Raul Sîmpetru
@RaulCSimpetru
Followers
29
Following
43
Media
8
Statuses
19
content creator for journals (PhD student), N² Lab, @UniFAU who needs safe spaces when you can have latent spaces?
Erlangen, Germany
Joined August 2022
How can non-invasive neural interfaces effectively decode motor intentions from individuals with spinal cord injuries and amputations to restore functional abilities?@ScienceAdvances "MyoGestic: EMG interfacing framework for decodingmultiple spared motor dimensions in
0
1
3
MyoGestic is officially published in #ScienceAdvances! To celebrate we updated the repository ( https://t.co/LsdwLORKgI) with better documentation and API. We're committed to making MyoGestic the cummunity's go-to framework and welcome your feedback! https://t.co/X644IafaDz
science.org
AI-powered EMG decoding with MyoGestic enables intuitive real-time control of spared movements for users with neural lesions.
Ever wanted to test a new myocontrol idea, but the thought of creating a responsive interface to experiment with the algorithm seemed daunting? MyoGestic 💪👑 the open-source EMG interfacing framework we developed can help you out. Preprint: https://t.co/scuMu4fiW9
0
1
4
We’re live at #Consumenta this week, showing how tech can help paralyzed individuals regain hand function! Stop by to see our #MyoGestic demos and explore the future of assistive tech. #AssistiveTech #Innovation
Ever wanted to test a new myocontrol idea, but the thought of creating a responsive interface to experiment with the algorithm seemed daunting? MyoGestic 💪👑 the open-source EMG interfacing framework we developed can help you out. Preprint: https://t.co/scuMu4fiW9
0
0
2
@IEEEembs Huge thanks to @DanielaOlivi for her invaluable input on figure aesthetics and text, Dr. Matthias Ponfick for providing access to the patients and answering my medical questions, and of course, @AlecsDelVecchio for his patience 😀 while I talked about circles for months!
0
0
2
@IEEEembs 3/3 Short summary Cool results: - While trained to proportionally predict movements, our model also learned to separate them. Classification accuracy of 98.3% in SCI. - The latent space forms circular patterns and the circularity correlates with prediction error.
1
0
1
@IEEEembs 2/3 Short summary We trained an AI model for each participant to predict the virtual hand kinematics from their EMG signals. We then reduced the dimensionality of the model's latent space (model internal > 1000D vector) for 2D analysis.
1
0
1
@IEEEembs We recorded EMG signals from 7 motor-complete spinal cord injured and 13 uninjured participants as they followed a virtual hand. Tasks included individual finger movements, grasps, and pinches.
2
0
1
Our paper on identifying spared movements after motor-complete SCI is now published in @IEEEembs TNSRE! 🎉 Discover how we use AI to predict individual finger movements and its latent space helps detect spared proportional control. https://t.co/h9QDuagncT 1/3 Short summary
Can we predict the movements of the paralyzed hand from the foream EMG activity without prior information? We introduce a method based on the latent manifolds, which are embedded in specific dimensions across tasks in injured and uninjured individuals https://t.co/w1NxQLJf8Z
1
2
11
@DominikIBraun 5/ For more details check out our preprint: https://t.co/scuMu4fiW9 Huge thanks to everyone who made this research possible! A special shoutout to our supervisor, @AlecsDelVecchio, for the unwavering support he provides day in and day out.
0
0
0
@DominikIBraun 4/ Our goal is for MyoGestic to become the community’s go-to choice for real-time myocontrol projects. Do not be shy to make an issue or drop us a message directly!
1
0
0
3/ My co-first author @DominikIBraun and I would be thrilled if you gave MyoGestic a try: https://t.co/LsdwLORKgI While the framework is still in its early stages, we welcome any input or suggestions you may have regarding its capabilities.
github.com
Software framework made to help the myocontrol community to develop and test new myocontrol algorithms. - NsquaredLab/MyoGestic
1
0
0
2/ MyoGestic is many things: developer-friendly, intuitive, and responsive, but most importantly, it's input/output/algorithm agnostic. The motor intent decoded from individuals with neural lesions can be used to control a virtual hand, an orthosis, a prosthesis, or a 2D cursor.
1
0
0
1/ In our new study, we paired a 32-channel EMG bracelet with a software framework we developed, called MyoGestic, to decode motor intent in the hands and legs of individuals with spinal cord injury, spinal stroke, and amputation.
1
0
0
Ever wanted to test a new myocontrol idea, but the thought of creating a responsive interface to experiment with the algorithm seemed daunting? MyoGestic 💪👑 the open-source EMG interfacing framework we developed can help you out. Preprint: https://t.co/scuMu4fiW9
We are about to open source a software to record and train *instantaneously* high-density EMG signals and move a virtual hand with an AI that takes <20 seconds for each movement as training input. It works with any EMG and in patients with paralysed hands due to SCI/stroke.
1
1
6