A Protocol For Uncovering Neural Mechanisms Of Neurotherapeutic Effects On Electroencephalography Using The Human Neocortical Neurosolver
May 19th, 2026
This protocol demonstrates how physics-based neural simulations can be used to interpret electrophysiological biomarkers of neurotherapeutics and uncover their effect on neural circuits, providing a mechanistically grounded approach for neurotherapeutic development.
We develop models of brain circuitry that simulate the neural generators of EEG and apply them to discover how neurotherapeutics impact brain dynamics. Current methods cannot fully explain how neurotherapeutics alter brain circuits. This protocol combines EEG with circuit simulations to reveal underlying mechanisms.
To begin, identify the treatment-induced electroencephalography event-related potential, or EEG ERP biomarker, and initialize the default Human Neocortical Neurosolver, or HNN, model. Then, start with the canonical HNN event-related potential simulation and its default parameters. Manually tune the scaling factor and exogenous drive parameters to fit the pre-treatment event-related potential.
Click the Load Data button on the menu bar located on the lower left portion of the Graphical User Interface window. In the File Browser window, select a CSV file or text file containing the event-related potential waveform to be modeled as the target waveform. Ensure the file is comma-delimited with two columns, time and source localized empirical dipole waveform.
In the Simulation Parameters panel, set the desired values of tstop to control the simulation length, dt for the integration time step, and trials for the number of repeated simulations with identical model parameters. Select Backend as either Joblib for embarrassingly parallel simulations over trials or MPI for serial simulations sped up with parallelization over computer cores. Click the Run button to start the default simulation of a canonical event-related potential.
Click the Visualization tab on the top left of the Graphical User Interface window. Then, click the dropdown menu labeled Data to compare and select the loaded target waveform from the previous step. Click Clear Axes to reset the plot.
Click Add Plot to generate a new plot with the simulated initial ERP waveform and the target waveform overlaid together. Display the automatically calculated correlation coefficient and root mean squared error between the two waveforms. Next, modify the scaling factor by manual hand tuning to approximately match the magnitudes of the simulated and empirical dipole wave forms.
For this simulation, set the default dipole scaling parameter in the Simulation tab to 3, 000. Identify the simulated ERP peaks that are misaligned in time with the empirical ERP waveform. Click the External Drives tab on the top left of the Graphical User Interface window.
Click the dropdown menu of the exogenous drive, whose mean time is closest to the misaligned peak. Modify the values in the mean time and standard deviation time text boxes to better match the timing and width of peaks in the target waveform. Set the mean time of the evprox1 external drive to 60 milliseconds, evdist1 to 100 milliseconds, and evprox2 to 150 milliseconds.
Next, manually tune the synaptic weights of exogenous drives to obtain a closer fit to the magnitudes of empirically-recorded ERP peaks. Identify the simulated ERP peaks that are misaligned in magnitude with the empirical ERP waveform. Click the External Drives tab on the top left of the Graphical User Interface window.
Click the dropdown menu of the exogenous drive, whose mean time is closest to the misaligned peak. Set the AMPA weights of the evdist1 drive to 0.01423 for L5_pyramidal and 0.0000007 for L2_pyramidal. Assign the NMDA weights of the evdist1 drive as 0.0080074 for L5_pyramidal and 0.0004317 for L2_pyramidal.
For the evprox2 drive, configure the AMPA weights to 0.0684013 for L5_pyramidal and 0.143884 for L2_pyramidal. After completing the modifications in the previous steps, click the Simulation tab and enter pre-treatment_handtuned in the Name text box. Click the Run button to simulate the modified parameter set.
Select the data to compare dropdown menu and regenerate the plot to compare the new simulation to the loaded data. Inspect the generated plot in the Figure panel. Access previous plots using the corresponding Figure tabs and save the model outputs.
Next, establish the post-treatment model fit and perform uncertainty quantification using a simulation-based interface, along with a separability assessment. Finally, examine, validate, and further constrain the model. After applying the simulation-based inference workflow, the parameter distributions for pre-treatment and post-treatment ERPs were visualized using a pair plot.
Thalamocortical synchrony exhibited the greatest pre-to post-treatment separability with the lowest overlap value of 0.07, which increased following treatment. Pre-treatment waveforms closely matched simulations generated from parameter distribution samples with a correlation coefficient of 0.99. Post-treatment waveforms closely matched simulations generated from parameter distribution samples with a correlation coefficient of 0.96.
Simulated ERPs sampled from pre-treatment and post-treatment parameter distributions were shown alongside corresponding cell-specific spiking activity. Reduced spiking activity in Layer 5 pyramidal neurons was observed following treatment. Our simulations enable multi-scale interpretation of EEG generation and can be used to build evidence about the mechanisms of neurotherapeutics on brain circuits.
The simulation experiments require well-grounded hypotheses on potential mechanisms of neurotherapeutics, as they provide targets for model fitting and examination. Future studies using invasive electrophysiology or certain types of imaging can inform modeling supporting an iterative simulation and validation process.
Overview
This article presents a protocol utilizing the Human Neocortical Neurosolver (HNN), an open-source biophysical modeling software, to investigate the neural mechanisms underlying EEG biomarkers. The workflow enables researchers to link EEG signals to their multiscale neural generators, facilitating the study of neurotherapeutic effects and the development of testable predictions for central nervous system disorders.
Key Study Components
Area of Science
- Neuroscience
- Electrophysiology
- Computational modeling
Background
- EEG and electrophysiological methods provide high temporal resolution biomarkers for CNS disorders.
- Understanding the neural mechanisms generating these biomarkers is limited.
- Improved mechanistic insight is crucial for developing diagnostics and therapeutics based on EEG signals.
- HNN software links EEG biomarkers to their underlying neural circuit activity.
Purpose of Study
- To demonstrate a hypothesis-driven workflow using HNN for testing neural mechanisms of neurotherapeutic-induced EEG biomarkers.
- To optimize model parameters for fitting simulated and empirical current source waveforms.
- To visualize and quantify multiscale cell- and circuit-level activity corresponding to EEG biomarkers.
Methods Used
- Biophysical modeling with the Human Neocortical Neurosolver (HNN).
- Parameter optimization to fit simulated data to empirical EEG current source waveforms.
- Visualization and quantification of cell- and circuit-level neural activity.
- Application to auditory evoked responses (P1, N1, P2 components) and neurotherapeutic-induced changes.
Main Results
- Demonstrated workflow for linking EEG biomarkers to neural circuit mechanisms using HNN.
- Example analysis of early event-related potential components in auditory evoked responses.
- Assessment of neurotherapeutic-induced modifications in neural circuit activity.
- Generation of testable predictions for follow-up empirical studies.
Conclusions
- The protocol enables simulation experiments that connect EEG biomarkers to underlying neural mechanisms.
- Provides validation targets for empirical studies based on model predictions.
- The workflow is adaptable for studying disease mechanisms or other therapeutic interventions.
Frequently Asked Questions
Executive Industry Relevance
Mechanistic de-risking of neurotherapeutic strategies requires linking EEG biomarkers to their underlying neural circuit generators. The Human Neocortical Neurosolver (HNN) protocol enables hypothesis-driven modeling to clarify how neurotherapeutic interventions modulate specific EEG features, supporting predictive confidence in CNS drug discovery. This workflow advances translational continuity by generating testable predictions for empirical validation and portfolio triage.
Strategic Applications in Biopharma R&D
Early Discovery & Target Validation
- Enables interrogation of therapeutic hypotheses by simulating neurotherapeutic effects on EEG biomarkers.
- Supports biological de-risking through mechanistic mapping of EEG signals to neural circuit activity.
- Facilitates functional target validation by quantifying cell- and circuit-level contributions to observed biomarkers.
- Improves predictive confidence for advancing CNS targets in the discovery pipeline.
Screening & Assay Development
- Prepares validated computational models for downstream screening of neuroactive compounds.
- Standardizes quantitative EEG biomarker outputs for reproducible assay development.
- Enables scalable simulation experiments to evaluate compound effects on neural circuit mechanisms.
- Supports reliable evaluation of candidate interventions by linking model outputs to empirical EEG data.
Translational & Preclinical Research
- Aligns EEG biomarker changes with disease-relevant neural mechanisms for translational biomarker development.
- Provides continuity from discovery-stage modeling to preclinical validation of neurotherapeutic effects.
- Enables risk-adjusted advancement decisions by generating mechanistically grounded predictions for empirical testing.
- Supports predictive de-risking of CNS portfolios by clarifying circuit-level drug effects.
Pipeline & Workflow Integration
The HNN protocol integrates from early discovery through preclinical research by connecting EEG biomarker analysis to mechanistic modeling and empirical validation.
- Discovery Biology: Supports hypothesis testing and pathway clarification by simulating neurotherapeutic modulation of EEG features.
- Screening: Provides reproducible, quantitative EEG biomarker outputs for compound evaluation.
- Analytics: Delivers cell- and circuit-level readouts to compare intervention effects across conditions.
- Translational Research: Aligns model predictions with disease-relevant EEG biomarkers for preclinical continuity.
- Enterprise Reuse: Establishes a reusable modeling framework adaptable to diverse CNS targets and interventions.
Operational & Enterprise Impact
- Scientific Value: Increases predictive confidence and reduces mechanistic ambiguity in CNS target validation.
- Operational Value: Standardizes modeling and analysis workflows for reproducibility and scalability.
- Strategic Value: Enables informed go/no-go decisions and capital-efficient portfolio management.
- Portfolio Impact: Supports risk-adjusted prioritization and advancement of neurotherapeutic programs.
Implementation Considerations
- Requires expertise in computational neuroscience and EEG biomarker analysis.
- Needs access to HNN software and supporting computational infrastructure.
- Demands cross-team standardization of model parameters and empirical data integration.
- Adaptable across different CNS disease models and neurotherapeutic interventions.
- Dependent on quality and resolution of empirical EEG data for model fitting.