Details
- ERC sector
- PE6 - Computer Science and Informatics
- ERC subsector
- PE6_7 - Artificial intelligence, intelligent systems, natural language processing
- Project start date
- CUP
- D53D23017460001
- Financial support received
- €79.369,00
Description and purpose
Motor rehabilitation and related clinical supervision are currently available only at specialized centers (e.g., sanitary gyms) and in the presence of healthcare professionals (e.g., physiotherapists), a problem that severely limits the availability to these services. The current project introduced an innovative technological framework to dispense with costly infrastructures, and to increase the number of patients a healthcare professional can supervise. To achieve this, the UNIPR Research Unit worked in concert with the other Research Units to integrate real-time and semi-autonomic monitoring, training, and supervision of rehabilitation activities potentially performable by a subject, irrespective of the physical contex where he can exercise, including his own house.
Purpose
To achieve this, we profited of the currently available AI methodologies for integration of sensor data from a subsymbolic (machine learning) framework into an ontology-based knowledge representation system that models the healthcare professional’s view of the patient. This approach leads to a symbolic avatar of the pathophysiological variables of the subject, represented as a digital twin. The SORTT project envisages a smart space where a variety of sensors acquire a grid of motor, physiological and stress-related conditions to describe and monitor the patient's state and activity during the motor rehabilitation session. The smart space collects the (dedicated and generic) data sources available in the environment, which may differ in type, frequency of data collection, and data quality. It also establishes whether the incoming data are sufficient to safely monitor the patient, and elaborates the data into digital biomarkers that are significant to healthcare professionals.
Expected results
In this research perspective, the UNIPR Unit contributed to set an innovative methodology to implement the chosen approach, defined as Ontology-based Modeling system. This descriptive and inferential frame is based on collection of biomarkers, that can be integrated through an ad hoc language or ontology with environmental data, leading to a comprehensive picture of the context where the subject performs the rehabilitation exercises. Simplification and generalization of classical anatomical descriptors play a substantial role in the area of the language related to the description of movements. As a result, the project and the activity of the UNIPR Research Unit brought significant knowledge advances in multiple fields (rehabilitation, IoT, applied ontology, AI for healthcare), and involved highly interdisciplinary collaborations with national academic centers (University of Bologna, CNR-ISTC) and international groups (Tufts University School of Medicine, Boston, MA, USA, and RIV Capital - Luxembourg and Dubai - UAE).
Achieved results
In summary, the project proved feasibility of the methodologies applied, and set a new perspective in digital health monitoring including studies on the patentability and marketability of the research results. International and National research publications on all these topics have been provided.
Scientific director
Modified on