Assessment of the Safety and Efficacy of the Felix NeuroAI Wristband in Essential Tremor: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Assessment of the Safety and Efficacy of the Felix NeuroAI Wristband in Essential Tremor (NCT06235190), a randomized na study. The primary outcome is Change in Tremor Research Group Essential Tremor Rating Assessment Scale (TETRAS) Modified Activities of Daily Living (mADL) Score From Baseline Pre-stimulation to Day 90.

2. Clinical Question

The registry summary frames the study as a comparison of safety and efficacy in essential tremor. The primary statistical comparison concerns the average change in a specified symptom and activity score over the treatment period. The listed secondary outcomes include additional TETRAS measures, clinician and patient global impressions, and quality of life measured with QUEST. Those measures broaden the outcomes under study, but the primary result remains specific to TETRAS mADL at Day 90. The record says participants completing the treatment period would be eligible for a long-term open-label study with Felix; the primary comparison is the randomized Felix-versus-sham period.

3. Trial Design

The registry record identifies a randomized, double-masked, parallel-group trial comparing the Felix NeuroAI Wristband with a sham device. The stated enrollment is 125. The record lists the study as completed, with a start date of 2024-02-27 and a primary completion date of 2024-10-31. The phase is listed as NA. These registry entries describe the planned or recorded trial structure; they do not establish how many participants contributed data to each outcome.

Participants were assigned to Felix or sham for up to 90 days. The record then says participants would be eligible to enter a long-term open-label study and receive Felix. The open-label follow-on is distinct from the randomized comparison and should not be treated as part of the primary Felix-versus-sham result. Arm descriptions beyond the intervention names are not reported in the registry record.

The primary endpoint is change in TETRAS mADL score from baseline pre-stimulation to Day 90. The posted outcome values are based on 62.0 participants in the Felix arm and 36.0 in the sham arm. These endpoint-specific counts are not the same as the enrollment figure and should not be read as a description of all participants enrolled, randomized, or followed. The record does not report the reasons that the endpoint counts differ from enrollment.

The registry lists 8 outcome measures, including the primary endpoint and secondary measures. The secondary outcomes include TETRAS mADL changes at earlier follow-up times, a TETRAS Performance Subscale dominant-hand score, clinician and patient global severity and improvement ratings, and QUEST quality of life. The record lists one analysis and reports a two-sided t-test for the primary comparison. Further details about the randomization procedure, masking implementation, allocation ratio, and analysis populations are not reported in the registry record.

4. Participant Flow and Accounting

The registry reports how participants moved through each stage of the study. All counts below are exactly as posted; missing counts are not inferred from other stages.

Overall Study

5. Baseline Characteristics

The table reproduces the baseline characteristics posted in the registry, which describe the participants analysed at baseline and help show whether the randomized groups were comparable.

6. Primary Endpoint

The registry lists the primary outcome as Change in Tremor Research Group Essential Tremor Rating Assessment Scale (TETRAS) Modified Activities of Daily Living (mADL) Score From Baseline Pre-stimulation to Day 90, measured over 90 days.

The primary endpoint is the change in TETRAS Modified Activities of Daily Living score from baseline pre-stimulation to Day 90. TETRAS mADL combines items from the TETRAS activities-of-daily-living subscale and the performance subscale. In the registry description, each item is scored from 0 to 4, with higher item scores indicating greater tremor severity. The total score is the sum of 13 items and ranges from 0 to 52.

This measure is used to quantify tremor-related difficulty in activities and observed performance in a structured way. Because it is a continuous score, the primary result is reported as a mean change for each arm and a mean difference between arms. The posted average changes are -6.9 points for Felix and -2.7 points for sham. A negative change indicates a lower score at follow-up than at baseline; given the scale description, that is a movement toward less severe scores on the measure. The posted comparison estimates the difference between the groups’ average changes, rather than describing the experience of every participant.

The endpoint focuses on a specified follow-up time and score. It does not, on its own, describe every aspect of tremor, daily functioning, quality of life, or safety. The registry lists other outcomes that address some of those domains, but the primary statistical result should be interpreted as evidence about the stated TETRAS mADL measure at Day 90. Details such as the handling of missing endpoint measurements and the exact analysis population are not reported in the registry record.

7. Results Posted in the Registry

• Felix NeuroAI Wristband: mean -6.90 versus -2.70 on control; difference -4.20 (95% CI -6.21 to -2.19), standardized effect d = -0.78; z-test p = <0.001.

8. Statistical Interpretation

9. Statistical Methods Explained

The statistical approach reported for the primary comparison is a two-sided t-test, with the result expressed as a net mean difference. The posted arm means and standard deviations also allow an unadjusted recalculation of the contrast and its uncertainty. Such a recalculation is based on summary statistics and is not a substitute for additional details about the trial’s analysis plan or participant-level data.

What does the mean change represent?

For a continuous outcome, a group mean describes the average value across the participants contributing to that outcome. Here, the reported values are average changes in TETRAS mADL score from baseline to Day 90, not average scores at Day 90 alone. The Felix mean change is -6.9 points and the sham mean change is -2.7 points. A negative change means the follow-up score is lower than the baseline score. Since the registry says higher item scores indicate greater tremor severity, a lower score represents a lower level of severity on this measure, on average.

The mean difference compares those two average changes. The posted difference of -4.2 points is the Felix group’s mean change relative to the sham group’s mean change under the stated comparison. It is a group-level contrast, not a count of participants who improved and not an estimate of the proportion who experienced a particular degree of change. A mean can be influenced by the spread and shape of the underlying participant-level results, neither of which can be fully assessed from group means alone.

What does the t-test assess?

A t-test evaluates whether the observed difference in means is compatible with a null comparison of no average difference, using assumptions about the sampling distribution of the mean difference. The registry labels its test as two-sided, meaning that differences in either direction are considered rather than testing only for a difference in one prespecified direction. The registry reports p < 0.001. The simple unadjusted recalculation reports p <0.001. Both are small p-values, but they are outputs of their respective reported and recalculated analyses and should be quoted with their own stated thresholds.

A p-value is a measure of how compatible the observed data are with a specified null model and its assumptions. It is not the probability that the null hypothesis is true, nor does it quantify the size or importance of the difference. The estimated mean difference and its confidence interval are needed to describe magnitude and precision alongside the p-value.

Why report a confidence interval?

A confidence interval gives a range of values compatible with the observed data and the method used to estimate uncertainty. The recalculated 95% confidence interval for the mean difference extends from -6.206 to -2.194 points. Because the range is wholly below zero, it is consistent with a negative average contrast under the comparison as posted. The interval also conveys that the estimate is not exact: values across the range represent different plausible magnitudes of the difference under the analysis assumptions. It does not guarantee that the true effect lies inside this particular interval.

What do the standard deviations add?

The posted standard deviations are 6.0 points for Felix and 4.1 points for sham. A standard deviation describes how dispersed participant changes are around the arm’s mean change. It is not the uncertainty in the mean itself and should not be read as a confidence interval. The reported spread indicates that individual change scores vary; the average difference does not imply that every participant’s change was close to the arm mean.

In general, greater variability can make a mean difference less precise when other design features are held constant, because the group averages carry more sampling uncertainty. General statistical considerations do not establish whether the variability is adequately modeled in this specific trial. The registry record does not report diagnostics for the t-test assumptions, distributional checks, or an alternative analysis for unusual score distributions.

What is and is not specified about the analysis?

The registry identifies one analysis and reports one primary comparison. It does not report the detailed statistical analysis plan, whether the primary analysis was adjusted for baseline score or other covariates, or which analysis population was used. The outcome counts are 62.0 in the Felix arm and 36.0 in the sham arm, but these counts alone do not identify whether the analysis followed an intention-to-treat, per-protocol, or other population definition. Those details are not reported in the registry record.

The recalculated statistics are simple unadjusted calculations from the posted summary statistics. They do not incorporate covariate adjustment, missing-data modeling, stratification, or any participant-level features that might have been part of a fuller analysis. The registry’s own reported test remains the sponsor’s analysis. The record does not report the original design assumptions behind the trial’s enrollment, sample-size justification, interim analyses, or procedures for multiplicity. Those omissions mean that the reported primary result can be described, but its relationship to every planned analysis decision cannot be reconstructed from the registry record alone.

In general, statistical interpretation depends not only on a test and its p-value but also on how outcomes and populations were defined, how missing values were handled, and whether analyses were selected before results were known. The registry record does not provide enough detail to resolve all of those questions. The reported result should therefore be interpreted as the posted primary comparison, with the unadjusted recalculation serving as a transparent summary-statistic check rather than a replacement for unavailable analysis documentation.

10. Confidence Intervals and Precision

The recalculated 95% confidence interval for the mean difference in TETRAS mADL change is -6.206 to -2.194 points. The estimate is -4.2 points. The interval is entirely on the negative side of zero, indicating that the estimated contrast is directionally consistent across the range of values covered by this method. This is compatible with a larger mean reduction in the Felix group than in the sham group for the posted contrast.

Precision is about the width of the range of plausible values, not simply whether a p-value crosses a threshold. The interval gives information about the uncertainty in the estimated group difference: it supports a range of magnitudes rather than a single exact value. Its endpoints should be considered together with the point estimate and the measurement scale. The record does not report a threshold for a clinically important TETRAS mADL difference, so the interval alone cannot determine how important its values are in clinical or everyday terms.

The interval is a recalculated, unadjusted interval based on the posted summary statistics. It does not account for adjustments or modeling details that are not specified in the registry record. It also does not express uncertainty about all possible outcomes, the safety profile, or longer-term open-label experience. The observed standard deviations of 6.0 points in the Felix arm and 4.1 points in the sham arm describe participant-level variation, whereas the confidence interval describes uncertainty about the estimated mean contrast. These are related but different quantities.

The registry’s own analysis reports the same mean difference, -4.2, but does not report a confidence interval. The interval above therefore belongs to the unadjusted recalculation and should not be attributed to the sponsor’s reported analysis. The sponsor’s reported p-value is p < 0.001, while the recalculated p-value is <0.001.

11. Planning a Similar Trial: Sample Size

This is a planning illustration, not a re-evaluation of the original trial. It asks: if a new study wanted to detect a difference like the one observed, or a smaller one, how large would it need to be? It is deliberately not “observed power” of the finished trial, which is not a meaningful quantity.

The planning figures are illustrations based on the observed mean difference, not a description of the trial’s original enrollment plan. They give participants per arm for 80% power and participants per arm for 90% power. They are not totals across arms and are not total-event figures; the event-based convention does not apply to this continuous outcome.

The stated planning assumptions are a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation. The effect to detect is set to a stated fraction of the observed effect. The row for 100% of the observed effect gives 26 participants per arm for 80% power and 35 participants per arm for 90% power. Planning for 75% of the observed effect gives 46 participants per arm for 80% power and 62 participants per arm for 90% power. Planning for 50% of the observed effect gives 104 participants per arm for 80% power and 139 participants per arm for 90% power.

In plain terms, the 80% power column gives the estimated participants per arm under the stated assumptions for a design intended to detect the chosen effect with 80% power. The 90% power column does the same for a design intended to detect that effect with 90% power. Greater power requires more participants under otherwise unchanged assumptions. The illustrations also show the general relationship between effect size and required enrollment: smaller effects need more participants to detect them at the same power and significance level.

These are planning illustrations based on the observed effect, not observed power. The original trial’s own design assumptions and sample-size justification are not reported in the registry record. The illustrations should not be taken as evidence that the trial was planned using these assumptions or that its enrollment was adequate for every outcome.

12. How the Sample Size Is Derived

Power is the probability that a study of a given size rejects the null hypothesis when a specific true effect exists. It depends on four things: the effect to detect, the variability of the outcome, the significance level (here a two-sided α of 0.05, so z_1−α/2 = 1.960) and the desired power (z_1−β = 0.842 for 80% and 1.282 for 90%).

Formula

n is the number of participants in each arm, Δ is the difference in means to detect and σ is the common standard deviation.

Worked calculation for this trial

1. Observed difference Δ = -4.200; pooled standard deviation σ = 5.386.
2. n = 2 × (1.96 + 0.842)² × 5.386² / -4.200² = 25.8, rounded up to 26 participants per arm for 80% power.

The smaller-effect rows in the table above repeat the same calculation with the effect multiplied by 0.75 and 0.5 (for ratios, applied to the logarithm of the ratio).

R code

alpha <- 0.05
z_a   <- qnorm(1 - alpha / 2)          # two-sided
delta <- -4.2                      # observed difference in means
sigma <- 5.38550330362292                     # pooled standard deviation
n_per_arm <- function(power, frac = 1) {
  ceiling(round(2 * (z_a + qnorm(power))^2 * sigma^2 / (delta * frac)^2, 6))
}
for (f in c(1, 0.75, 0.5))
  cat(sprintf("%3.0f%% of observed effect: n per arm = %s (80%% power), %s (90%% power)\n", 100 * f,
              format(n_per_arm(0.80, f), big.mark = ","), format(n_per_arm(0.90, f), big.mark = ",")))

# Check against R's built-in function (t-based, so very slightly larger)
cat("power.t.test:", round(power.t.test(delta = delta, sd = sigma, power = 0.80)$n, 1), "\n")

# Simulation: exact two-sample t-test using simulated means and variances
set.seed(6235190)
n    <- n_per_arm(0.80)
reps <- 20000
m_diff <- rnorm(reps, delta, sigma * sqrt(2 / n))             # difference in sample means
s2     <- sigma^2 * rchisq(reps, 2 * (n - 1)) / (2 * (n - 1))   # pooled sample variance
tstat  <- m_diff / sqrt(s2 * 2 / n)
cat(sprintf("Simulated power at n = %s per arm: %.3f (%d replicates)\n", format(n, big.mark = ","), mean(abs(tstat) > qt(0.975, 2 * (n - 1))), reps))

Output from running the code

100% of observed effect: n per arm = 26 (80% power), 35 (90% power)
 75% of observed effect: n per arm = 46 (80% power), 62 (90% power)
 50% of observed effect: n per arm = 104 (80% power), 139 (90% power)
power.t.test: 26.8 
Simulated power at n = 26 per arm: 0.788 (20000 replicates)

The simulation draws many sets of group means and variances at the planned sample size and counts how often the two-sample t-test rejects the null. It should land close to the target power; at very small required sample sizes it can fall below it, because the normal approximation is rough there.

13. Analyses Reported by the Sponsor

The registry reports the sponsor’s primary analysis as a two-sided t-test with a net mean difference of -4.2 and p < 0.001. The registry record does not give a confidence interval for that reported result. A separate unadjusted recalculation from the posted arm means and standard deviations also gives a mean difference of -4.2, with a 95% confidence interval from -6.206 to -2.194 and p <0.001.

The record does not flag a discrepancy between the sponsor’s reported result and the recalculated comparison. The point estimates agree as reported. The p-values are expressed using different thresholds, so they should not be treated as identical numerical values. The confidence interval is available from the recalculation, not from the sponsor’s stated analysis in the registry record.

The recalculation is a simple check based on posted summary statistics. It does not reproduce any unreported adjustment, analysis-population rule, or participant-level modeling that might underlie the sponsor’s analysis. The sponsor’s reported analysis is the result to cite as the registry’s own analysis; the recalculation adds an unadjusted estimate of precision and should be labeled accordingly.

14. Multiplicity, Missing Data and Analysis Populations

The registry lists one primary analysis, and the reported primary comparison is the Felix-versus-sham difference in change in TETRAS mADL score at Day 90. It also lists secondary outcomes that include TETRAS measures at other time points, dominant-hand performance, clinician and patient global severity and improvement ratings, and quality of life. The record does not report a multiplicity adjustment, a hierarchy for testing outcomes, or a procedure for controlling the chance of false-positive findings across multiple outcomes or time points.

In general, when multiple outcomes or comparisons are examined, the chance of observing at least one apparently notable result can increase if no adjustment or prespecified testing strategy is used. Whether that issue affects interpretation depends on the planned analysis and how results are reported. Because the record does not state the multiplicity strategy, no conclusion about adjustment can be made. The registry entries for this analysis give the primary result but do not provide secondary outcome results, so their estimates and uncertainty are not reported in the registry record.

The handling of missing outcome data is also not reported in the registry record. The primary outcome counts are 62.0 for Felix and 36.0 for sham, while the total enrollment is stated as 125. These figures do not reveal why some participants are absent from the primary outcome counts or how missing measurements were handled. The registry does not report imputation, sensitivity analyses, withdrawal patterns, or whether the analysis used all randomized participants.

The exact analysis population is not reported in the registry record. The endpoint-specific arm counts should not be taken as proof of an intention-to-treat or per-protocol analysis. Baseline adjustment, covariate handling, and any subgroup analyses are also not reported. The record identifies a two-sided t-test and one analysis, but does not provide the detailed statistical analysis plan needed to assess all prespecification and analysis decisions.

Interim analyses and stopping rules are not reported in the registry record. The original sample-size justification and the trial’s planned assumptions are also not reported. These missing details do not change the arithmetic of the posted comparison, but they limit assessment of the full analysis process and of how the primary result fits within all outcomes and decisions considered during the trial.

15. Safety Results

The registry reports serious adverse events as affected participants divided by participants at risk.

Deaths

Other (non-serious) adverse events

Participants with at least one non-serious event at or above the registry's 0% reporting threshold.

Most frequent non-serious adverse event terms

The registry lists serious adverse events for 0 participants at risk out of 83 in the Felix group and 0 participants at risk out of 42 in the sham group. Deaths are also listed as 0 in each group. These are the serious-event counts recorded for the listed safety groups. The registry does not list serious adverse-event terms.

These safety denominators differ from the primary endpoint counts of 62.0 for Felix and 36.0 for sham. They describe the groups used for the adverse-event entries and should not be substituted for the outcome-specific counts. The record also lists other adverse events affecting 28 participants in the Felix group and 2 in the sham group; these are separate from the serious adverse-event figures and should not be conflated with them.

A zero count in the posted serious-event entries means no serious events are recorded in those entries; it does not establish that no possible safety issue exists. The registry information does not provide event details beyond the counts, the duration of safety follow-up, or additional context needed to assess the nature and timing of events. The figures should be read as reported safety counts, not as a complete characterization of risk.

16. What the Result Does, and Does Not, Mean

The posted primary result describes a difference in average TETRAS mADL change between Felix and sham at Day 90. Mean change was -6.9 points in the Felix group and -2.7 points in the sham group, with a posted mean difference of -4.2 points. The unadjusted recalculated 95% confidence interval extends from -6.206 to -2.194, and the recalculated p-value is <0.001. The registry’s own analysis reports p < 0.001.

This is evidence about a particular measured outcome and comparison. It does not show that every participant experienced a reduction, quantify the number of participants who improved, or establish the experience of any individual. The scale is a composite score that combines activities-of-daily-living and performance items; it is not a direct account of every aspect of essential tremor or quality of life. The registry lists additional measures, but their results are not reported in the registry record in the registry record for this analysis.

The result also does not by itself establish the clinical importance of the mean difference. The registry record does not state a threshold for a clinically important change on TETRAS mADL. Statistical evidence against a zero mean difference is distinct from evidence about how a score difference translates into meaningful benefit for participants. The confidence interval gives a range of estimated differences under an unadjusted method, but it cannot resolve that clinical interpretation on its own.

The randomized, double-masked design supports a comparative reading of the treatment period, but the available record does not include all details needed to evaluate the analysis. The analysis population, missing-data handling, baseline adjustment, and multiplicity strategy are not reported in the registry record. Nor does the primary outcome establish what happens during the long-term open-label period mentioned in the registry summary.

The safety record lists no serious adverse events or deaths in either listed safety group, alongside other adverse events in both groups. Those entries should be interpreted narrowly as registry counts, not as proof of an absence of risk. Taken together, the posted information describes a statistically distinct average change on the primary measure and provides limited safety counts. It does not answer every question about durability, individual response, all secondary outcomes, or the clinical meaning of the observed score difference.

17. Why This Trial Matters Statistically

This trial’s primary comparison is statistically informative because it contrasts a device with a sham condition in a randomized, double-masked, parallel-group design and specifies a continuous symptom-related outcome at a defined follow-up. The reported mean difference and confidence interval make it possible to discuss both the estimated direction and the uncertainty in the average contrast, rather than relying on a p-value alone.

The endpoint also makes the interpretation concrete: TETRAS mADL combines ratings of activities and performance, and its items are scored in the direction of increasing tremor severity. That structure permits a group-level comparison of score changes over the treatment period. At the same time, the outcome is a measure with a defined scope, not a comprehensive summary of every relevant experience or outcome.

Statistically, the result illustrates why the estimate, interval, and analysis context all matter. The confidence interval is entirely below zero, while the registry’s own p-value and the recalculated p-value are both reported using threshold notation. The recalculation supplies an interval based on summary statistics, but it cannot fill gaps about analysis populations, missing data, baseline adjustment, or multiplicity. Those omissions are important when judging how fully the posted statistical result can be independently evaluated.

The record also distinguishes the randomized treatment comparison from an intended open-label follow-on. That distinction matters because a controlled comparison and an open-label experience answer different statistical questions. The planning illustrations show how assumed effect size and desired power relate to enrollment requirements, but they are not evidence about the trial’s original design rationale. Overall, the record supports a focused interpretation of the primary comparison while leaving important questions about the wider analysis plan and longer-term outcomes unresolved.

18. Limitations

• The posted summary statistics are the only data these results rest on; patient-level data, full protocols and statistical analysis plans are not available.
• Recalculated results are simple, unadjusted calculations from posted summaries. They do not replace the sponsor's prespecified analysis.

The registry record provides a primary result but leaves several trial-specific details unavailable. The primary endpoint counts are 62.0 for Felix and 36.0 for sham, compared with stated enrollment of 125. Reasons for the difference between enrollment and outcome counts, participant flow, and the handling of missing measurements are not reported in the registry record. This limits assessment of whether the observed data represent the intended analysis population and whether missingness could affect the estimate.

The record does not define the analysis population or state whether the primary comparison was adjusted for baseline score or other covariates. The reported sponsor analysis is identified as a two-sided t-test, but diagnostics, distributional checks, and sensitivity analyses are not reported. The unadjusted recalculation uses posted summary statistics and cannot address these issues.

The registry lists multiple secondary outcomes, but their results are not reported in the registry record for this analysis. A multiplicity strategy is not reported, so it is not possible to determine how the primary and secondary outcomes were organized for inference. The original sample-size rationale and design assumptions are also not reported; the planning illustrations based on the observed effect cannot establish what assumptions guided enrollment.

The record does not provide a clinically important difference threshold for TETRAS mADL, limiting interpretation of the practical meaning of the estimate. It also does not provide detailed arm descriptions, implementation details for masking, or details about the long-term open-label follow-on. Safety entries include event counts but not serious-event terms or broader context about event timing and follow-up. These limitations do not change the reported primary estimate, but they constrain what can be concluded from the registry record about robustness, clinical importance, and the broader benefit-risk picture.

19. Related Tutorials

20. Related Statistical Calculators

21. Sources

• ClinicalTrials.gov: Assessment of the Safety and Efficacy of the Felix NeuroAI Wristband in Essential Tremor — NCT06235190.
• Registry record last updated 2026-07-09; status last verified 2026-06.
• Results first posted 2026-07-09.
• Data retrieved from ClinicalTrials.gov on 2026-11-09.
• Group sizes, means, event counts and sponsor-reported p-values are reproduced as posted. Effect estimates, confidence intervals and p-values not attributed to the sponsor are independently calculated from the posted summary statistics.

22. Record Summary

The registry record describes a completed randomized, double-masked, parallel-group study of the Felix NeuroAI Wristband compared with a sham device in patients with essential tremor. Stated enrollment is 125, and the treatment period was up to 90 days. The primary endpoint is change from baseline pre-stimulation to Day 90 in TETRAS mADL score.

The posted mean changes are -6.9 points for Felix and -2.7 points for sham, with a mean difference of -4.2. The registry reports a two-sided t-test with p < 0.001. An unadjusted recalculation gives a 95% confidence interval from -6.206 to -2.194 and p <0.001. Serious adverse events and deaths are listed as 0 in both safety groups.

Important details, including missing-data handling, analysis populations, multiplicity procedures, and the original sample-size rationale, are not reported in the registry record. The results therefore support a focused reading of the posted Day 90 primary comparison, with limits on interpretation of clinical importance and the full analysis process.