CLAS with ZMax and a special 3rd party solution

Closed-Loop Auditory Stimulation with ZMax

Closed-loop auditory stimulation (CLAS) is an established method for modulating slow-wave activity during deep sleep. By detecting ongoing slow oscillations and delivering brief auditory stimuli at a specific phase - typically the up-state - researchers can influence sleep dynamics in a controlled and time-locked manner. Despite its scientific promise, CLAS has traditionally required custom signal-processing pipelines, real-time programming expertise, and laboratory-grade EEG systems. Implementing reliable phase-locked stimulation often meant building and validating a complex technical setup before the first participant could even be tested.

Today, that barrier is significantly lower.

When combined with a real-time research platform for multimodal brain and behavioral studies, ZMax becomes a ready-to-run solution for slow-wave CLAS. Automatic slow-wave detection, predictive phase tracking, and latency-compensated stimulation are integrated into a configurable research workflow that can run on standard PCs or portable tablets. No custom MATLAB scripts. No external DSP toolboxes. No complex hardware integration. CLAS protocols can be configured in minutes. Labs already running ZMax can add closed-loop stimulation in software alone - the same headband, the same recording workflow.

A complete closed-loop solution: ZMax streams EEG to a laptop running the CLAS software, which detects slow waves and locks audio tones to a selected target phase - all on the same machine.


Selective Stimulation During Deep Sleep

Closed-loop stimulation is only meaningful if it is delivered during the appropriate sleep stage. The software continuously tracks slow delta waves and restricts stimulation to deep NREM sleep, enabling controlled and reproducible CLAS protocols.

Overnight hypnogram (white) with stimulation events (green) clustered during deep sleep. The heatmap at the bottom shows the power in the delta range over time. Data in this and the following figures come from an expert-scored overnight recording from a sleep database, replayed through the CLAS software.

Selectivity is near-total: in the above example, 98 % of the 3368 stimuli were delivered in deep sleep (N3-N4), and fewer than 2 % in light sleep.

Precise Phase Targeting of Slow Oscillations

Once slow waves are detected, the software predicts the optimal moment within the oscillation and delivers auditory stimulation at the selected target phase(s) - for example, at the up-state peak used in classical CLAS protocols. Phase locking operates with millisecond resolution and automatically compensates for hardware latencies to ensure accurate biological timing.

Zoomed slow-wave segment showing the EEG signal (beige: raw; orange: filtered) and phase-locked stimuli (green).

Real-Time Control and Monitoring

Each stimulus comes from a model fitted to the ongoing oscillation, and various quality thresholds can be applied before a stimulus is allowed. Accuracy and precision are reported live, so a protocol can be checked and tuned while the session runs.

The phase-locking panel during a live session in the 8-12 Hz band, targeting the middle of the rising slope: stimulation thresholds, accuracy statistics and controls are on the left, while the fitted model, filtered input signal and session statistics are on the right.t.

Quantifying Phase-Locking Accuracy

Every stimulus is logged with the phase it actually hit, so locking accuracy is measured rather than assumed. The companion analysis tool provides a full offline analysis of the night; among its outputs, two numbers summarize phase-locking performance - how many stimuli were delivered, and how tightly they clustered around the target phase.

The two trade off against each other. Relaxed thresholds accept weaker, less regular waves - more stimulations, wider spread. Strict thresholds admit only well-formed oscillations - tighter phase, fewer stimulations. Researchers set the operating point to suit the study, and can test it offline on an existing recording before running a participant.

The same night analyzed at three detection settings. Target phase 90 degrees; the radial axis gives the number of stimuli per bin (64 bins of 5.6 degrees); the green line marks the mean lock angle and the shaded sector its dispersion +- 1/2 SD).

Flexibility and Control

CLAS protocols can be configured using built-in presets, customized as needed, and executed on standard PC or tablet hardware. Signal and stimulus latencies are compensated automatically. Researchers can replay previously recorded data to optimize parameters before running participants. The millisecond processing speed of the phase-locking algorithm makes it capable of running many streams in parallel - allowing synchronized locking on separate channels, in separate frequency bands, and even across several participants on a single machine.

Results can be reviewed in detail and exported for further analysis. Centralized or cloud-based data collection is available optionally for remote and multi-site studies.

Ready to Implement CLAS?

Interested in implementing closed-loop stimulation with ZMax? Contact us to get started:

  • Request a trial version of the CLAS software with phase-locking support to configure and test your own protocol.
  • Share a sample EEG recording and receive a free simulation report with recommended CLAS parameters.
  • Discuss your study design and technical setup with our team.

Whether you are planning a pilot experiment or a full study, we can help you move from concept to implementation quickly.

Or, if you want to roll your own solution, click here.