Fabio Cassano

Classification of healthy subjects and Alzheimer's disease patients with dementia from cortical sources of resting state EEG rhythms: A study using artificial neural networks

Previous evidence showed a 75.5% best accuracy in the classification of 120 Alzheimers disease (AD) patients with dementia and 100 matched normal elderly (Nold) subjects based on cortical source current density and linear lagged connectivity estimated by eLORETA freeware from resting state eyes-closed electroencephalographic (rsEEG) rhythms (Babiloni et al., 2016a). Specifically, that accuracy was reached using the ratio between occipital delta and alpha1 current density for a linear univariate classifier (receiver operating characteristic curves). Here we tested an innovative approach based on an artificial neural network (ANN) classifier from the same database of rsEEG markers. Frequency bands of interest were delta (2–4 Hz), theta (4–8 Hz Hz), alpha1 (8–10.5 Hz), and alpha2 (10.5–13 Hz). ANN classification showed an accuracy of 77% using the most 4 discriminative rsEEG markers of source current density (parietal theta/alpha 1, temporal theta/alpha 1, occipital theta/alpha 1, and occipital delta/alpha 1). It also showed an accuracy of 72% using the most 4 discriminative rsEEG markers of source lagged linear connectivity (inter-hemispherical occipital delta/alpha 2, intra-hemispherical right parietal-limbic alpha 1, intra-hemispherical left occipital-temporal theta/alpha 1, intra-hemispherical right occipital-temporal theta/alpha 1). With these 8 markers combined, an accuracy of at least 76% was reached. Interestingly, this accuracy based on 8 (linear) rsEEG markers as inputs to ANN was similar to that obtained with a single rsEEG marker (Babiloni et al., 2016a), thus unveiling their information redundancy for classification purposes. In future AD studies, inputs to ANNs should include other classes of independent linear (i.e., directed transfer function) and non-linear (i.e., entropy) rsEEG markers to improve the classification.

Optimizing Feed-Forward Neural Network Topology by Multi-objective Evolutionary Algorithms: A Comparative Study on Biomedical Datasets

The design of robust classifiers, for instance Artificial Neural Networks (ANNs), is a critical aspect in all complex pattern recognition or classification tasks. Poor design choices may undermine the ability of the system to correctly classify the data samples. In this context, evolutionary techniques have proven particularly successful in exploring the complex state-space underlying the design of ANNs. Here, we report an extensive comparative study on the application of several modern Multi-Objective Evolutionary Algorithms to the design and training of an ANN for the classification of samples from two different biomedical datasets. Numerical results show that different algorithms have different strengths and weaknesses, leading to ANNs characterized by different levels of classification accuracy and network complexity.

A P300 Clustering of Mild Cognitive Impairment Patients Stimulated in an Immersive Virtual Reality Scenario

In this paper, we present an innovative framework useful for clustering patients affected by a mild cognitive impairment and designed to improve the living environment and the lifestyle of patients in order to delay their cognitive state decline. The cognitive state changes are evaluated by means the event - related potentials elicited by environmental stimuli administered in several Virtual Reality scenarios. In particular, we formerly describe our innovative Virtual Reality environment, the protocol of stimuli administration, the procedure to measure the P300 latency in the response signal and finally the Self Organizing Map used to cluster the data. This research finds application in the fields of re-qualification of the environments for patients and healthcare introducing a new method for evaluation of best living conditions through VR.

Towards an IoT model for the assessment of smart devices

The current Internet of Things (IoT) market proposes a wide variety of devices with complex design and different functionality. In addition, the same IoT device can be used in different domains, from home to industry, to healthcare. The management of such devices occurs in different ways, for example through visual interaction using high level programming languages (e.g. Event-Condition-Action rules) or through high level API. Generally, end users are not technical experts and are not able to configure their IoT devices, thus they need external tools (or visual interaction paradigm) to exploit and better control them. In this work, we present a model for IoT devices which allows to assess those devices and their suitability for a certain domain according to four dimensions: communication, target, data manipulation and development. The model aims at better understanding the device capabilities and, consequently, facilitating the choice of the devices that better suit the domain in which they should be used.

Gamification and Learning Analytics to Improve Engagement in University Courses

Gamification is one of the most used techniques to improve active participation and engagement in different kinds of contexts. The use of game techniques is effective in pushing subjects to be involved in an activity. Since the early childhood, indeed, the promises of rewards are useful to affect specific behaviors. On the other hands, the learning analytics have been largely implemented in education in order to improve the assessment and the self-assessment of students, above all in e-learning settings. The research presented in this work aims at combining gamification techniques and learning analytics to improve the engagement in University courses. The paper describes a model of gamification and a learning dashboard defined based on data in Moodle e-learning platform. A pilot test of an app android in which both the solutions have been implemented pointed out promising results.

Supporting end users to control their smart home: design implications from a literature review and an empirical investigation

Abstract Designing tools that allow end users to easily control and manage a smart home is a critical issue that researchers in Ambient Intelligence and Internet of Things have to address. Because of the variety of available solutions, with their advantages and limitations, it is not straightforward to understand which are the requirements that must be satisfied to effectively support end users. This paper aims to contribute to this topic through a systematic and rigorous activity based on two main pillars of the empirical research in software engineering: i) a literature review addressing design and evaluation of tools for smart home control oriented to end users, and ii) an experimental study in which three tools, that emerged from the literature review as the most suitable and widespread, were compared in order to identify the interaction mechanisms that end users appreciate most. On the basis of the obtained results, a set of design implications that may drive the development of future tools for smart home control and management are presented.

EUDroid: a formal language specifying the behaviour of IoT devices

Recent technologies are offering today many possibilities to end users, which ask for continuous support in a variety of situations. Internet of things (IoTs) and the proliferation of smart devices are offering many opportunities that raise the need to standardise protocols for their interoperability and interaction languages for their management. This study proposes EUDroid, a system composed of a mobile application and an IoT device used as a pill reminder to allow the patients to correctly take their prescribed drugs. A web server stores and manages the therapies that can be defined by the end users. The web server also manages the communication between the app and the device. In order to validate the management of the therapies, a formal language has been proposed. It describes the behaviour of different components of the IoT device, such as LEDs or buzzers, and defines when, with which delay, and for how long time a given event will last, to manage technical concepts related to smart devices for supporting them in following therapies more accurately.

A Homemade Pill Dispenser Prototype Supporting Elderly.

People, and mainly elderly people, need a continuous support for different reasons. Recent technologies are offering many possibilities that was not possible to conceive in the past. In particular, the proliferation of IoT devices raise the need to standardize protocols and interaction languages. The aim of this work is to create a device for the management of pills according to the user’s therapy, with Internet of things (IoT) devices and by allowing users to manage the pill dispenser by themselves. The work falls into two main areas of current research: the End-user development (EUD) and the Internet of things (IoT). The main issue we cope with such device is to allow the different therapies for each person and for each drug. We propose the EUDroid system, which provides the end user with the possibility to easily activate LEDs and buzzer related to pills from the users’ smartphone. The user chooses the type of pill to be associated to each LED, the day and time of activation and some other property. A formal language to configure the device has been adopted in order to allow users to build complex conditions for remind to follow the therapy.