity and specificity to diagnose neurodegenerative diseases.

Abstract
The aim of this study is to analyze the typical process of the practical part of software development courses at universities and to evaluate whether the current process meets the expectations of students or whether the needs of students would be better met by the benefits of automated code feedback. A semiautomated survey was conducted in German, involving bachelor students from different universities who had attended an introductory programming class within the last 6 semesters. The results clearly show that the students would like to have individual assignments instead of the due to time constraints, usual group exercises as they see more advantages for their learning progress. The use of automated code feedback could not only solve this time problem but would also bring other benefits.

Abstract
Decision trees are often preferred when implementing Machine Learning in embedded systems for their simplicity and scalability. Hoeffding Trees are a type of Decision Trees that take advantage of the Hoeffding Bound to allow them to learn patterns in data without having to continuously store the data samples for future reprocessing. This makes them especially suitable for deployment on embedded devices. In this work we highlight the features of a HLS implementation of the Hoeffding Tree. The implementation parameters include the feature size of the samples (D), the number of output classes (K), and the maximum number of nodes to which the tree is allowed to grow (Nd). We target a Xilinx MPSoC ZCU102, and evaluate: the design's resource requirements and clock frequency for different numbers of classes and feature size, the execution time on several synthetic datasets of varying sizes (N) and the execution time and accuracy for two datasets from UCI. For a problem size of D=3, K=5, and N=40000, a single decision tree operating at 103MHz is capable of 8.3x faster inference than the 1.2 GHz ARM Cortex-A53 core. Compared to a reference implementation of the Hoeffding tree, we achieve comparable classification accuracy for the UCI datasets.