Marcello Ferro

An android for enhancing social skills and emotion recognition in people with autism

It is well documented that the processing of social and emotional information is impaired in people with autism. Recent studies have shown that individuals, particularly those with high functioning autism, can learn to cope with common social situations if they are made to enact possible scenarios they may encounter in real life during therapy. The main aim of this work is to describe an interactive life-like facial display (FACE) and a supporting therapeutic protocol that will enable us to verify if the system can help children with autism to learn, identify, interpret, and use emotional information and extend these skills in a socially appropriate, flexible, and adaptive context. The therapeutic setup consists of a specially equipped room in which the subject, under the supervision of a therapist, can interact with FACE. The android display and associated control system has automatic facial tracking, expression recognition, and eye tracking. The treatment scheme is based on a series of therapist-guided sessions in which a patient communicates with FACE through an interactive console. Preliminary data regarding the exposure to FACE of two children are reported.

Personal Health System architecture for stress monitoring and support to clinical decisions

Developments in computational techniques including clinical decision support systems, information processing, wireless communication and data mining hold new premises in Personal Health Systems. Pervasive Healthcare system architecture finds today an effective application and represents in perspective a real technological breakthrough promoting a paradigm shift from diagnosis and treatment of patients based on symptoms to diagnosis and treatment based on risk assessment. Such architectures must be able to collect and manage a large quantity of data supporting the physicians in their decision process through a continuous pervasive remote monitoring model aimed to enhance the understanding of the dynamic disease evolution and personal risk. In this work an automatic simple, compact, wireless, personalized and cost efficient pervasive architecture for the evaluation of the stress state of individual subjects suitable for prolonged stress monitoring during normal activity is described. A novel integrated processing approach based on an autoregressive model, artificial neural networks and fuzzy logic modeling allows stress conditions to be automatically identified with a mobile setting analysing features of the electrocardiographic signals and human motion. The performances of the reported architecture were assessed in terms of classification of stress conditions.

Real-time discrimination of multiple cardiac arrhythmias for wearable systems based on neural networks

This paper aims at developing a wearable system able to recognize the most significant cardiac arrhythmias through an efficient algorithm, in terms of low computational cost and memory usage, implementable in a portable, real-rime hardware. In addition, it must respect the specifications of good specificity and sensitivity, in order to permit a positive clinical validation. The hardware is constituted of a general propose microcontroller, which is able to acquire electro-cardiogram signal (ECG), perform analog to digital conversion and extract QRS complex. The algorithm classifies QRS complexes as normal or pathologic by means of selected features obtained from discrete fourier transform (DFT). Furthermore, a spatial wavelet pre-filter is also investigated to obtain an enhanced QRS complex discrimination. In particular, pattern recognition of QRS complex is performed from binding minimal architecture of neural network as Kohonen self organizing map (KSOM). Experimental results were validated by means of MIT-BIH arrhythmias database obtaining specificity and sensitivity up to 98%.

A Sensing Seat for Human Authentication

This work is focused on the design and the realization of a sensing seat system for human authentication. Such a system may be used for security purposes in trucks, cars, offices, and scenarios where human subject authentication is needed and a seat is available. The sensing seat is realized by a seat coated with a removable Lycra sensing cover equipped with a piezoresistive sensor network. Since each sensor consists of a conductive elastomer composite rubber screen printed onto a cotton Lycra fabric, the sensing cover is able to respond to simultaneous deformations in different areas. This technology avoids the use of rigid electronic components and enables the realization of different cover layouts according to different types of seats. The algorithms for the enrollment, authentication, and monitoring tasks are discussed. A measurement campaign was carried out using data from 40 human subjects. The authentication capabilities of the system are reported in terms of acceptance and rejection rates, showing a high degree of correct classification.

Exhaled air temperature in asthmatic children: a mathematical evaluation

Recently, the exhaled breath temperature has been proposed as a potential marker for the evaluation of airway inflammation in asthma. The purpose of this study was to verify the ability to distinguish asthmatics from normal controls by a dedicated detailed mathematical evaluation of the exhaled air curve. Analysis was performed in the different phases of the curve of exhaled temperature, i.e. the rate of temperature increase (Δe°T) and the mean plateau value. Principal components analysis (PCA) and artificial neural networks (ANNs) were used for the evaluation of the data in 90 asthmatic children and in 33 healthy age‐matched controls. Both PCA and ANNs showed that a separation between patients and controls can be obtained only by the evaluation of the plateau phase of the curve, which better reflects the periphery of the airway.

A neuron-astrocyte transistor-like model for neuromorphic dressed neurons

Experimental evidences on the role of the synaptic glia as an active partner together with the bold synapse in neuronal signaling and dynamics of neural tissue strongly suggest to investigate on a more realistic neuron-glia model for better understanding human brain processing. Among the glial cells, the astrocytes play a crucial role in the tripartite synapsis, i.e. the dressed neuron. A well-known two-way astrocyte-neuron interaction can be found in the literature, completely revising the purely supportive role for the glia. The aim of this study is to provide a computationally efficient model for neuron-glia interaction. The neuron-glia interactions were simulated by implementing the Li-Rinzel model for an astrocyte and the Izhikevich model for a neuron. Assuming the dressed neuron dynamics similar to the nonlinear input-output characteristics of a bipolar junction transistor, we derived our computationally efficient model. This model may represent the fundamental computational unit for the development of real-time artificial neuron-glia networks opening new perspectives in pattern recognition systems and in brain neurophysiology.

FACE: facial automaton for conveying emotions

The human face is the main organ of expression, capable of transmitting emotions that are almost instantly recognised by fellow beings. In this paper, we describe the development of a lifelike facial display based on the principles of biomimetic engineering. A number of paradigms that can be used for developing believable emotional displays, borrowing from elements of anthropomorphic mechanics and control, and materials science, are outlined. These are used to lay down the technological and philosophical premises necessary to construct a man-machine interface for expressing emotions through a biomimetic mechanical head. Applications in therapy to enhance social skills and understanding emotion in people with autism are discussed.

Implementation of Fowler's method for end-tidal air sampling.

The design, realization and testing of a CO(2)-triggered breath sampler, capable of a separate collection of dead space and end-tidal air on multiple breaths, is presented. This sampling procedure has advantages in terms of the sample volume, insights regarding the origin of compounds, increased reproducibility and higher concentrations of compounds. The high quality of design and the speed of the components ensure a breath-by-breath estimate of dead volume, as well as the comfort and safety of the subject under test. The system represents a valid tool to contribute to the development of a standardized sampling protocol needed to compare results obtained by the various groups in this field.

Assessment of Bioinspired Models for Pattern Recognition in Biomimetic Systems

The increasing complexity of the artificial implementations of biological systems, such as the so-called electronic noses (e-noses) and tongues (e-tongues), poses issues in sensory feature extraction and fusion, drift compensation and pattern recognition, especially when high reliability is required. In particular, in order to achieve effective results, the pattern recognition system must be carefully designed. In order to investigate a novel biomimetic approach for the pattern recognition module of such systems, the classification capabilities of an artificial model inspired by the mammalian cortex, a cortical-based artificial neural network (CANN), are compared with several artificial neural networks present in the e-nose and e-tongue literature, a multilayer perceptron (MLP), a Kohonen self-organizing map (KSOM) and a fuzzy Kohonen self-organizing map (FKSOM). Each network was tested with large datasets coming from a conducting polymer-sensor-based e-nose and a composite array-based e-tongue. The comparison of results showed that the CANN model is able to strongly enhance the performances of both systems.

A system for automatic detection of momentary stress in naturalistic settings.

Prolonged exposure to stressful environments can lead to serious health problems. Therefore, measuring stress in daily life situations through non-invasive procedures has become a significant research challenge. In this paper, we describe a system for the automatic detection of momentary stress from behavioral and physiological measures collected through wearable sensors. The systems architecture consists of two key components: a) a mobile acquisition module; b) an analysis and decision module. The mobile acquisition module is a smartphone application coupled with a newly developed sensor platform (Personal Biomonitoring System, PBS). The PBS acquires behavioral (motion activity, posture) and physiological (hearth rate) variables, performs low-level, real-time signal preprocessing, and wirelessly communicates with the smartphone application, which in turn connects to a remote server for further signal processing and storage. The decision module is realized on a knowledge basis, using neural network and fuzzy logic algorithms able to combine as input the physiological and behavioral features extracted by the PBS and to classify the level of stress, after previous knowledge acquired during a training phase. The training is based on labeling of physiological and behavioral data through self-reports of stress collected via the smartphone application. After training, the smartphone application can be configured to poll the stress analysis report at fixed time steps or at the request of the user. Preliminary testing of the system is ongoing.