Analyze-Me
Analyze-Me: Software tools for processing physiological signals from patients with non-verbal autism
Phase 3
Type of project: Feasibility study
Disabilities concerned: Autism and Pervasive Developmental Disorders, Mental disability, Polydisability
Topics: Autonomy, Services and communication, Health, Interpersonal relations
Statut : In progress
Development of software for the detection and visualisation of physiological signals linked to stress factors in patients with a non-verbal form of autism.
Patients with a non-verbal form of autism are unable to communicate with others and are at risk of developing difficult behaviours (self-aggression, hetero-aggression, shouting, destruction, motor instability). Some of these behaviours can be dangerous not only for the patient but also for their carers. When crises occur, it is essential to detect them so that appropriate action can be taken to change the environmental conditions and mitigate their consequences. Given that verbal communication is very limited, if not impossible in some cases, it is difficult to understand the reasons for a seizure. It is difficult to determine whether
they are an alternative way of expressing the person’s needs or whether they reflect a state of consciousness or a somatic experience, stress or pain. We have already received support from the Innovation Booster for exploratory research and a prototype. The aim of our project is to develop software to detect and visualise SNS activation, known to correlate with acute pain/stress or alternative communication, by analysing physiological signals collected using the Empatica E4 wristband. These signals include measurements such as electrodermal activity (EDA), blood volume pulse (BVP), temperature, and 3D acceleration (ACC).
As a result of our previous project, we have developed a comprehensive graphical user interface for physiological signal processing, enabling us to extract over 100 different signal characteristics from BVP and EDA signals. We have also received ethics committee approval for data collection and have already started collecting data. On the basis of the initial data, we have been able to classify stressful events against calm periods with 98% accuracy, and to differentiate stressful events from attempts at non-verbal communication (which is difficult for untrained staff to achieve) with 97% accuracy.
