Benyamin Meschede-Krasa
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Benyamin Meschede-Krasa has a total of 66 co-authors. Here are the top co-authors by number of shared publications:
- Donald T. Avansino
Department of Neurosurgery, Stanford University, Stanford, CA, USA
3 shared publications - Jaimie M. Henderson
Department of Neurosurgery, Stanford University, Stanford, USA
3 shared publications - Leigh R. Hochberg
Massachusetts General Hospital
3 shared publications - Nick Hahn
Department of Neurosurgery, Stanford University, Stanford, USA
3 shared publications - Brandon G. Jacques
Wallace H. Coulter Department of Biomedical Engineering, Emory University and Georgia Institute of Technology, Atlanta, GA, USA
2 shared publications - Carrina Iacobacci
University of California, Davis
2 shared publications
Latest Publications
- Open Access
Neural decoding of speech using deep neural ensembles
Speech brain-computer interfaces (BCIs) can restore rapid communication to people with paralysis, but decoding errors still limit performance. In recent brain-to-text decoding competitions, deep ensemble methods, which combine predictions from multiple independently trained decoders, have delivered striking accuracy improvements and account for the largest gains over baseline approaches. However, these methods have not previously been tested in real-time, require substantial computational resources, and their performance under various clinically relevant constraints remains poorly understood. Here, we present the first closed-loop test of deep ensembles in a participant with bilateral intracortical microelectrode arrays, demonstrating a reduction in word error rate from 33.7% to 26.0% on a large-vocabulary task. Using additional data from three participants, we then assess how these gains depend on baseline error rate, training dataset size, and ensemble size, including the resource-accuracy tradeoffs most relevant for real-world deployment. Finally, we introduce a computationally efficient pseudoensembling approach based on test-time augmentation that improves decoding accuracy while requiring only a single base decoder, greatly reducing the computational burden of ensembling. Together, these results show that the benefits of deep ensembling can be realized in real time and under practical resource constraints, bringing speech BCIs closer to broader clinical adoption.
Open AccessNeural decoding of speech using deep neural ensembles
Speech brain-computer interfaces (BCIs) can restore rapid communication to people with paralysis, but decoding errors still limit performance. In recent brain-to-text decoding competitions, deep ensemble methods, which combine predictions from multiple independently trained decoders, have delivered striking accuracy improvements and account for the largest gains over baseline approaches. However, these methods have not previously been tested in real-time, require substantial computational resources, and their performance under various clinically relevant constraints remains poorly understood. Here, we present the first closed-loop test of deep ensembles in a participant with bilateral intracortical microelectrode arrays, demonstrating a reduction in word error rate from 33.7% to 26.0% on a large-vocabulary task. Using additional data from three participants, we then assess how these gains depend on baseline error rate, training dataset size, and ensemble size, including the resource-accuracy tradeoffs most relevant for real-world deployment. Finally, we introduce a computationally efficient pseudoensembling approach based on test-time augmentation that improves decoding accuracy while requiring only a single base decoder, greatly reducing the computational burden of ensembling. Together, these results show that the benefits of deep ensembling can be realized in real time and under practical resource constraints, bringing speech BCIs closer to broader clinical adoption.
- Open Access
Premotor cortex uses a compositional neural geometry to plan words
Speech requires precise serial ordering of words and phonemes into novel combinations. To accomplish this, the brain is believed to flexibly prepare utterances before producing them, even allowing pronunciation of never-before spoken words. To discover how neural populations achieve this, intracortical activity from premotor cortex was recorded while two speech neuroprosthesis pilot clinical trial participants attempted to speak factorially-balanced phoneme sequences. During preparation, activity encoded not only the next-phoneme, but multiple upcoming phoneme positions spanning whole words. We found that word-level plans were formed by compositionally combining phoneme representations, a mechanism that may enable efficient planning of novel sequences. When utterances contained more than one word, premotor cortex activity was largely limited to the first word, suggesting that articulatory planning is segmented by higher-order features. Together, these results reveal a compositional, hierarchically-segemented planning geometry, potentially a universal neural strategy for sequence organization across higher levels of language.
Open AccessPremotor cortex uses a compositional neural geometry to plan words
Speech requires precise serial ordering of words and phonemes into novel combinations. To accomplish this, the brain is believed to flexibly prepare utterances before producing them, even allowing pronunciation of never-before spoken words. To discover how neural populations achieve this, intracortical activity from premotor cortex was recorded while two speech neuroprosthesis pilot clinical trial participants attempted to speak factorially-balanced phoneme sequences. During preparation, activity encoded not only the next-phoneme, but multiple upcoming phoneme positions spanning whole words. We found that word-level plans were formed by compositionally combining phoneme representations, a mechanism that may enable efficient planning of novel sequences. When utterances contained more than one word, premotor cortex activity was largely limited to the first word, suggesting that articulatory planning is segmented by higher-order features. Together, these results reveal a compositional, hierarchically-segemented planning geometry, potentially a universal neural strategy for sequence organization across higher levels of language.
- Open Access
Improved interpretability in LFADS models using a learned, context-dependent per-trial bias
The computation-through-dynamics perspective argues that biological neural circuits process information via the continuous evolution of their internal states. Inspired by this perspective, Latent Factor Activity using Dynamical systems (LFADS, [1]) identifies a generative model consistent with the neural activity recordings. LFADS models neural dynamics with a recurrent neural network (RNN) generator, which results in excellent fit to the data. However, it has been difficult to understand the dynamics of the LFADS generator. In this work, we show that this poor interpretability arises in part because the generator implements complex, multi-stable dynamics. We introduce a simple modification to LFADS that ameliorates issues with interpretability by providing an inferred per-trial bias (modeled as a constant input) to the RNN generator, enabling it to contextually adapt a simpler dynamical system to individual trials. In both simulated neural recordings from pendulum oscillations and real recordings during arm movements in nonhuman primates, we observed that the standard LFADS learned complex, multi-stable dynamics, whereas the modified LFADS learned easier-to-understand contextual dynamics. This enabled direct analysis of the generator, which reproduced at a single-trial level previous results shown only through more complex analyses at the trial average. Finally, we applied the per-trial inferred bias LFADS model to human intracortical brain computer interface recordings during attempted finger movements and speech. We show that modifying neural dynamics using linear operations of the per-trial bias addresses non-stationarity and identifies the extent of behavioral variability, problems known to plague BCI. We call our modification to LFADS as “contextual LFADS”.
Open AccessImproved interpretability in LFADS models using a learned, context-dependent per-trial bias
The computation-through-dynamics perspective argues that biological neural circuits process information via the continuous evolution of their internal states. Inspired by this perspective, Latent Factor Activity using Dynamical systems (LFADS, [1]) identifies a generative model consistent with the neural activity recordings. LFADS models neural dynamics with a recurrent neural network (RNN) generator, which results in excellent fit to the data. However, it has been difficult to understand the dynamics of the LFADS generator. In this work, we show that this poor interpretability arises in part because the generator implements complex, multi-stable dynamics. We introduce a simple modification to LFADS that ameliorates issues with interpretability by providing an inferred per-trial bias (modeled as a constant input) to the RNN generator, enabling it to contextually adapt a simpler dynamical system to individual trials. In both simulated neural recordings from pendulum oscillations and real recordings during arm movements in nonhuman primates, we observed that the standard LFADS learned complex, multi-stable dynamics, whereas the modified LFADS learned easier-to-understand contextual dynamics. This enabled direct analysis of the generator, which reproduced at a single-trial level previous results shown only through more complex analyses at the trial average. Finally, we applied the per-trial inferred bias LFADS model to human intracortical brain computer interface recordings during attempted finger movements and speech. We show that modifying neural dynamics using linear operations of the per-trial bias addresses non-stationarity and identifies the extent of behavioral variability, problems known to plague BCI. We call our modification to LFADS as “contextual LFADS”.
- Open Access
- 1 April 2025
- Published by IOP Publishing in Journal of Neural Engineering
- Vol. 22 (2) , 026050
5-year follow-up of a fully implanted brain–computer interface in a spinal cord injury patient
Spinal cord injury (SCI) affects over 250 000 individuals in the US. Brain–computer interfaces (BCIs) may improve quality of life by controlling external devices. Invasive intracortical BCIs have shown promise in clinical trials but degrade in the chronic period and tether patients to acquisition hardware. Alternatively, electrocorticography (ECoG) records data from electrodes on the cortex, and studies evaluating fully implanted BCI-ECoG systems are scarce. Objective. We seek to address this need using a fully implanted ECoG-based BCI that allows for home use in SCI. Approach. The patient used a long-term BCI system, initially controlling an functional electrical stimulation orthosis in the lab and later using an external mechanical orthosis at home. To evaluate its long-term viability, electrode contact impedance, signal quality, and decoder performance were measured. Signal quality was assessed using signal-to-noise ratio and maximum bandwidth of the signal. Decoder performance was monitored using the area under the receiver operator characteristic curve (AUROC). Main results. The study analyzed data from the patient’s home environment over 54 months, revealing that the device was used at home for 38 ± 24 min on average daily. After six months, we observed stable event-related desynchronization that aided in determining the onset of motor intention. The decoder’s average AUROC across months was 0.959. Importantly, 40 months of the data collected was gather from the subject’s home or community environment. The results indicate long-term ECoG recordings were stable for motor-imagery classification and motor control in the community environment in a case of an individual with SCI. Significance. This study presents the long-term feasibility and viability of an ECoG-based BCI system that persists in the home environment in a case of SCI. Future research should explore larger electrode counts with more participants to confirm this stability. Understanding these trends is crucial for clinical utility and chronic viability in broader patient populations.
- 1 April 2025
- Published by IOP Publishing in Journal of Neural Engineering
- Vol. 22 (2) , 026050
Open Access5-year follow-up of a fully implanted brain–computer interface in a spinal cord injury patient
Spinal cord injury (SCI) affects over 250 000 individuals in the US. Brain–computer interfaces (BCIs) may improve quality of life by controlling external devices. Invasive intracortical BCIs have shown promise in clinical trials but degrade in the chronic period and tether patients to acquisition hardware. Alternatively, electrocorticography (ECoG) records data from electrodes on the cortex, and studies evaluating fully implanted BCI-ECoG systems are scarce. Objective. We seek to address this need using a fully implanted ECoG-based BCI that allows for home use in SCI. Approach. The patient used a long-term BCI system, initially controlling an functional electrical stimulation orthosis in the lab and later using an external mechanical orthosis at home. To evaluate its long-term viability, electrode contact impedance, signal quality, and decoder performance were measured. Signal quality was assessed using signal-to-noise ratio and maximum bandwidth of the signal. Decoder performance was monitored using the area under the receiver operator characteristic curve (AUROC). Main results. The study analyzed data from the patient’s home environment over 54 months, revealing that the device was used at home for 38 ± 24 min on average daily. After six months, we observed stable event-related desynchronization that aided in determining the onset of motor intention. The decoder’s average AUROC across months was 0.959. Importantly, 40 months of the data collected was gather from the subject’s home or community environment. The results indicate long-term ECoG recordings were stable for motor-imagery classification and motor control in the community environment in a case of an individual with SCI. Significance. This study presents the long-term feasibility and viability of an ECoG-based BCI system that persists in the home environment in a case of SCI. Future research should explore larger electrode counts with more participants to confirm this stability. Understanding these trends is crucial for clinical utility and chronic viability in broader patient populations.
- Open Access
Representation of verbal thought in motor cortex and implications for speech neuroprostheses
Speech brain-computer interfaces show great promise in restoring communication for people who can no longer speak 1–3 , but have also raised privacy concerns regarding their potential to decode private verbal thought 4–6 . Using multi-unit recordings in three participants with dysarthria, we studied the representation of inner speech in the motor cortex. We found a robust neural encoding of inner speech, such that individual words and continuously imagined sentences could be decoded in real-time This neural representation was highly correlated with overt and perceived speech. We investigated the possibility of "eavesdropping" on private verbal thought, and demonstrated that verbal memory can be decoded during a non-speech task. Nevertheless, we found a neural "overtness" dimension that can help to avoid any unintentional decoding. Together, these results demonstrate the strong representation of verbal thought in the motor cortex, and highlight important design considerations and risks that must be addressed as speech neuroprostheses become more widespread.
Open AccessRepresentation of verbal thought in motor cortex and implications for speech neuroprostheses
Speech brain-computer interfaces show great promise in restoring communication for people who can no longer speak 1–3 , but have also raised privacy concerns regarding their potential to decode private verbal thought 4–6 . Using multi-unit recordings in three participants with dysarthria, we studied the representation of inner speech in the motor cortex. We found a robust neural encoding of inner speech, such that individual words and continuously imagined sentences could be decoded in real-time This neural representation was highly correlated with overt and perceived speech. We investigated the possibility of "eavesdropping" on private verbal thought, and demonstrated that verbal memory can be decoded during a non-speech task. Nevertheless, we found a neural "overtness" dimension that can help to avoid any unintentional decoding. Together, these results demonstrate the strong representation of verbal thought in the motor cortex, and highlight important design considerations and risks that must be addressed as speech neuroprostheses become more widespread.
- Open Access
- 6 May 2021
- Published by Public Library of Science (PLoS) in PLOS ONE
- Vol. 16 (5) , e0246165
Machine learning of EEG spectra classifies unconsciousness during GABAergic anesthesia
In current anesthesiology practice, anesthesiologists infer the state of unconsciousness without directly monitoring the brain. Drug- and patient-specific electroencephalographic (EEG) signatures of anesthesia-induced unconsciousness have been identified previously. We applied machine learning approaches to construct classification models for real-time tracking of unconscious state during anesthesia-induced unconsciousness. We used cross-validation to select and train the best performing models using 33,159 2s segments of EEG data recorded from 7 healthy volunteers who received increasing infusions of propofol while responding to stimuli to directly assess unconsciousness. Cross-validated models of unconsciousness performed very well when tested on 13,929 2s EEG segments from 3 left-out volunteers collected under the same conditions (median volunteer AUCs 0.99-0.99). Models showed strong generalization when tested on a cohort of 27 surgical patients receiving solely propofol collected in a separate clinical dataset under different circumstances and using different hardware (median patient AUCs 0.95—0.98), with model predictions corresponding with actions taken by the anesthesiologist during the cases. Performance was also strong for 17 patients receiving sevoflurane (alone or in addition to propofol) (median AUCs 0.88—0.92). These results indicate that EEG spectral features can predict unconsciousness, even when tested on a different anesthetic that acts with a similar neural mechanism. With high performance predictions of unconsciousness, we can accurately monitor anesthetic state, and this approach may be used to engineer infusion pumps to intelligibly respond to patients’ neural activity.
- 6 May 2021
- Published by Public Library of Science (PLoS) in PLOS ONE
- Vol. 16 (5) , e0246165
Open AccessMachine learning of EEG spectra classifies unconsciousness during GABAergic anesthesia
In current anesthesiology practice, anesthesiologists infer the state of unconsciousness without directly monitoring the brain. Drug- and patient-specific electroencephalographic (EEG) signatures of anesthesia-induced unconsciousness have been identified previously. We applied machine learning approaches to construct classification models for real-time tracking of unconscious state during anesthesia-induced unconsciousness. We used cross-validation to select and train the best performing models using 33,159 2s segments of EEG data recorded from 7 healthy volunteers who received increasing infusions of propofol while responding to stimuli to directly assess unconsciousness. Cross-validated models of unconsciousness performed very well when tested on 13,929 2s EEG segments from 3 left-out volunteers collected under the same conditions (median volunteer AUCs 0.99-0.99). Models showed strong generalization when tested on a cohort of 27 surgical patients receiving solely propofol collected in a separate clinical dataset under different circumstances and using different hardware (median patient AUCs 0.95—0.98), with model predictions corresponding with actions taken by the anesthesiologist during the cases. Performance was also strong for 17 patients receiving sevoflurane (alone or in addition to propofol) (median AUCs 0.88—0.92). These results indicate that EEG spectral features can predict unconsciousness, even when tested on a different anesthetic that acts with a similar neural mechanism. With high performance predictions of unconsciousness, we can accurately monitor anesthetic state, and this approach may be used to engineer infusion pumps to intelligibly respond to patients’ neural activity.
- Top Cited PapersOpen Access
- 2 January 2020
- Published by Springer Nature in Nature Communications
- Vol. 11 (1) , 1-15
An early phase of instructive plasticity before the typical onset of sensory experience
While early experience with moving stimuli is necessary for the development of direction selectivity in visual cortex of carnivores, it is unclear whether experience exerts a permissive or instructive influence. To test if the specific parameters of the experienced stimuli could instructively sculpt the emergent responses, visually naive ferrets were exposed to several hours of experience with unusual spatiotemporal patterns. In the most immature ferrets, cortical neurons developed selectivity to these patterns, indicating an instructive influence. In animals that were 1–10 days more mature, exposure to the same patterns led to a developmentally-typical increase in direction selectivity. We conclude that visual development progresses via an early phase of instructive plasticity, when the specific patterns of neural activity shape the specific parameters of the emerging response properties, followed by a late phase of permissive maturation, when sensory-driven activity merely serves to enhance the response properties already seeded in cortical circuits.
- 2 January 2020
- Published by Springer Nature in Nature Communications
- Vol. 11 (1) , 1-15
Top Cited PapersOpen AccessAn early phase of instructive plasticity before the typical onset of sensory experience
While early experience with moving stimuli is necessary for the development of direction selectivity in visual cortex of carnivores, it is unclear whether experience exerts a permissive or instructive influence. To test if the specific parameters of the experienced stimuli could instructively sculpt the emergent responses, visually naive ferrets were exposed to several hours of experience with unusual spatiotemporal patterns. In the most immature ferrets, cortical neurons developed selectivity to these patterns, indicating an instructive influence. In animals that were 1–10 days more mature, exposure to the same patterns led to a developmentally-typical increase in direction selectivity. We conclude that visual development progresses via an early phase of instructive plasticity, when the specific patterns of neural activity shape the specific parameters of the emerging response properties, followed by a late phase of permissive maturation, when sensory-driven activity merely serves to enhance the response properties already seeded in cortical circuits.