Measuring stress level objectively is crucial for personalized health monitoring. While traditional methods require a clinical setting, wearables provide a valuable alternative. In this paper, we approach stress assessment as a regression task, focusing on stress exposure, and evaluate Functional Data Analysis (FDA) to extract richer information from physiological signals. We apply scalar-on-function regression and functional clustering to WESAD, a public dataset which contains signals from wearables and psychometric questionnaires that we use as a ground truth for stress. We compare the results obtained by applying FDA with those achieved by methods using features extracted from signals rather than the signals themselves. The comparison reveals that FDA excels in capturing signal variations and their association with stress, offering new insights into how this association changes with different stressful activities. While non-functional techniques suffice for some analyses, FDA is key to capture overtime patterns linked to stress levels.

The tenth IEEE International Conference on Smart Computing (SmartComp 2024), sponsored by the IEEE Computer Society, will be held in-person in Osaka, Japan. It continues the tradition of the previous editions in presenting high-quality research and technology in smart and connected computing.

Smartphones and wearable devices, along with Artificial Intelligence, can represent a game-changer in the pandemic control, by implementing low-cost and pervasive solutions to recognize the development of new diseases at their early stages and by potentially avoiding the rise of new outbreaks. Some recent works show promise in detecting diagnostic signals of COVID-19 from voice and coughs by using machine learning and hand-crafted acoustic features. In this paper, we decided to investigate the capabilities of the recently proposed deep embedding model L3-Net to automatically extract meaningful features from raw respiratory audio recordings in order to improve the performances of standard machine learning classifiers in discriminating between COVID-19 positive and negative subjects from smartphone data. We evaluated the proposed model on 3 datasets, comparing the obtained results with those of two reference works. Results show that the combination of L3-Net with hand-crafted features overcomes the performance of the other works of 28.57% in terms of AUC in a set of subject-independent experiments. This result paves the way to further investigation on different deep audio embeddings, also for the automatic detection of different diseases.

Malnutrition represents a major public health con-cern worldwide, and it particularly harms older adults since it is frequently associated with several chronic health disorders. It significantly increases in institutionalised subjects, especially in presence of cognitive impairments. For this reason, it is essential to early detect nutritional deficiencies and unhealthy dietary habits and to timely trigger proper feedback and interventions. Malnutrition assessment in clinical settings is generally based on standard screening tools including questionnaires, rating scales, and biometrics. Technological solutions based on IoT and mobile devices, along with AI techniques for data analysis, could provide important advantages in the risk assessment and prevention. In this paper we present a Decision Support System for the early detection of malnutrition risk based on data collected by a m-health application for nutritional and body composition monitoring. The application has been in use in a nursing home in Italy from March 2018 and, considering drop-outs and the impact of Covid-19 pandemic, we have been able to collect consistent data over three different trial periods. In collaboration with a medical specialist, we performed feature engineering to estimate daily intake for the major food components, meal completeness, variability, considering also physiological data. Then, we ran several Machine Learning models using the results of Mini Nutritional Assessment rating scale as ground truth, and applying SMOTE and cost-sensitive learning to deal with the dataset imbalance. Obtained results indicate that the best performing ML models for malnutrition risk prediction reach median accuracy and recall values of 94% and 92%, respectively.

High-resolution stress detection is an essential requirement for designing time- and event-based stress monitoring systems as a building block for mobile and e-health systems aimed at supporting personalised treatments, both in clinical and remote settings. However, most of the existing solutions focus on binary or few-class stress detection, thus providing a limited feedback and reducing their utility and applicability in real- world scenarios. In this paper we present an alternative approach that overcomes the standard formulation of stress detection as supervised classification problem, by using ensemble learners and recurrent neural networks (RNNs) as the most relevant models for solving time series regression tasks. We trained and tested models using WESAD, a public multimodal wearable dataset for stress and affect detection, and we defined and computed stress scores based on various validated questionnaires stored in the dataset to be used as ground truth. Leave-One- Subject-Out (LOSO) cross-validation scheme has been applied to test the generalisation capabilities of each model in predicting individual stress scores. Results show that Nonlinear AutoRegressive network with eXogenous inputs (NARX), Random Forest (RF), and Least-Squares Gradient Boosting (LSBoost) provide high-resolution personalised stress predictions for the majority of analysed subjects. The proposed predictive models may be integrated as support to decision making into a Decision Support System (DSS) for online stress monitoring, with the main goal to design personalised stress management and alleviation strategies related to the inferred stress severity.

In the last decade, ageing has become a worldwide increasing phenomenon leading to an increased need for specialised help. Continuous monitoring, training and rehabilitation improve the quality of life of older adults and discard the risk of depression and social isolation. To this end, the use of mobile and e-health personalised services at home and in residential long-term care facilities can help to stabilise the health conditions of the subjects, in terms of physical, mental, and social capabilities. In this context, we propose a set of personalised monitoring and rehabilitation services based on mobile, wearable, free contact, and touch screen technologies designed to provide an integrated care and monitoring programme for elderly frail subjects. We evaluated the proposed solutions by deploying the services in a nursing home in Italy and defined customised protocols to involve both guests (primary users) and nursing care personnel (secondary users). In this paper, we present technical details of the proposed solutions and the results obtained by the recently conducted surveys on the Quality of Experience and user acceptance of both user categories after 4 months from the deployment.

Wi-Fi Direct is a promising technology for the support of device-to-device communications (D2D) on commercial mobile devices. However, the standard as-it-is is not sufficient to support the real deployment of networking solutions entirely based on D2D such as opportunistic networks. In fact,WiFi Direct presents some characteristics that could limit the autonomous creation of D2D connections among users’ personal devices. Specifically, the standard explicitly requires the user’s authorization to establish a connection between two or more devices, and it provides a limited support for inter-group communication. In some cases, this might lead to the creation of isolated groups of nodes which cannot communicate among each other. In this paper, we propose a novel middleware-layer protocol for the efficient configuration and management of WiFi Direct groups (WiFi Direct Group Manager, WFD-GM) to enable autonomous connections and inter-group communication. This enables opportunistic networks in real conditions (e.g., variable mobility and network size). WFD-GM defines a context function that takes into account heterogeneous parameters for the creation of the best group configuration in a specific time window, including an index of nodes’ stability and power levels. We evaluate the protocol performances by simulating three reference scenarios including different mobility models, geographical areas and number of nodes. Simulations are also supported by experimental results related to the evaluation in a real testbed of the involved context parameters. We compare WFD-GM with the state-of-the-art solutions and we show that it performs significantly better than a Baseline approach in scenarios with medium/low mobility, and it is comparable with it in case of high mobility, without introducing additional overhead.

The current availability of interconnected portable devices, and the advent of the Web 2.0, raise the problem of supporting anywhere and anytime access to a huge amount of content, generated and shared by mobile users. On the one hand, users tend to be always connected for sharing experiences and conducting their social interactions with friends and acquaintances, through so-called Mobile Social Networks, further improving their social inclusion. On the other hand, the pervasiveness of communication infrastructures spreading data (cellular networks, direct device-to-device contacts, interactions with ambient devices as in the Internet-of-Things) makes compulsory the deployment of solutions able to filter off undesired information and to select what content should be addressed to which users, for both (i) better user experience, and (ii) resource saving of both devices and network. In this work, we propose a novel framework for pervasive social networks, called Pervasive PLIERS (p-PLIERS), able to discover and select, in a highly personalized way, contents of interest for single mobile users. p-PLIERS exploits the recently proposed PLIERS tag-based recommender system (Arnaboldi et al., 2016) as a context reasoning tool able to adapt recommendations to heterogeneous interest profiles of different users. p-PLIERS effectively operates also when limited knowledge about the network is maintained. It is implemented in a completely decentralized environment, in which new contents are continuously generated and diffused through the network, and it relies only on the exchange of single nodes’ knowledge during proximity contacts and through device-to-device communications. We evaluated p-PLIERS by simulating its behavior in three different scenarios: a big event (Expo 2015), a conference venue (ACM KDD’15), and a working day in the city of Helsinki. For each scenario, we used real or synthetic mobility traces and we extracted real datasets from Twitter interactions to characterize the generation and sharing of user contents.

Mobile systems are characterized by several dynamic components such as user mobility, device interoperability, and interactions among users and their devices. In this scenario, context-awareness and the emerging concept of social-awareness become a fundamental requirement to develop optimized systems and applications. In this paper we present CAMEO, a light-weight context-aware middleware platform for mobile devices designed to support the development of real-time mobile social network (MSN) applications. MSNs extend the paradigm of online social networks with additional interaction opportunities generated by user mobility and opportunistic wireless communications among users which share interests, habits, and needs. Specifically, CAMEO is designed to collect and reason upon multidimensional context information, derived by the local device, the local user, and their physical interactions with other devices and users. It provides a common application programming interface to MSN applications through which they can exploit context- and social-aware functionalities to optimize their features. CAMEO has been implemented on an Android platform together with a real example of an MSN application. Validation and performance evaluation have been conducted through an experimental testbed.

In this paper we present DroidOppPathFinder, a Mobile Social Network application designed to generate and share contents about paths for fitness activity in a city. The application is able to recommend the best path in a specific area by analyzing the user's preferences and real-time environmental characteristics collected by heterogeneous sensing devices and services through opportunistic sensing mechanisms. To this aim, DroidOppPathFinder is developed on top of our middleware CAMEO, which provides context- and social-aware functionalities to improve both the application's performances and the user experience. This work represents a real example of opportunistic sensing service as additional support to the development of MSN applications. In addition, it demonstrates an efficient management of heterogeneous sensing data and services on mobile devices in order to further enrich the context of both local and remote nodes.

Opportunistic Mobile Social Networks represent the emerging trend in mobile applications combining the novel paradigm of opportunistic networking with the need of users to generate and share contents anywhere and anytime. In this demo we present a real implementation of a context-aware middleware for the development of optimized opportunistic Mobile Social Networks. To highlight the advantages of this platform we present also an innovative application for tourists aimed at enriching the touristic experience with useful information and virtual social interactions with other users.

Welcome to the third edition of the International Workshop on Interdisciplinary Research on E-Health Services and Systems, being held on Monday, June 20th, 2011 in Lucca, Italy, in conjunction with IEEE WoWMoM conference. It has been our great pleasure to organize this one-day workshop, dealing with a variety of research advances and challenges around e-health services and systems.

Opportunistic networks are mainly characterized by nodes intermittently connected among them. Available applications designed for mobile ad hoc networks are not suitable for such an environment since we cannot assume to have a stable path between pairs of nodes. Network protocols and applications themselves must be enhanced to exploit all possible communication opportunities to deliver messages on the network. In this demo we present an enhanced file sharing application based on the exchange of context information between nodes. In this case the context is defined as a combination of the user personal information, interests, and social relationships in order to implement cooperative downloading mechanisms. Besides reducing the impact of intermittent connectivity and high mobility on multi-hop communications, exploiting context also allows us to avoid flooding, thus resulting in a very efficient approach.

Several structured p2p systems exist in literature and many applications have been developed on top of them. To use these applications on top of different overlays without changing their implementations, a common API was proposed in. However, since that specification is meagre, current implementations of structured p2p systems have customized it reducing the portability of applications. In addition, in mobile environments, the possibility to exploit cross-layer interactions considerably improves overall performace. In fact, a cross-layer p2p system, called CrossROAD, has been recently designed to optimize structured overlays on MANETs. It directly interacts with a proactive routing protocol, and it can provide cross-layer information to upper-layer applications to further optimize their behavior. In this paper we propose a cross-layer extension of the common API pointing out advantages of the cross-layer approach even at the application layer