Search Results

The Felix™ NeuroAI™ Wristband is indicated to aid in tremor-related functional limitations in the upper limbs in adults with essential tremor.

Fasikl's Felix™ NeuroAI™ Wristband is a wrist-worn, noninvasive, transcutaneous neurostimulation system. It is intended to be used by adult patients with essential tremor (ET) on a daily basis to aid in tremor-related functional limitations in the upper limbs. The system continuously monitors tremor frequency and tremor amplitude and automatically adjusts stimulation settings. The Felix system consists of the following components: Felix wristband (NeuroAI™ stimulator with integrated strap and detachable connector band), disposable electrode band, wireless charger, and smartphone application.

The Felix NeuroAI System is an external upper limb tremor stimulator cleared via a 510(k) pathway. The device's acceptance criteria are based on a superiority study (TRANQUIL study) against a sham control, demonstrating its effectiveness in aiding tremor-related functional limitations.

1. Table of Acceptance Criteria and Reported Device Performance

The core acceptance criterion for the Felix NeuroAI System was demonstrating superiority over sham stimulation in the primary effectiveness endpoint.

Note on Secondary Endpoints (Descriptive Results):

While not formal acceptance criteria for statistical superiority, the secondary endpoints provided reinforcing evidence of the device's performance:

• Change in TETRAS mADL score over time: The Felix group showed continued improvement over 90 days, while the sham group plateaued at 14 days.
• CGI-I (Clinical Global Impression of Improvement): 69.4% of Felix users achieved "any improvement" vs. 44.4% in sham at 90 days.
• PGI-I (Patient Global Impression of Improvement): 67.7% of Felix users achieved "any improvement" vs. 50.0% in sham at 90 days.
• QUEST Total Score: Felix group showed a greater reduction in QUEST total score (-6.0 ± 10.77) compared to sham (-3.4 ± 7.87), indicating better quality of life improvement.
• Responder Analysis: Felix group consistently showed higher responder rates across various thresholds of TETRAS mADL improvement (e.g., ≥10% relative reduction: Felix 71.0% vs. Sham 38.9%).

2. Sample Size Used for the Test Set and Data Provenance

• Test Set (Clinical Study - TRANQUIL study):
  • Total Randomized: 125 subjects
  • Felix Group: 83 subjects
  • Sham Group: 42 subjects
  • Analysis Populations for Primary Endpoint:
    • Prespecified Primary Analysis (excluding missing data): Felix (N=62), Sham (N=36)
    • Supplementary Post-hoc Primary Analysis (MMRM imputation): Felix (N=83), Sham (N=42)
• Data Provenance: The study was a multi-center, multi-region clinical trial (MRCT).
  • Country of Origin: Eight (8) sites in the US and four (4) sites outside the US (in China).
  • Study Design: Prospective, randomized, sham-controlled, double-blinded.

3. Number of Experts Used to Establish the Ground Truth for the Test Set and Qualifications of Those Experts

• Experts for Ground Truth Establishment (TETRAS assessments): The "raters assessing the effectiveness endpoints" were blinded to the treatment assignment. The text explicitly states that TETRAS PS items were rated by a physician. For the TETRAS mADL, it is a composite score, where some components are patient-rated (ADL items) and some are physician-rated (PS items).
• Number of Experts: Not explicitly stated as a specific number of unique experts across the entire study. It refers to "raters" and "physicians" implying multiple individuals depending on clinical site staffing.
• Qualifications: "Movement disorder specialist" for the initial clinical diagnosis of ET (inclusion criteria). For the TETRAS PS items, the evaluators were "physicians". No specific years of experience or board certifications are detailed for the raters/physicians.

4. Adjudication Method for the Test Set

• Adjudication Method: Not explicitly stated for the TETRAS scores. The study was double-blinded, meaning both patients and raters assessing the effectiveness endpoints were blinded. This blinding helps mitigate bias in assessment. However, a formal adjudication process for discordant ratings (if multiple raters were to assess the same patient) is not described.

5. Multi-Reader Multi-Case (MRMC) Comparative Effectiveness Study

• Was an MRMC study done? No, the study was a clinical trial comparing the device (Felix) to a sham control in human subjects, not a reader study evaluating human performance with and without AI assistance for interpretation tasks. The AI in this device is for closed-loop tremor stimulation control, not for diagnostic or interpretive purposes that would typically involve a multi-reader study.
• Effect Size of Human Readers Improvement with AI vs. Without AI Assistance: Not applicable, as this was not an MRMC comparative effectiveness study focused on human reader performance.

6. Standalone Performance (Algorithm Only Without Human-in-the-Loop)

• Was a standalone performance done? The clinical study (TRANQUIL study) evaluated the performance of the Felix NeuroAI system (device + AI algorithm) against a sham device. The AI algorithm's role is specifically to control the stimulation based on tremor measurements, not to provide an interpretable output or diagnosis.
• Performance: The primary outcome (TETRAS mADL change) quantifies the effect of the system (AI-controlled stimulation) as a whole. While the AI is an integral part of the device's function, its "standalone" performance in isolation from the stimulation output and overall patient effect is not separately reported in terms of a diagnostic metric (e.g., accuracy, sensitivity, specificity). The non-clinical data section (9.1) mentions "ML/AI Algorithm Verification Testing" as part of software testing, but this is a technical verification, not a clinical standalone performance evaluation in the sense of a diagnostic algorithm.

7. Type of Ground Truth Used for the Test Set

• Type of Ground Truth: Expert Consensus (Clinical Assessments) and Patient Reported Outcomes.
  • The primary effectiveness endpoint (TETRAS mADL score) is a composite measure involving both patient ratings (ADL items based on the prior week) and physician ratings (PS items, specifically Archimedes spirals and handwriting). This combines subject report with professional clinical assessment.
  • The clinical diagnosis of ET was established by a "movement disorder specialist".

8. Sample Size for the Training Set

• Training Set Sample Size: Not explicitly stated in the provided text. The text describes the clinical study (TRANQUIL) as the test set for device effectiveness. Information regarding the training data for the AI algorithm (e.g., number of subjects, diversity of data) is typically found in the software documentation for the submission, not usually summarized in this level of detail in a 510(k) summary focused on clinical efficacy. The AI algorithm's purpose is "to control the stimulation state of the Felix device" and is "NOT intended to be used for any diagnostic purposes."

9. How the Ground Truth for the Training Set Was Established

• Ground Truth Establishment for Training Set: Not explicitly stated in the provided text. Given the AI's role is device control (closed-loop stimulation based on internal IMU measurements of tremor), its "ground truth" for training would likely relate to mapping IMU data to optimal stimulation parameters for tremor suppression. This would heavily rely on biofeedback and iterative optimization within an engineering context, potentially involving human subject data collected during development. The software verification for closed-loop control (IEC 60601-1-10) is mentioned, suggesting engineering standards for development.

Ask a specific question about this device