A Digital Twins of Human Neuromusculoskeletal System: Model-Informed Personalized Neurorehabilitation

One billion people worldwide suffer from various neurological disorders.

The principle of ‘subject-centered’ therapy has been suggested of great importance in developing effectively neurorehabilitation technology.

However, there are two major obstacles in personal-tailored therapy: in vivo monitoring of the spinal neural circuitries and incorporating an individual’s voluntary interfacing with intervention, i.e., neurostimulations.

Di-NeuR proposes a digital twins (DT) platform of the human neuromusculoskeletal system which addresses these challenges for improving clinical diagnosis, monitoring therapy outcomes, and developing novel neuro-rehabilitation technologies.

By integrating wearable sensors, neuromusculoskeletal modeling and machine learning, the detailed functionality of the proposed DT platform includes: 1) real-time investigation of the interplay between motor neuron control and musculoskeletal functions; 2) in vivo estimation of non-measurable intrinsic motor neuron properties; and 3) real-time closed-loop control strategy development for neurostimulations.

The scientific novelty and ground-breaking impact will be demonstrated by 1) in vivo decoding of neural input by accessing the motoneuron firing behavior and properties; and 2) developing a deep reinforcement learning-based adaptive functional electrical stimulation (FES) controller in both digital and physical environments.

Di-NeuR might pioneer a new generation of evidence-based neurorehabilitation practice.