AISN vision

Integrating AI in Stroke Neurorehabilitation

Health services are increasingly moving towards a treatment continuum aligned with the patient journey. This transition will critically depend on the successful deployment of trustworthy AI-enhanced technologies that are accurate, secure, and trusted. The AISN project will develop and validate operating procedures and guidelines for integrating AI in a healthcare continuum, focussing on post-stroke rehabilitation.

AISN delivers a representative AI health platform built from integrating validated platforms for data acquisition and access, clinical interpretation, whole-brain simulation, clinically validated intervention delivery and optimization and model-based prediction. The AISN integrated platform will be validated in the clinical context of rehabilitation in the outpatient and at-home phase and facilitate a concrete assessment of the fundamental ways in which AI-enhanced clinical decision-support will change the care pathway and the formulation of novel AI compatible treatment guidelines. AISN ensures an ethical approach by developing legal and ethical guidelines for the robust, fair, and trustworthy deployment of AI in health and validating acceptance and transparency of its solutions. Building on the AISN platform, the project will test current standard operating procedures for integrating AI in health care and formulate and validate new ones where needed. The AISN guidelines and procedures will emphasize the evidence base and safety of clinical interventions, transparency, prognostics at varying time-scales, personalization of interventions, access to disease-specific information by clinicians, patients and their carers, and assure that the potential of AI is fully developed in the service of value-based medicine satisfying standards of security and safety. We will go beyond currently available guidelines and frameworks by emphasizing explainability, AI tools with evolving performance, and the dynamic interaction between users and algorithms.

AISN challenge

for clinical decision-support and intervention delivery

AISN Health services are increasingly moving towards a treatment continuum aligned with the patient journey from the clinic to the home environment. This transition will critically depend on the successful deployment of trustworthy AI-enhanced technologies for data processing, diagnostics, prognostics and patient-tailored intervention that are accurate, secure, and trusted by clinicians and patients alike.

The AISN platform for clinical decision-support and intervention delivery is built from the integration of three validated platforms for data acquisition and access (EBRAINS' knowledge graph), clinical interpretation, and whole-brain simulation (CHARITE's Virtual Research Environment, VirtualBrainCloud), and intervention planning, delivery and optimization (EODYNE's Rehabilitation Gaming System) with advanced AI tools for the extraction of predictive information from complex high-dimensional and heterogeneous medical data (SADDLEPOINT’s Bayesian Inference Engine). The resulting AISN platform is a representative testbed for the introduction of AI in healthcare because it both includes AI-enhanced components and modules in the cloud and the edge for data processing and interpretation, and prognostics, with clinically validated technology for automated control of interventions, personalization of treatment protocols, and patient-specific signal processing and behaviour and performance analysis.

Recent research has shown that AI systems can be remarkably persuasive - even more effective than humans at changing people's minds during conversations. While this could be beneficial for promoting healthy behaviors during rehabilitation, it also raises concerns about potential manipulation. We need to ensure that AI systems maintain a balance between supporting patient decisions and unduly influencing them.

AISN Rehabilitation Gaming System

Eodyne RGS

AISN simulations

BrainX3

With the integration of whole-brain modeling, BrainX3 enables clinicians to simulate the potential development of brain lesions and other abnormalities. This simulation allows for the prediction of brain activity post-lesion, helping to identify significant changes in neural dynamics and connectivity patterns. By understanding these changes, clinicians can develop more precise and personalised treatment strategies, thereby enhancing prognostic accuracy.

The signal processing module within BrainX3 facilitates precise analysis of brain activity, which is essential for diagnosing neurological conditions. This module supports preprocessing, denoising, and detailed analysis of EEG and MEG data, allowing clinicians to extract meaningful insights from raw signals. A key feature of this module is the Power Spectral Density (PSD) visualisation, which correlates strongly with behavioural patterns. This correlation aids in identifying biomarkers crucial for monitoring disease progression and treatment response. The integration of signal processing capabilities ensures that BrainX3 can provide a comprehensive pipeline for biomedical signal analysis, advancing both diagnostics and prognostics.