Docket #: S15-182
Online brain machine-interface auto-delete based on motor cortical activity
Stanford researchers at the Shenoy Lab have tested a method that can detect and predict the outcome of brain machine interface (BMI) tasks using motor cortical brain activity. It can improve performance of BMI by auto-deleting incorrect selections using a classifier on neural spiking activity from motor cortex to decode task outcome.
This method has been successfully implemented in an online non-human primate BMI experiment. In real time experiment with closed-loop BMI control, the team identified motor cortical neural signals indicative of task error occurrence. Task outcomes were detected in real time with 93% accuracy and upcoming task outcomes were predicted with 83% accuracy using neural activity alone. This novel strategy can help increase the performance and clinical viability of BMIs.
| Stanford researchers develop brain-controlled typing for people with paralysis. |
Stage of Research:
Applications
- Incorporate the algorithm in human brain machine interface (BMI) to increase performance (e.g., virtual keyboard typing rate).
Advantages
- Increased performance/typing rate by 20%
- High accuracy – 93% accuracy for detecting task outcomes and 83% accuracy for predicting upcoming task outcomes in real time
- Error detection is based on neural activity alone
- Improves upon current BMI technology:
- Current BMI only decode kinematic signals, not neural activity
- Current BMI have reduced typing rates due to wrong characters and errors
- Use of error signal for auto-delete can help increase the ease of use and clinical adoption of BMIs
Publications
- Even-Chen N, Stavisky S, Kao JC, Ryu SI, Shenoy KV (2015), "An auto-correcting brain machine interface: Error detection using spiking neuronal activity in the motor cortex." 37th Annual International Conference of the IEEE Engineering in Medicine and Biology Society, Milan, Italy: 71-75.
Related Links
Patents
- Issued: 10,779,746 (USA)
Similar Technologies
-
Inductive Light Generator (ILG) S06-068Inductive Light Generator (ILG)
-
Improved Virtual-to-Real Alignment for Augmented Reality S19-059Improved Virtual-to-Real Alignment for Augmented Reality
-
Ultrasonic neuromodulation with Pattern Interference Radiation Force (PIRF) S16-382Ultrasonic neuromodulation with Pattern Interference Radiation Force (PIRF)