Postdoc position in the lab of Prof. Gregoire Courtine at EPFL (Lausanne, Switzerland)

Machine learning techniques to develop and enhance clinical treatments based on electrical stimulation of the spinal cord

Location:

The laboratory of Prof. Gregoire Courtine at the Swiss Federal Institute of Technology (EPFL) in Lausanne, Switzerland, is looking to fill a fully funded postdoc position. The qualified candidate will benefit from joining a very dynamic and multidisciplinary group working at the interface of computational neuroscience, neuroengineering, prosthetics and biology. EPFL provides state-of-the-art facilities and is one of the leading technical universities worldwide. Postdoc salaries at EPFL rank the highest in the world.

Opportunity:

The offered position will be based at the Defitech Center for interventional Neurotherapies (NeuroRestore) – a research and innovation center joining EPFL’s lab of Prof. Gregoire Courtine and the University Hospital of Lausanne (CHUV) lab of Prof. Jocelyne Bloch. NeuroRestore conceives, develops and applies medical therapies aimed to restore neurological functions. To this end, NeuroRestore integrates implantable neurotechnologies with innovative treatments developed through rigorous preclinical and clinical studies. By working with our network of vibrant high-tech start-ups and established medical technology companies, NeuroRestore is committed to validate our medical therapy concepts. The overarching goal of NeuroRestore is to see our medical therapies used every day in hospitals and rehabilitation clinics worldwide.

Description:

Therapies based on epidural electrical stimulation (EES) of the spinal cord can restore the ability to walk to people paralyzed by spinal cord injury, and alleviate gait deficits of people with Parkinson’s disease. EES does this by recruiting sensory axons within dorsal spinal roots that enter the spinal cord between the vertebrae to increase the activation of the spinal motor pools that, in turn, move the muscles. Yet, the efficacy of the EES-based therapies relies on synchronizing users’ movement intentions with the spatiotemporal stimulation protocols that reliably and accurately generate paralyzed movements. Due to the large state space of all the stimulation parameters (location, amplitude, frequency, etc.) efficacy of the therapy depends on the fast and accurate initialization of the stimulation protocols. As the patients use the stimulation, small movements of the array, as well as changes in spine position due to users’ posture can reduce the usability of the stimulation. Stimulation efficacy can be enhanced by dynamically adjusting the stimulation protocols to changes to the way how to users’ spinal cord reacts to stimulation. Finally, the functional use of the stimulation largely depends on the accurate timing of stimulation delivery. Machine learning approaches that infer users’ intentions based on behavioral, physiological or neural recordings can vastly improve the synchronization between intended and therapy-supported movements and, therefore, play a critical role in achieving functional recovery of patients. While the current medical devices mostly support block-based stimulation protocols that remain constant for hundreds of milliseconds, upcoming devices will enable changes of stimulation at a millisecond resolution, thus opening a new field for machine learning approaches that exploit these capabilities.

The successful candidate will work to develop, implement and apply machine learning algorithms and approaches to enhance EES-based therapies. Specifically, he will:

• Design the mapping procedures for the generation of transfer functions that relate the continuously-controlled stimulation to the evoked muscle activity.
• Develop algorithms that automatically adjust these transfer functions as the interaction between patients and their EES-based therapy evolves.
• Implement machine learning techniques that utilize users’ behavioral, physiological and neural signals to continuously synchronize the delivery of stimulation with the users’ movement intentions.
• Lead the team that develops machine learning methods to initialize and adjust block-based EES protocols.
• Assist and oversee the development and implementation of machine learning methods that use inference of discrete motor events to synchronize block-based EES protocols with the users’ intentions.

By integrating well-equipped and expertly staffed rodent, non-human primate and clinical research facilities, NeuroRestore provides an ideal substrate for rapidly developing, integrating and clinically validating cutting-edge machine learning concepts within medical therapies, with the capacity to push successfully proven concepts into the technology transition phase. The successful candidate will have access to these animal platforms and will work within the framework of multiple NeuroRestore clinical trials with people with spinal cord injury and Parkinson’s disease. They will benefit from the possibility of validating their concepts in animal experiments and implementing them within the therapies being tested in the clinical trials.

Prerequisites:

• Doctoral degree (PhD)
• Proficiency in Python, Matlab and C++
• Strong background in quantitative data analysis
• Experience with applying multiple machine learning techniques to behavioral, physiological, biological and/or neural datasets
• Good written and verbal skills in English

Contact:

Applications including a CV and a cover letter describing your background and interest should be sent to tomislav.milekovic at epfl.ch. Informal inquiries are welcome.