Electroencephalography Biomarkers

The measurable signal reflects coordinated postsynaptic activity across populations of cortical neurons rather than the activity of a single cell. Electrode recordings capture this collective electrical behavior, which can then be analyzed for recurring patterns. These patterns provide engineering systems with measurable indicators of neural states or processes, although their reliability depends on how consistently the underlying activity is recorded.

These measures describe different properties of the same EEG data. Frequency-band power summarizes activity distributed across frequency ranges, while event-related potentials characterize responses associated with events. Connectivity examines relationships among recorded signals, and signal complexity quantifies aspects of signal structure. Selecting among them depends on whether an application needs state characterization, event detection, relational information, or a broader description of signal behavior.

EEG biomarker estimates can be affected by unwanted components introduced during signal acquisition, so artifact reduction helps separate relevant neural patterns from contamination. Validation then tests whether an extracted feature reliably characterizes the intended neural state or process. Together, these steps reduce the risk that an engineering system bases classifications or device-control commands on recording conditions rather than meaningful brain activity.

The choice should match the information required by the system. Frequency-band power may support analysis of activity across frequency ranges, event-related potentials may be appropriate when responses are linked to events, and connectivity or complexity may provide relational or structural descriptions. Comparing these representations helps engineers identify features that best support the intended classification, communication function, or control task.

A basic workflow begins with EEG signal acquisition through scalp electrodes, followed by artifact reduction to improve signal quality. Engineers then calculate selected features, such as power, event-related potentials, connectivity, or complexity, and validate their relationship to the neural state or process of interest. The resulting features can support classification or translation into device-control commands.

They are useful when an engineering system must infer meaningful information from neural activity without relying solely on conventional physical input. Extracted features can help characterize cognitive or neurological function, support brain-computer interfaces, and contribute to assistive communication technologies. In these settings, the biomarker serves as an input to systems designed to classify neural activity or control a device.

Analysis can produce quantitative features for characterizing neural states or processes and can support classifications derived from recorded brain activity. In neurotechnology, these outputs may be translated into meaningful system inputs or device-control commands. Their value depends on careful acquisition, feature analysis, artifact reduction, and validation, which connect the recorded signal to the intended cognitive, neurological, or engineering outcome.