Skip to main content

NEUROPREDICT – Artificial Intelligence to optimize rehabilitation for patients with acquired brain injury (ABI)

General description of the project

NEUROPREDICT project seeks to optimise motor rehabilitation processes for people with Acquired Brain Injury (ABI), integrating brain activity and movement analysis tools with artificial intelligence techniques. Stroke is the leading cause of ABI in the adult population. In addition, 85% of patients who survive continue to experience motor function disorders. Many patients achieve full or partial motor recovery through therapeutic rehabilitation processes.  

Despite its proven effectiveness, post-stroke rehabilitation still presents significant challenges. Despite advances in post-stroke rehabilitation, current processes rely heavily on clinical decisions based on standardised scales which, although useful, are subject to a degree of subjectivity and have limited sensitivity to capture the complex and heterogeneous neuroplastic changes that occur during motor recovery.  

Various studies have demonstrated the existence of techniques that enable the measurement of brain function reorganisation during rehabilitation. These measurements serve to evaluate the effectiveness of interventions and facilitate clinical decision-making. Compared to neuroimaging techniques (functional magnetic resonance imaging, positron emission tomography), which are costly, difficult to access and have certain contraindications, electroencephalography (EEG) devices are affordable, portable and non-invasive. On the other hand, at the motor level, the use of inertial sensors allows for the objective measurement of the execution and range of a movement during rehabilitation. 

This project will use a combination of electroencephalography (EEG) data and inertial sensors to objectively characterise the changes that occur at the cerebral and motor levels in people with Acquired Brain Injury during rehabilitation processes. The joint analysis of these signals will provide sensitive indicators of functional evolution and response to therapeutic interventions. The incorporation of artificial intelligence techniques will facilitate the identification of relevant patterns and the generation of predictive metrics, contributing to a better understanding of recovery mechanisms and the development of rehabilitation strategies that are better adapted to the condition and evolution of each patient. 

Objectives

The project aims to study the neural activity associated with rehabilitation processes after a stroke and develop an AI-based programme for data integration and analysis that identifies relevant relationships between variables and supports decision-making aimed at prescribing the optimal treatment for each patient.

To achieve this objective, the project aims to:

  • Build a multimodal and longitudinal database that integrates brain, motor and clinical data collected from patients throughout the different phases of the rehabilitation process following a stroke. This database will enable the characterisation of the type of brain damage and analysis of its evolution as therapy progresses, constituting a significant contribution to the systematic monitoring of recovery processes in neurorehabilitation.
  • Analyse the relationship between brain activity and motor activity during rehabilitation, using machine learning techniques to identify patterns, correlations and dynamics of change in the recorded signals, as well as their location in the brain, and connect them with the functional assessments commonly used by rehabilitation professionals.
  • Define objective and reliable indicators that enable the quantification of both the initial degree of impairment and progress in functional recovery in people who have suffered a stroke, based on integrated analysis of brain, motor and clinical signals.
  • Generate knowledge aimed at personalising rehabilitation treatments by identifying the most relevant variables that influence the evolution of people with acquired brain injury, with the aim of improving the suitability of therapeutic interventions and optimising recovery times and outcomes.

Contribution of NAIR Center

NAIR Center will contribute its expertise in advanced signal analysis, complex data management and exploitation, and the development of artificial intelligence models applied to the clinical field to the project.  

Firstly, it will undertake the processing of electroencephalography and movement signals acquired using various experimental devices, characterised by their high complexity and the presence of noise and unwanted components. To this end, automatic processes for reading data in proprietary formats will be developed and advanced digital processing techniques will be applied to improve the signal-to-noise ratio, as well as analysis in the temporal and frequency domains to extract relevant characteristics. This processing will transform the raw signals into reliable, informative and comparable representations, which will serve as the basis for the subsequent stages of the project. 

Based on these processed signals, NAIR Center will contribute to the design, structuring and maintenance of a multimodal database that longitudinally integrates the brain and motor activity records obtained during the rehabilitation process. It will ensure that data collection is systematic, rigorous and reproducible, and that data storage and access are secure and efficient.   

Finally, NAIR Center will lead the application of artificial intelligence techniques aimed at identifying patterns and biomarkers.  

The integration of these developments will enable progress towards a clinical decision support system aimed at personalising rehabilitation treatments and improving their effectiveness. 

Partners

Funding


More projects