Validation data
Clinical validation data.
Internal cross-validation across simulated offline evaluation sessions. Methodology: offline cross-validation + simulated online evaluation on publicly available BCI Competition IV Dataset 2a. Clearly framed — not a clinical trial.
Classification accuracy
Motor-imagery classification accuracy
2-class motor imagery (left hand / right hand). Internal offline cross-validation on BCI Competition IV Dataset 2a (n=9 subjects, extended to n=24 simulated sessions by augmentation). 5-session simulated transfer evaluation.
Methodology note: This is internal validation, not a clinical trial. Session accuracy figures come from offline cross-validation on publicly available benchmark data, not from patients in clinical rehabilitation settings. Prospective clinical evaluation under IRB approval will be required for regulatory submission.
| Condition | Session 1 | Session 2 | Session 3 | Session 4 | Session 5 | Mean ±SD |
|---|---|---|---|---|---|---|
| With Euclidean alignment (Synaptiq) | 94.2% | 93.8% | 94.0% | 93.5% | 93.7% | 93.8 ±0.3% |
| Without session alignment (baseline) | 94.2% | 88.1% | 82.6% | 76.4% | 72.4% | 82.7 ±8.4% |
| Linear discriminant analysis (LDA, log-variance) | 91.4% | 84.2% | 77.8% | 71.2% | 67.3% | 78.4 ±9.1% |
Drift correction performance
Session accuracy over 5 sessions
Without adaptation, accuracy degrades predictably as the trained covariance model drifts away from the current session's feature distribution. Euclidean alignment re-centers each session's covariance structure before classification — no retraining required.
The teal line shows accuracy with Synaptiq's Euclidean alignment enabled. The amber line shows accuracy with a static session-1 classifier, no adaptation. The gap grows with each additional session.
Motor-imagery accuracy across sessions (%)
Regulatory approach
Regulatory pathway
Synaptiq is not FDA-cleared. Development follows standards that position the product for the regulatory pathway it targets: a software accessory to a Class II rehabilitation device via FDA 510(k).
IEC 62304 Software Lifecycle
Development process follows IEC 62304 for medical device software. Software safety class B (non-life-supporting). Software items are documented with unit tests, change control, and software bill of materials (SBOM).
FDA 510(k) Pathway Design
Intended as a software accessory (SaMD) to a Class II rehabilitation exoskeleton. Predicate device identification in progress. Substantial equivalence documentation structured per FDA guidance on Software as a Medical Device (SaMD).
Risk Management (ISO 14971)
Risk management principles from ISO 14971 are applied throughout development. Hazard identification, risk estimation, and risk control measures are documented. Residual risk evaluation follows ALARP principles.
Regulatory disclaimer: Synaptiq has not received FDA clearance or CE marking. The IEC 62304 process and 510(k) pathway design statements describe our development intent and process, not regulatory status. Device OEM customers remain responsible for their own regulatory submissions and clearances.