Video Assisted Speech Technology to Enhance Motor Planning for Speech: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Video Assisted Speech Technology to Enhance Motor Planning for Speech (NCT04764539), a randomized na study. The primary outcome is Change in Mean Length of Utterance (MLU).

2. Clinical Question

3. Trial Design

This was a randomized, parallel-group study with no masking. Six participants were enrolled, with three assigned to the 2D iPad format and three to the 3D VR format. The record lists the study as completed. The sponsor was iTherapy, LLC. The recorded study start was 2019-12-01, and primary completion was 2020-11-30. The phase is listed as NA.

Both arms received Video-Assisted Speech Therapy stimuli, but the delivery formats differed. One arm used a tablet-based 2D presentation on an iPad Pro. The other used a 3D VR-integrated prototype with VR goggles and bone conduction headphones. The arm descriptions state that three children with autism spectrum disorder, aged 4 to 8 years, participated in each format. They describe a 14-session study; the primary outcome timeframe says sessions were held twice a week over seven weeks.

The first and last sessions were reserved for pre-test and post-test language sample collection and assessment. The outcome description says participants completed pre- and post-test language samples and that gain from pre-test to post-test was compared. The registry does not report the randomization method, allocation concealment, masking of assessors, or a detailed statistical analysis plan. It also does not report whether the same staff delivered both formats or whether session attendance and treatment exposure were complete for every participant.

Three secondary outcomes are listed: parent perceptions of communication changes resulting from study participation, change in type-token ratios, and increase in response rate to treatment stimuli. Six outcome measures and fourteen analyses are listed overall. The record does not provide enough detail to infer how these outcomes related to the listed analyses or whether those analyses were prespecified.

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

Analysis population for the primary outcome, as defined in the registry: All participants for each respective group.

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 Mean Length of Utterance (MLU), measured over Seven weeks--each subject participated in the study twice a week over a 7-week period for a total of 14 sessions. The first and last sessions (session #1 and session #14) were reserved for pre-test and post-test language sample collection and assessment..

The primary endpoint was change in mean length of utterance, abbreviated MLU, measured in morphemes per utterance. A morpheme is a component of a word; the registry description counts bound and free morphemes in spoken utterances. The measure summarizes utterance length in a language sample rather than counting only words. The registry presents the endpoint as a continuous measure, for which results are reported as means and standard deviations.

The outcome description says a pre-test and post-test language sample was collected and that the gain was compared. It describes two licensed speech-language pathologists observing a parent interacting and talking with the child during a five-minute period. The pathologists transcribed speech and calculated the number of morphemes per utterance by dividing the total morphemes by the total utterances. The description also refers to a pre- and post-test 15-minute language sample. These different stated durations are not reconciled in the registry record.

General: MLU is one way to describe speech production in a language sample. A change measure can express how a participant's measured value differs between two assessments, while a comparison between groups asks whether the observed changes or values differ by assigned format. Interpretation depends on consistent sampling and scoring. The registry does not report participant-level scores or enough detail to determine how variability between individual participants contributed to the reported group results.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

General: The primary outcome is continuous: MLU is expressed as morphemes per utterance, and each participant can have a value on a numerical scale. A group mean describes the average of the values in that group. A standard deviation describes how dispersed the individual values are around the group mean. It is not the same as a confidence interval, which describes uncertainty around an estimated quantity.

The registry reports group means and standard deviations for each arm. For the 2D arm, the mean is 0.539 and the standard deviation is 0.412; for the 3D arm, the mean is 0.299 and the standard deviation is 0.498. Each arm has three participants. With this small group size, each participant's observation can have substantial influence on the group summary. The summaries do not reveal the individual observations, their distribution, or whether a single observation strongly affected a mean.

What does the confidence interval express?

General: A confidence interval gives a range of values compatible with the data and the statistical method used, while representing sampling uncertainty. A narrower interval indicates greater precision than a wider interval, under the assumptions of that method. It is not a range that guarantees the true effect lies within it, and it does not describe the range of outcomes for individual participants.

The posted between-arm ANOVA result is a mean difference in final values of -0.24, with a 95% confidence interval from -1.276 to 0.796. Because this interval includes zero, the posted estimate does not distinguish a nonzero difference from no difference at the stated confidence level. The interval also spans values on either side of zero, so the data are compatible with effects in either direction. The width of the interval matters: the point estimate alone would give an incomplete account of uncertainty.

What does the ANOVA comparison address?

General: Analysis of variance, or ANOVA, is a method for comparing means under a specified model. A two-group ANOVA can assess whether the group means differ, but the interpretation depends on how the outcome and comparison were defined and on the assumptions of the analysis. The registry labels the reported between-arm quantity as a mean difference in final values. It does not describe the model formula, baseline adjustment, diagnostics, or any covariates. Those features should not be inferred from the name of the method alone.

The outcome description emphasizes pre-test to post-test gain, but the posted comparison is identified as a comparison of final values. This distinction affects the question being answered. A comparison of final values is not automatically the same as a comparison of change scores or a model that adjusts final values for baseline. The registry record does not clarify which exact data entered the reported ANOVA or how the stated gain was incorporated.

How should the p-values be read?

General: A p-value describes how unusual data at least as incompatible with a specified null model would be if that model and the analysis assumptions were correct. It is not the probability that the null hypothesis is true, and it is not a measure of effect size. It also does not tell whether an effect is important in practice. The posted p-value for the between-arm comparison is p < 0.555. The confidence interval is essential for understanding the range and uncertainty of the estimated difference; the p-value alone cannot supply that information.

The within-arm p-values are p < 0.213 for the 2D arm and p < 0.533 for the 3D arm. Those tests concern the respective arm-level results as reported and do not test whether the arms differ from each other. Separate within-arm tests cannot substitute for the randomized between-arm comparison. The record does not explain the null hypotheses or the exact construction of these within-arm ANOVA results, so their interpretation should remain tied to the labels in the registry.

Why does the very small sample matter?

General: With only a few observations, estimates of both means and variability can be unstable. Confidence intervals can be wide, and standard assumptions used by a parametric method can be difficult to assess from the observed data. Random assignment helps define a comparison, but it does not ensure that small groups will be similar on every relevant characteristic. The registry record gives no baseline characteristics by arm, so the balance of participant features cannot be assessed from the posted information.

For continuous outcomes, departures from model assumptions and influential observations can matter. The record does not report distributional checks, residual diagnostics, robust methods, or a sensitivity analysis. Those omissions do not demonstrate that assumptions were violated; they mean the record does not allow readers to evaluate them. Similarly, no participant-level data are posted in the registry record described by the registry results, so the reported means and standard deviations cannot be independently checked at the individual level.

What does a planning illustration mean?

General: Sample-size planning asks how many participants might be required under chosen assumptions to detect a specified effect with a chosen statistical power and significance level. Such a calculation is prospective and depends on assumptions about the effect, variability, allocation, and analysis. An illustration based on an observed effect is not an observed-power result and is not evidence that the completed study had adequate power. The planning assumptions recorded for this trial specify a two-sided significance level of 0.05, equal allocation, and a normal approximation, with the target effect expressed as a fraction of the observed effect.

The registry does not report the trial's own sample-size justification or say that these planning assumptions governed the original design. The calculated illustrations should therefore be read only as hypothetical planning scenarios, not as the number enrolled, a protocol requirement, or a measure of the study's ability to detect an effect. No planning count should be interpreted as an observed result.

10. Confidence Intervals for Each Arm

No two-arm comparison is shown for this trial, but each arm's mean can be given with a t-based 95% confidence interval computed from the posted mean, spread and number of participants.

R code and output

arm <- c("Stimuli Administered Via 2D Format on an iPad Pro", "Stimuli Administered in 3D Format Via VR Goggles a")
n   <- c(3, 3)
m   <- c(0.5387, 0.2987)
sp  <- c(0.4119, 0.4977)
sd  <- sp      # posted standard deviations
for (i in seq_along(n)) {
  se <- sd[i] / sqrt(n[i]); h <- qt(0.975, n[i] - 1) * se
  cat(sprintf("CARM|%s|%d|%.3f|%.3f|%.3f|%.3f\n", arm[i], n[i], m[i], sd[i], m[i] - h, m[i] + h))
}

CARM  Stimuli Administered Via 2D Format on an iPad Pro  3  0.539  0.412  -0.485  1.562
CARM  Stimuli Administered in 3D Format Via VR Goggles a  3  0.299  0.498  -0.938  1.535

11. Analyses Reported by the Sponsor

The registry reports the sponsor's ANOVA results, including the between-arm mean difference in final values of -0.24, its 95% confidence interval from -1.276 to 0.796, and p < 0.555. These are the posted analyses to use when describing the sponsor's reported comparison. The record also gives separate ANOVA results for each arm, but those are not substitutes for the between-arm estimate.

A recalculated two-arm comparison is not available because the record indicates that values are missing or not estimable for that comparison. Accordingly, there is no independent unadjusted recalculated estimate to set beside the sponsor's analysis. The record does not flag a discrepancy between a recalculation and the sponsor's result. No claim about agreement or disagreement between those analyses is warranted.

The group means and standard deviations remain useful descriptive information, but they do not resolve the uncertainty in the inferential comparison. The registry's between-arm analysis is labeled as a comparison of final values, while the endpoint description refers to pre-to-post gain. The record does not explain that distinction or provide enough detail to reconstruct the analysis. The sponsor's reported result should therefore be presented with its stated label and uncertainty, without recasting it as a different estimand.

12. Multiplicity, Missing Data and Analysis Populations

The registry lists six outcome measures and fourteen analyses. It identifies three secondary outcomes: parent perceptions of communication changes resulting from study participation, change in type-token ratios, and increase in response rate to treatment stimuli. The record does not explain which analyses correspond to each outcome, which tests were primary or secondary, or whether any multiplicity adjustment was planned or used. It does not report a hierarchy of hypotheses or an analysis plan that would establish how to interpret the collection of tests.

General: When many outcomes or analyses are examined, the chance of obtaining an apparently unusual result can increase, depending on how the analyses relate to one another and how results are selected for emphasis. Methods for addressing multiplicity can include prespecified testing sequences or adjusted significance thresholds. Their relevance depends on the study's aims and analysis plan. Because those details are not reported in the registry record, the posted p-values should not be treated as if a multiplicity strategy were known.

The record also does not report the number of participants with missing primary outcome data, whether all randomized participants were included in the analysis, or how missing values were handled. It does not report an intention-to-treat, per-protocol, or other named analysis population. The total enrollment and the arm-level counts do not establish that every participant contributed complete pre-test and post-test measurements.

General: Missing data can affect a result when availability of measurements is related to participant characteristics or outcomes. Analysis choices can also affect the estimate and its uncertainty. Without the amount, timing, and handling of missingness, readers cannot assess whether the analysis represents all randomized participants or a subset with available measurements. The registry does not report sensitivity analyses for missing data, reasons for missingness, or a comparison of included and excluded participants.

These absences limit evaluation of both the multiplicity of the analyses and the population represented by the reported results. They do not demonstrate that an adjustment was omitted in the original study or that data were missing; they mean those details are not reported in the registry record.

13. 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.

The registry reports zero serious adverse events among the three participants at risk in the 2D iPad arm and zero among the three participants at risk in the 3D VR arm. It also reports zero other adverse events in each arm and zero deaths in both arms. No serious adverse event terms are listed.

These figures describe the events recorded in this study. They do not establish that no adverse events could occur, and they provide limited information about how frequently uncommon events might occur. The safety counts are based on three participants at risk in each arm, so the absence of a recorded event in this small study cannot precisely characterize safety in a broader population.

The registry does not report the period over which adverse events were collected, how events were defined or assessed, or whether all participants completed safety follow-up. It also does not provide event-level descriptions beyond the absence of listed serious-event terms. Therefore, the figures should be read as registry-reported counts, with their limited scope, rather than as evidence of a general safety profile.

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

The results describe a small randomized comparison of two VAST delivery formats and a continuous speech measure. The posted arm means for change in MLU are 0.539 for the 2D iPad format and 0.299 for the 3D VR format. The reported between-arm ANOVA comparison of final values is -0.24, with a 95% confidence interval from -1.276 to 0.796 and p < 0.555. The interval is compatible with values on either side of zero, leaving substantial uncertainty about the direction and magnitude of a difference.

These results do not establish that the formats are equivalent. Equivalence requires an appropriate design and a prespecified margin defining what differences would count as practically similar; such a margin is not reported in the registry record. The results also do not establish that either format improves communication or language ability. The study compares two formats, and the posted analysis is not a comparison with no intervention or usual care. The registry lists parent perceptions and other measures as secondary outcomes, but no secondary outcome results are included in the registry record.

The registry's own outcome description refers to pre-to-post gain, whereas its reported between-arm analysis is labeled as a comparison of final values. Without the analysis details, the posted estimate should not be interpreted as a baseline-adjusted contrast or as a definitively analyzed difference in change. No threshold for a meaningful MLU change is reported, so statistical estimates cannot be translated into a claim about practical importance.

General: A randomized study can support a comparison between assigned groups, but the strength and reach of that comparison depend on sample size, conduct, outcome measurement, analysis, and follow-up. With six participants overall, the estimates are imprecise and may be sensitive to individual observations. The results are descriptive and hypothesis-generating, and should not be generalized beyond the population and setting represented by this study without further evidence. They are not a basis for individual treatment decisions.

15. Why This Trial Matters Statistically

This trial raises a useful statistical question: how should a continuous language measure be compared when a small randomized study evaluates two ways of delivering the same video-modeling intervention? The record makes the contrast explicit—2D tablet delivery versus a 3D VR format—while also showing why a group mean alone is not enough. The reported means are accompanied by standard deviations and an inferential comparison, allowing readers to consider both the observed values and uncertainty.

The study also illustrates the importance of defining the estimand: the exact quantity a statistical analysis aims to estimate. The primary outcome description emphasizes change from pre-test to post-test, while the posted between-arm analysis is labeled as a comparison of final values. A final-value comparison, a comparison of changes, and a baseline-adjusted comparison are distinct approaches. The record does not specify which relationship between these descriptions applies, so the analysis label matters when interpreting the result.

The confidence interval is especially informative in a study with three participants per arm. Its span across negative and positive values makes the uncertainty visible and discourages overinterpretation of the point estimate. The example also shows why a non-small p-value should not be read as proof of no difference or as evidence of equivalence. Precision and the range of compatible effects are central to interpretation.

Finally, the registry lists multiple outcomes and analyses without describing a multiplicity strategy, analysis populations, or missing-data handling. Those reporting details are important to understanding how robust and selective the reported findings may be. For biostatisticians and clinical researchers, the record highlights the value of aligning outcome descriptions, analysis labels, and prespecified plans so readers can tell precisely what was estimated and how much uncertainty remains.

16. 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 enrollment was six, with three participants in each arm. This makes the reported group means and standard deviations sensitive to the values of individual participants and limits precision. No participant-level results or baseline characteristics by arm are reported, so it is not possible to examine the distribution of MLU, the balance of participant features, or the influence of individual observations.

The endpoint description is not fully consistent about language-sample duration: it refers to pre- and post-test 15-minute samples, while the scoring description refers to observation during a five-minute period. The registry does not clarify whether these descriptions concern different stages of measurement or how the final scores were obtained. The between-arm analysis is labeled as a comparison of final values even though the outcome description says pre-to-post gain was compared. The analysis details do not resolve this distinction.

The record does not report the randomization method, allocation concealment, assessor masking, treatment fidelity, attendance, or a detailed protocol for administration. It does not report the sample-size justification, analysis plan, model diagnostics, missing-data handling, or analysis population. The record also lists several outcomes and analyses without explaining their mapping or any multiplicity procedure. These limitations constrain assessment of the analysis and its robustness; they do not prove that the methods were inadequate.

The registry gives no threshold for a meaningful MLU change and does not report results for the listed secondary outcomes in the information described. Adverse-event counts are recorded, but the collection period and assessment procedures are not. Generalizability is also limited by the narrow study population and the very small number of participants. The registry record does not provide enough information to determine how findings might apply to other ages, populations, settings, or implementations of the technology.

17. Related Tutorials

18. Related Statistical Calculators

19. Sources

• ClinicalTrials.gov: Video Assisted Speech Technology to Enhance Motor Planning for Speech — NCT04764539.
• Registry record last updated 2023-05-09; status last verified 2023-04.
• Results first posted 2023-05-09.
• 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.

20. Record Summary

NCT04764539 is a completed randomized, parallel-group study sponsored by iTherapy, LLC. It enrolled six children aged 4 to 8 years, with three participants assigned to each VAST delivery format: 2D stimuli on an iPad Pro or 3D stimuli using VR goggles and bone conduction headphones. The recorded study start was 2019-12-01, and primary completion was 2020-11-30. The study was unmasked.

The primary outcome was change in mean length of utterance, measured in morphemes per utterance. The posted means were 0.539 for the 2D arm and 0.299 for the 3D arm. The reported ANOVA comparison of final values was -0.24, with a 95% confidence interval from -1.276 to 0.796 and p < 0.555. The interval indicates substantial uncertainty about the between-arm difference. The record lists three secondary outcomes, six outcome measures, and fourteen analyses. It does not report a sample-size justification, detailed analysis plan, or missing-data approach.