Invasive Brain-Computer Interfaces for Attention

The goal of this interventional study is to compare if the use of a brain-machine interface (BCI) therapy can improve the symptoms of attentional deficit by producing brain changes in the networks that modulate attention. The investigators intend to work with epileptic participants who do not respond to pharmacological treatment, who will undergo neurosurgery. The questions the study sets out to answer are:

1. is there an improvement of symptoms in an experimental group receiving the treatment versus a sham group receiving a simulation of the treatment?
2. does the application of the therapy before surgery reduce the recovery times of post-surgery cognitive deficits described in the literature?

Making use of the information recorded from brain electrodes implanted before a participant's epilepsy surgery, the investigators will create a BCI decoder that works with the available activity sources to establish the level of attention of each participant when performing tasks. Participants:

• will perform an offline phase first, which will consist of one day of evaluation, in which they will be familiarized with an attentional task.
• will perform a training phase later, which will consist of several days of evaluation, where they will learn to modulate their level of attention. This modulation will be facilitated by the BCI decoder, which will classify the level of attention directly from the brain and provide visual feedback that the participant will use as a guide.

If the participant is part of the experimental group (or BCI group), the feedback will work as described and should be easy to follow, but if the participant is part of the Sham group, the feedback will not work according to the brain activity of the actual participant, but according to that of another person. Because of this, a mismatch will be created between the moments a brain experiences inattention, and participants believe they are experiencing inattention.

This is a randomized, double-blind study, in which the experimenters will evaluate how the effect of the attentional therapy with BCI affects an BCI group and a Sham group.

Study Overview

• Status: Withdrawn
• Conditions: Epilepsy; Epilepsy in Children
• Intervention / Treatment: Behavioral: Attention Intervention using a customized BCI decoder; Behavioral: Attention Intervention without using a customized BCI decoder

Detailed Description

This research differs from others available in that it is among the first of its kind to be performed on participants with invasive electrodes in a hospital setting. Additionally, it focuses on epileptic participants, who already have a set of invasive electrodes in place, so there is no need for any additional surgical intervention.

Also, the age range of participants for the study (between 8 and 21 years old) usually presents a high incidence of attentional disorders, so it is considered a good group to carry out this research.

This research does not require any additional intervention of any kind, except for the participant willingness to participate, with the possibility of improving their baseline attentional level, or at least of recovering their baseline attentional level faster after surgery, which usually decreases it.

• Study Type: Interventional
• Phase: Not Applicable

Contacts and Locations

Study Locations

• United States
  • Texas
    • Austin, Texas, United States, 78723
      • Dell Children's Medical Center

Participation Criteria

Eligibility Criteria

• Ages Eligible for Study: Child; Adult
• Accepts Healthy Volunteers: No

Description

Inclusion Criteria:

• Children and adolescents (8-21 years)
• Confirmed diagnosis of drug-refractory epilepsy
• iEEG implants on the GK network (ventro-lateral prefrontal cortex and executive network). Also desirable in areas related with attention and Action Phase processing
• Normal to corrected vision
• Ability to understand instructions to follow protocols
• Able to read and understand English or Spanish (all evaluations will be conducted depending on the mother tongue of the participant)
• Able to assent together with his/her legal guardian (below 18 years old) or approve (18 years old or older) informed consent

Exclusion Criteria:

• Prior history of seizure focus removal
• Prior history of ischemic or hemorrhagic stroke
• Prior history of traumatic brain injury
• Prior history of color blindness
• Intracranial implants
• Headaches disorders
• Neurological infections
• Neurological pain or malnutrition disorders
• Severe mental disorders: depression, anxiety, among other psychiatric diseases
• Severe intellectual and learning disabilities
• Compromised consciousness
• Severe physical impairment (i.e. inability to mobilize upper extremities by oneself)
• Severe co-morbidities (active cancer within 5 years, cardiovascular diseases, severe metabolic diseases, hepatic or kidney failure, recent major surgery, infectious diseases)
• Substance or alcohol abuse
• Pregnancy
• Criteria identified in safety guidelines for MRI, in particular metallic implants. Participants who are unable to perform MRI will be completely excluded from the study

Study Plan

How is the study designed?

Design Details

• Primary Purpose: Treatment
• Allocation: Randomized
• Interventional Model: Parallel Assignment
• Masking: Triple

Number of Arms

2

Arms and Interventions

Participant Group / Arm	Intervention / Treatment
Experimental: This arm will receive visual feedback controlled by its own BCI	Behavioral: Attention Intervention using a customized BCI decoder; During the offline phase of the intervention, participants will perform an attentional task while intracranial brain activity is recorded. Data from this session will be used to train a personalized decoder capable of classifying attentional engagement. During the training phase, participants will receive real-time visual feedback contingent on their brain activity when attentional engagement is detected. This closed-loop feedback aims to reinforce successful attention and enhance performance over repeated sessions.
Sham Comparator: This arm will receive visual feedback from another randomly selected participant	Behavioral: Attention Intervention without using a customized BCI decoder; During the offline phase, participants will perform an attentional task while intracranial brain activity is recorded. A personalized decoder will be created for each participant but will not be used during the training phase sessions. During the training phase, participants will receive visual feedback while performing attentional tasks; however, the feedback will not be contingent on their brain activity. Instead, feedback will be non-contingent and unrelated to actual attentional engagement. This group is not expected to experience improvements in attentional performance through the training sessions.

What is the study measuring?

Primary Outcome Measures

Outcome Measure	Measure Description	Time Frame
Sustained attention as assessed by Conners Continuous Performance Test, 3rd Edition	The CPT-3 is a task-oriented computerized assessment used to evaluate attention-related problems in individuals aged 8 years and older. The test provides objective information about an individual's performance in attention tasks. T-scores: ~30 (Min) / 100+ (Max); higher score = worse performance; Confidence Index: 0 (Min) / 1.00 (Max); higher score = worse performance (closer to 1 = more likely atypical)	Perioperative/periprocedural, and 3 months after hospital discharge
Working memory as assessed by WISC-IV/WISC-V or WAIS-III/WAIS-IV	The WISC is a standardized test used to assess the intellectual ability of children aged 6 to 16 years. It evaluates multiple cognitive domains, including verbal comprehension, visual-spatial reasoning, working memory, processing speed, and fluid reasoning (added in WISC-V). The WAIS is the adult counterpart to the WISC and is used to assess the intelligence of individuals aged 16 to 90 years. Like the WISC, it evaluates cognitive abilities across domains such as verbal comprehension, perceptual reasoning, working memory, and processing speed. Full Scale IQ & Index Scores: 40 (Min) / 160+ (Max); higher score = better cognitive ability; Subtest Scaled Scores: 1 (Min) / 19 (Max); higher score = better performance on that subtest	Perioperative/periprocedural, and 3 months after hospital discharge
Executive function as assessed by Verbal Fluency and Trail Making, of the Delis-Kaplan Executive Function System	The D-KEFS Verbal Fluency evaluates an individual's verbal productivity, cognitive flexibility, and executive control over language. It includes Letter Fluency, Category Fluency, and Category Switching. The D-KEFS Trail Making assesses visual attention, psychomotor speed, sequencing, cognitive flexibility, and set-shifting. The Trail Making subtest is useful for detecting executive dysfunction and is frequently used in evaluating individuals with brain injuries, neurodevelopmental disorders, and neurodegenerative conditions. Verbal Fluency, Scaled Score: 1 (Min) / 19 (Max); higher score = better performance; Verbal Fluency, Error Rates 0 - ∞ (raw count); higher score = worse performance; Trail Making, Scaled Score: 1 (Min) / 19 (Max); higher score = better performance; Trail Making, Error Rates 0 - ∞ (raw count); higher score = worse performance	Perioperative/periprocedural, and 3 months after hospital discharge
Executive function as assessed by Behavior Rating Inventory of Executive Function, 2nd edition	The BRIEF-2 is a standardized questionnaire-based assessment designed to evaluate executive function behaviors in everyday settings. It is typically completed by parents, teachers, or the individual (self-report) and is used for children and adolescents aged 5 to 18 years. It assesses multiple domains of executive functioning-such as inhibition, working memory, emotional control, task initiation, and cognitive flexibility-and provides composite scores like the Behavioral Regulation Index, Emotion Regulation Index, and Cognitive Regulation Index. T-scores (Scales & Indexes): 30 (Min) / 100+ (Max); higher score = worse executive functioning; Global Executive Composite (GEC): 30 (Min) / 100+ (Max); higher score = worse executive functioning	Perioperative/periprocedural, and 3 months after hospital discharge
Subjective Attention Self-Report Visual Analog Scale (SASR-VAS)	The SASR-VAS is a brief, self-administered tool to assess how the participant experiences attentional problems and whether they have noticed any changes over time. The scale consists of single-item visual analog ratings, scored on a 0 to 10 Visual Analog Scale (VAS). Scores: 0 (Min) / 10 (Max); higher score = better perceived attentional functioning and greater perceived improvement	Perioperative/periprocedural, and 3 months after hospital discharge
Markers of plasticity as assessed by Functional MRI	Participants will undergo two fMRI sessions to assess changes in brain activity patterns associated with neuroplasticity. Each session will be conducted before and after the whole intervention. fMRI session will be divided into two parts: an initial resting phase, and an intervention phase (where participants will perform the same tasks as during the intervention/online sessions). By measuring blood-oxygen-level-dependent (BOLD) signals during rest and/or task performance, fMRI can identify percent (%) signal changes in specific Regions of Interest (ROI) over time.	Perioperative/periprocedural
Change on activity as assessed by iEEG Recordings: Attentive	The investigators can measure changes in participants' brain activity to understand how the brain responds to training. The target activity consists of the increase in amplitude (µV) or power (µV²/dB) in the high-gamma frequency [50-150] Hz in Regions of Interest (ROI) around the dorsolateral prefrontal (dlPFC) and ventrolateral prefrontal (vlPFC) cortices during attention trials.	Perioperative/periprocedural
Change on activity as assessed by iEEG Recordings: Resting	The investigators can measure changes in participants' brain activity to understand how the brain responds to training. The target activity consists of the decrease or lack in amplitude (µV) or power (µV²/dB) in the high-gamma frequency [50-150] Hz in Regions of Interest (ROI) around the dorsolateral prefrontal (dlPFC) and ventrolateral prefrontal (vlPFC) cortices during rest trials.	Perioperative/periprocedural

Secondary Outcome Measures

Outcome Measure	Measure Description	Time Frame
BCI Performance: Accuracy	The investigators can calculate different contingency metrics, based on how well each participant's decoder (either from the BCI or Sham group) interprets their attentional intent or mental state. All metrics are extracted from confusion matrices. Between the most relevant, Accuracy: 0 (Min) / 1 (Max); lower score = model makes all predictions incorrectly, higher score = model makes all predictions correctly	Perioperative/periprocedural
BCI Performance: Sensitivity	The investigators can calculate different contingency metrics, based on how well each participant's decoder (either from the BCI or Sham group) interprets their attentional intent or mental state. All metrics are extracted from confusion matrices. Between the most relevant, Sensitivity: 0 (Min) / 1 (Max); lower score = model misses all actual positives (only false negatives), higher score = model detects all actual positives (no false negatives)	Perioperative/periprocedural
BCI Performance: Specificity	The investigators can calculate different contingency metrics, based on how well each participant's decoder (either from the BCI or Sham group) interprets their attentional intent or mental state. All metrics are extracted from confusion matrices. Between the most relevant, Specificity: 0 (Min) / 1 (Max); lower score = model misses all actual negatives (only false positives), higher score = model correctly identifies all negatives (no false positives)	Perioperative/periprocedural
BCI Performance: Precision	The investigators can calculate different contingency metrics, based on how well each participant's decoder (either from the BCI or Sham group) interprets their attentional intent or mental state. All metrics are extracted from confusion matrices. Between the most relevant, Precision: 0 (Min) / 1 (Max); lower score = all predicted positives are wrong (only false positives), higher score = all predicted positives are correct (no false positives)	Perioperative/periprocedural
BCI Performance: F1 Score	The investigators can calculate different contingency metrics, based on how well each participant's decoder (either from the BCI or Sham group) interprets their attentional intent or mental state. All metrics are extracted from confusion matrices. Between the most relevant, F1 Score: 0 (Min) / 1 (Max); lower score = no balance between precision and sensitivity (either is 0), higher score = perfect balance of precision and sensitivity (both = 1)	Perioperative/periprocedural
BCI Performance: Matthews Correlation Coefficient (MCC)	The investigators can calculate different contingency metrics, based on how well each participant's decoder (either from the BCI or Sham group) interprets their attentional intent or mental state. All metrics are extracted from confusion matrices. Between the most relevant, Matthews Correlation Coefficient (MCC): -1 (Min) / 0 (Chance Level) / +1 (Max); lower score = perfect inverse prediction of the model, higher score = perfect prediction of the model	Perioperative/periprocedural
Experiment Performance: Correct Trials per Session	Correct Trials per Session will be assessed according to the number of correct answers per run and session. There will be multiple trials within a run, and several runs within a session. Session data will be used to evaluate: within-subject changes over time (e.g., session-to-session improvements), using dependent comparisons; and between-group differences, particularly comparing participants in the BCI intervention group and the Sham control group, using independent comparisons. Number of trials per run: 0 (Min) / 20 (Max); higher score = greater engagement or attentional control within a run Number of runs per session: ~ 4 (Min) / 8+ (Max); higher score = greater engagement or attentional control within a session Number of sessions per participant: ~ 3 (Min) / 7+ (Max); higher score = greater engagement or attentional control between sessions	Perioperative/periprocedural

Collaborators and Investigators

• Sponsor: University of Texas at Austin
• Investigators: Principal Investigator: Diego Mac-Auliffe, Postdoc, The University of Texas at Austin

Publications and helpful links

General Publications

• Ezzyat Y, Wanda PA, Levy DF, Kadel A, Aka A, Pedisich I, Sperling MR, Sharan AD, Lega BC, Burks A, Gross RE, Inman CS, Jobst BC, Gorenstein MA, Davis KA, Worrell GA, Kucewicz MT, Stein JM, Gorniak R, Das SR, Rizzuto DS, Kahana MJ. Closed-loop stimulation of temporal cortex rescues functional networks and improves memory. Nat Commun. 2018 Feb 6;9(1):365. doi: 10.1038/s41467-017-02753-0.
• Ossandon T, Vidal JR, Ciumas C, Jerbi K, Hamame CM, Dalal SS, Bertrand O, Minotti L, Kahane P, Lachaux JP. Efficient "pop-out" visual search elicits sustained broadband gamma activity in the dorsal attention network. J Neurosci. 2012 Mar 7;32(10):3414-21. doi: 10.1523/JNEUROSCI.6048-11.2012.
• Reilly C, Atkinson P, Das KB, Chin RF, Aylett SE, Burch V, Gillberg C, Scott RC, Neville BG. Neurobehavioral comorbidities in children with active epilepsy: a population-based study. Pediatrics. 2014 Jun;133(6):e1586-93. doi: 10.1542/peds.2013-3787.
• Perrone-Bertolotti M, El Bouzaidi Tiali S, Vidal JR, Petton M, Croize AC, Deman P, Rheims S, Minotti L, Bhattacharjee M, Baciu M, Kahane P, Lachaux JP. A real-time marker of object-based attention in the human brain. A possible component of a "gate-keeping mechanism" performing late attentional selection in the Ventro-Lateral Prefrontal Cortex. Neuroimage. 2020 Apr 15;210:116574. doi: 10.1016/j.neuroimage.2020.116574. Epub 2020 Jan 23.
• Mac-Auliffe D, Chatard B, Petton M, Croize AC, Sipp F, Bontemps B, Gannerie A, Bertrand O, Rheims S, Kahane P, Lachaux JP. The Dual-Task Cost Is Due to Neural Interferences Disrupting the Optimal Spatio-Temporal Dynamics of the Competing Tasks. Front Behav Neurosci. 2021 Aug 19;15:640178. doi: 10.3389/fnbeh.2021.640178. eCollection 2021.

Study record dates

Study Major Dates

• Study Start (Estimated): May 1, 2025
• Primary Completion (Estimated): April 1, 2026
• Study Completion (Estimated): December 1, 2026

Study Registration Dates

• First Submitted: March 19, 2025
• First Submitted That Met QC Criteria: April 18, 2025
• First Posted (Actual): April 23, 2025

Study Record Updates

• Last Update Posted (Actual): May 1, 2026
• Last Update Submitted That Met QC Criteria: April 27, 2026
• Last Verified: April 1, 2026

More Information

Terms related to this study

• Keywords: Brain-Computer Interface; Epilepsy; Attention; Brain Computer Interface; Attention deficit; BCI
• Additional Relevant MeSH Terms: Brain Diseases; Central Nervous System Diseases; Nervous System Diseases; Mental Disorders; Neurodevelopmental Disorders; Attention Deficit and Disruptive Behavior Disorders; Epilepsy; Attention Deficit Disorder with Hyperactivity
• Other Study ID Numbers: STUDY00003744

Plan for Individual participant data (IPD)

• Plan to Share Individual Participant Data (IPD)?: YES
• IPD Plan Description: All anonymized data will be made available upon request by the online publication date
• IPD Sharing Time Frame: Data will be made available by the online publication date
• IPD Sharing Access Criteria: Data will be shared upon request to the PIs

Drug and device information, study documents

• Studies a U.S. FDA-regulated drug product: No
• Studies a U.S. FDA-regulated device product: No
• product manufactured in and exported from the U.S.: No