This page provides an independent statistical analysis and educational interpretation of publicly reported results. ClinicalTrials.gov provides the official trial registry record. Numerical trial results on this page are restricted to the results posted in the ClinicalTrials.gov record; calculations added by Clinical Biostats are labelled as such.
1. Trial at a Glance
This page covers the registry results for Functionality of the Pulse Oximeter and Accelerometer Portions of a Novel Wearable Device (NCT05197790), a na na study. The primary outcome is Mean Absolute Error of Dove Device Compared to FDA Approved SpO2 Sensor.
| Feature | Registry value |
|---|---|
| ClinicalTrials.gov identifier | NCT05197790 |
| Phase | NA |
| Condition | Hypoxic Events |
| Allocation | Not applicable (single group) |
| Design model | Single Group |
| Masking | None (open label) |
| Enrollment | 27 |
| Arms in the primary outcome | n/a |
| Lead sponsor | University of Pennsylvania |
| Results posted | Yes |
2. Clinical Question
The study asks whether the pulse oximeter portion of a novel wearable device can measure oxygen saturation in a way that agrees with an FDA-approved SpO2 sensor. The primary endpoint is the mean absolute error between the Dove device and the FDA-approved sensor, expressed in percentage points over a 15 hours timeframe. The registry description also says skin tone will be incorporated as a factor in assessing SpO2 function and whether the device can reasonably detect hypoxia. No numerical endpoint result or skin-tone analysis is reported in the registry record.
Population
The study enrolled participants in a single-center assessment of a wearable device prototype. For the primary endpoint, 2 participants had a corrupted data file and 1 patient had the device applied without an SD card; the remaining 24 patients were analyzed.
Intervention
The intervention is the FDA-approved pulse oximeter and DOVE device prototype. The study description focuses on the prototype’s pulse oximeter and accelerometer portions.
Comparator
The primary endpoint compares the Dove device with an FDA-approved SpO2 sensor. This is a measurement reference, not a separate randomized treatment arm.
Primary question
What is the mean absolute error between the prototype’s SpO2 readings and the FDA-approved sensor’s readings over 15 hours, and how is skin tone incorporated into assessment of SpO2 function?
The record also lists a questionnaire about device comfort as a secondary outcome. It does not report results for that questionnaire or a separate numerical result for the accelerometer function. The question is therefore about prototype functionality and measurement comparison, rather than the effect of one treatment versus another.
3. Trial Design
This is a completed, single-center study of a wearable device prototype, registered as NCT05197790. The registry lists the study as starting on 2022-05-20 and having a primary completion date of 2024-12-21. The enrollment is 27. The study is listed as phase NA, with allocation NA, masking NONE, and a SINGLE GROUP model. The arm description says everyone will be in the same arm, named “Assess device prototype functionality.”
The registry does not describe a randomized comparison between treatment groups. The FDA-approved SpO2 sensor functions as the measurement reference for the primary endpoint; it is not a like-for-like control arm. The primary endpoint description says that at each time point, the difference between the device and the FDA-approved standard will be calculated. It then identifies mean absolute error as a single summary statistic and Bland-Altmann analysis as a way to represent limits of agreement graphically and numerically.
The primary endpoint timeframe is 15 hours. The registry summary says the study’s purpose is to assess the function of an oxygen sensor and a movement sensor in a novel device prototype. It also lists a questionnaire about comfort of the device as a secondary outcome. Although the study description names both pulse oximeter and accelerometer portions, the posted primary endpoint is specifically about SpO2 measurement error. A separate accelerometer endpoint result is not reported in the registry record.
The record gives no further detail about the measurement schedule, how many readings were obtained, how the device was applied beyond the stated SD-card issue, or the exact procedures for comparing the sensors. A more detailed protocol, including any statistical design rationale or planned analysis procedures beyond the endpoint description, is not reported in the registry record. The registry status is COMPLETED, but completion alone does not establish that every planned assessment has a posted numerical result.
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.
Before assignment: 3 patients had technical failures in the device's storage mechanism, and so their demographics and survey results are used, but their SpO2 data is not.
Overall Study
| Stage | Primary Arm |
|---|---|
| STARTED | 27 |
| COMPLETED | 27 |
| NOT COMPLETED | 0 |
Analysis population for the primary outcome, as defined in the registry: 2 participants had a corrupted data file. 1 patient had the device applied without an SD card. The remaining 24 patients were analyzed.
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.
| Characteristic | Unit | Primary Arm |
|---|---|---|
| Participants analysed | 27 | |
| Age, Continuous | years | 55 (38.5 to 69.5) |
| Sex: Female, Male: Female | Participants | 11 |
| Sex: Female, Male: Male | Participants | 16 |
| Race (NIH/OMB): American Indian or Alaska Native | Participants | 0 |
| Race (NIH/OMB): Asian | Participants | 2 |
| Race (NIH/OMB): Native Hawaiian or Other Pacific Islander | Participants | 0 |
| Race (NIH/OMB): Black or African American | Participants | 8 |
| Race (NIH/OMB): White | Participants | 17 |
| Race (NIH/OMB): More than one race | Participants | 0 |
| Race (NIH/OMB): Unknown or Not Reported | Participants | 0 |
| Region of Enrollment: United States | Participants | 27 |
| Body mass index | kilogram per meter squared | 28.5 (26.5 to 35.1) |
6. Primary Endpoint
The registry lists the primary outcome as Mean Absolute Error of Dove Device Compared to FDA Approved SpO2 Sensor, measured over 15 hours.
| Element | Registry entry |
|---|---|
| Result type | Mean |
| Measure of spread | 95% Confidence Interval |
| Unit | percentage points |
| Endpoint classification | Continuous Endpoint |
The primary endpoint is the mean absolute error of the Dove device compared with an FDA-approved SpO2 sensor. Its unit is percentage points, its parameter is a mean, and its timeframe is 15 hours. The endpoint description says that the difference between the device and the standard will be calculated at each time point, with mean average error used as a single summary statistic. It also states that skin tone will be incorporated as a factor in assessing SpO2 function and whether the device can reasonably detect hypoxia.
In general, absolute error describes the size of a difference without retaining whether the device reading is higher or lower than its reference. Averaging absolute errors gives a summary of typical discrepancy in the measured units. This makes the endpoint useful for describing how close device readings are to the reference on average, but it does not by itself describe the direction of any systematic bias or the range of differences for individual readings.
The registry also names Bland-Altmann analysis to graphically and numerically represent limits of agreement. In general, this type of analysis complements an average error by examining the distribution and spread of paired differences. Those summaries can address a different question from correlation: two devices may track changes similarly yet still differ in their readings. No mean absolute error, confidence interval, or limits of agreement are reported in the registry record, so the magnitude and precision of agreement cannot be assessed from the posted results.
7. Why No Re-analysis Is Shown
8. Statistical Interpretation
The posted record identifies a clear measurement question but does not provide the numerical result needed to answer it. The primary endpoint is the mean absolute error between the Dove device and an FDA-approved SpO2 sensor, expressed in percentage points, over 15 hours. However, the registry record does not report the estimated mean, its stated 95% Confidence Interval, or numerical limits of agreement. The record therefore does not show how large the average discrepancy was or how precisely it was estimated.
A mean absolute error is a summary of the size of the differences between device and reference readings. In general, a smaller value would indicate smaller average absolute differences in the observed comparisons, but that interpretation requires the actual estimate and context about acceptable measurement performance. Neither a numerical estimate nor a criterion for acceptable error is reported in the registry record. No conclusion about the level of agreement, or about the ability to detect hypoxia, can be drawn from the endpoint description alone.
The registry says skin tone will be incorporated as a factor in assessing SpO2 function. That is a stated analytic consideration, not a reported finding. No skin-tone-specific estimates, group descriptions, interaction estimates, or results are reported in the registry record. The result also does not include a numerical summary for the accelerometer function, despite the study’s broader purpose referring to both oxygen and movement sensors.
The analysis population described for the primary endpoint is the remaining 24 patients after 2 participants had corrupted data files and 1 patient had the device applied without an SD card. This makes clear that the primary endpoint did not use all enrolled participants as described in the record. The record does not explain the timing or pattern of the missing or unusable data, or whether the endpoint analysis accounted for missing readings within the analyzed participants.
The study is a single-group functionality assessment. The FDA-approved sensor is a reference for comparing measurements, not a randomized treatment comparator. Thus, the design can address agreement between the prototype and reference measurements as analyzed, but it does not provide a randomized comparison of clinical outcomes. Since the numerical endpoint result is not reported in the registry record, even the measurement-agreement question remains quantitatively unresolved in the posted information.
The registry lists 0 serious adverse events, 0 other adverse events, and 0 deaths among the 27 listed as at risk in the primary arm. These entries describe the safety counts in the record; they do not supply a detailed account of surveillance procedures, reporting completeness, or the relationship of any event to the device. The absence of posted primary endpoint values and the limited detail on analysis mean that neither accuracy nor safety should be inferred beyond the recorded counts and descriptions.
Any planning illustration based on an observed effect should be read as a planning exercise, not as a measure of observed power or as a substitute for the missing outcome estimate. The registry’s sample-size assumptions describe a hypothetical two-arm planning calculation, while the study itself is listed as a single-group design. Those assumptions do not establish a planned randomized design for this study.
9. Statistical Methods Explained
The primary endpoint is a continuous measurement summarized as a mean. The registry labels its parameter MEAN and its dispersion as 95% Confidence Interval. The analysis description says the device-to-reference difference will be calculated at each time point, mean average error will be used as a single summary statistic, and Bland-Altmann analysis will be used to represent limits of agreement. These descriptions identify the intended forms of summary, but the registry record does not report the numerical analysis results or additional implementation details.
What does mean absolute error describe?
In general, for each paired device and reference reading, an error can be defined as the device reading minus the reference reading. Taking the absolute value removes the sign, so the error records the size of the difference rather than whether the device reads above or below the reference. Averaging those absolute differences yields mean absolute error. In this study, the unit is percentage points. This summary is intuitive as a measure of average discrepancy, but it does not show whether the device tends to over-read or under-read. It also does not describe the full spread of individual differences.
The registry uses the phrase “Mean Absolute Error” in the endpoint title and “Mean average error” in the endpoint description. Both refer in context to a mean summary of absolute differences. The posted record does not include the actual calculation, the number of readings contributing to it, or whether readings were averaged first within participants. Those details matter because repeated readings from one participant are not generally equivalent to independent readings from different participants. The analysis approach to within-participant correlation is not reported in the registry record.
Why is a confidence interval useful?
In general, a confidence interval expresses uncertainty around an estimated quantity under the assumptions of the analysis. For a mean absolute error, it would provide information about the precision of the estimated average discrepancy, beyond the point estimate alone. The registry lists a 95% Confidence Interval as the endpoint’s dispersion measure, but the interval’s lower and upper values are not reported in the registry record. Without those values, the precision of the estimated mean cannot be judged. A confidence interval is not a guarantee that a particular range contains the true value; its interpretation depends on the statistical procedure and assumptions used.
The registry record does not state whether the interval accounts for clustering of multiple readings within participants, how missing readings were handled, or which method was used to calculate the interval. It also does not report a prespecified threshold for acceptable measurement error. Those omissions prevent a reader from assessing whether the interval was constructed appropriately for the data structure or whether the resulting estimate met a planned performance criterion.
What do limits of agreement add?
In general, Bland–Altman methods examine the differences between two measurement methods in relation to their average readings. A mean difference can describe systematic bias, while limits of agreement describe the range within which many observed differences may lie under the method’s assumptions. This gives information about the distribution of individual discrepancies that a mean absolute error alone cannot provide. A device can have a modest average error yet still have wide variation in individual differences; therefore, average error and limits of agreement address related but distinct aspects of agreement.
The registry states that Bland-Altmann analysis will be used graphically and numerically to represent limits of agreement. It does not report the mean difference, the limits, the plot, or the method used to calculate the limits. Details such as whether the differences vary with the average measurement, and whether the distributional assumptions are reasonable, are not reported in the registry record. No numerical interpretation of the limits is therefore possible from the posted results.
How should the reference comparison be understood?
The FDA-approved SpO2 sensor is the reference named for the primary endpoint. In general, comparing a new measurement device with a reference is a method-comparison exercise, not a randomized treatment comparison. The purpose is to characterize how measurements from the two methods differ. The registry’s single-group design and lack of a randomized comparison should be kept distinct from the paired measurement comparison described in the endpoint. The record does not give enough procedural detail to establish exactly how readings were paired or timed beyond stating that differences would be calculated at each time point.
Skin tone is named as a factor in the endpoint description. In general, examining a factor can help assess whether measurement performance varies across that characteristic. However, the registry record does not state how skin tone was measured or categorized, how it entered the analysis, or whether subgroup estimates or interaction tests were planned. No results by skin tone are reported, so the stated consideration cannot be treated as evidence that performance was similar or different across skin tones.
The study description refers to an accelerometer as well as a pulse oximeter, but the primary endpoint is specific to SpO2 measurement. No accelerometer performance metric or numerical result is reported in the registry record. The listed comfort questionnaire is a secondary outcome; its instrument, scoring approach, and results are not reported. Thus, the methods described for the primary endpoint should not be assumed to cover the movement sensor or comfort assessment.
What can be learned from planning calculations?
In general, a sample-size calculation based on an observed effect is a planning illustration: it asks what sample size might be needed under stated assumptions to detect an effect of a specified size. It is not observed power and cannot replace the outcome estimate or its uncertainty. The registry’s planning assumptions describe two-sided significance level 0.05, equal allocation between two arms, and a normal approximation, with the detectable effect defined as a fraction of the observed effect. These are planning assumptions, not the listed single-group study design. The record does not report the trial’s own sample-size justification, and the planning assumptions do not supply the missing primary endpoint result.
10. Multiplicity, Missing Data and Analysis Populations
The registry lists 2 outcome measures and identifies the primary endpoint as mean absolute error for the Dove device compared with an FDA-approved SpO2 sensor. It also lists a questionnaire about device comfort as a secondary outcome. The record reports 0 analyses. Numerical results for the primary endpoint and the comfort questionnaire are not reported in the registry record. A separate accelerometer result is also not reported, although the study’s purpose refers to a movement sensor.
Multiplicity arises when several outcomes, comparisons, time points, or subgroup analyses are examined. In general, the more analyses that are performed, the greater the chance that at least one result may appear noteworthy by chance if no adjustment or prespecified hierarchy is used. The registry record does not report a multiplicity strategy, a hierarchy among outcomes beyond identifying the primary endpoint, or which analyses were actually conducted. The analysis plan and any adjustment for multiple testing are not reported in the registry record.
For the primary endpoint, the record states that 2 participants had a corrupted data file and 1 patient had the device applied without an SD card; the remaining 24 patients were analyzed. This describes the endpoint analysis population and the stated reasons for not analyzing those records. It does not explain whether missingness occurred at the participant or reading level beyond those descriptions, whether other readings were incomplete, or whether the reasons were related to measurement quality or participant characteristics. The handling of missing readings within analyzed participants is not reported in the registry record.
The record does not describe an intention-to-analyze population, a sensitivity analysis, imputation, or another method for assessing the effect of missing data. It also does not report whether the comfort questionnaire or any accelerometer assessment used the same analysis population. Consequently, the extent to which the posted analyses represent all enrolled participants across all outcomes cannot be determined from the registry record.
11. Safety Results
The registry reports serious adverse events as affected participants divided by participants at risk.
| Arm | Participants with serious adverse events | Participants at risk | Percentage |
|---|---|---|---|
| Primary Arm | 0 | 27 | 0.0% |
Deaths
| Arm | Deaths (participants affected of at risk) |
|---|---|
| Primary Arm | 0 of 27 |
Other (non-serious) adverse events
Participants with at least one non-serious event at or above the registry's 0% reporting threshold.
| Arm | Participants affected of at risk |
|---|---|
| Primary Arm | 0 of 27 |
The registry lists one arm, “Primary Arm,” with 27 at risk for serious adverse events and 0 affected. It also lists 27 at risk for other adverse events and 0 affected, with 0 deaths. No serious adverse-event terms are listed. These are the recorded counts in the registry; they indicate that the record contains zero events in the listed categories for this arm.
These counts should be read as descriptive safety reporting, not as proof that no event occurred outside the information captured in the registry. The record does not report the adverse-event collection method, observation period for safety outcomes, definitions used for events, or whether events were assessed for relation to the device. It also does not provide event-level details beyond the zero counts and the absence of listed serious-event terms.
The study has no separate randomized comparison arm, so the safety figures do not estimate a comparative difference between treatments. The registry record also does not report a statistical analysis of adverse events. The counts should therefore be stated plainly and kept within their scope: the posted record lists 0 serious adverse events, 0 other adverse events, and 0 deaths in the primary arm.
12. What the Result Does, and Does Not, Mean
The study is designed to assess the functionality of a wearable device prototype, with the primary endpoint focused on agreement between the device’s SpO2 readings and an FDA-approved sensor. That is a measurement question. It is different from asking whether use of the device changes a clinical outcome, and it is different from comparing two randomized treatment strategies. The record describes the intended comparison and its summary methods, but it does not report the numerical findings.
Without a reported mean absolute error or 95% Confidence Interval, readers cannot determine the average size of the observed discrepancies or the precision of that estimate. Without numerical limits of agreement, the variation in individual differences cannot be assessed. The endpoint title and analysis description explain what was intended to be measured; they do not establish the observed level of agreement or whether any performance criterion was met. No acceptable-error threshold is reported in the registry record.
The description says skin tone will be incorporated as a factor in assessing SpO2 function and reasonable detection of hypoxia. This intention does not demonstrate that performance was evaluated successfully across skin tones. No skin-tone-specific results or numerical hypoxia-detection measure is reported in the registry record. Likewise, the record’s broader stated purpose includes the accelerometer, but no separate movement-sensor result is reported. A comfort questionnaire is listed, but its results are also absent.
The primary endpoint population excludes the records described as corrupted data or lacking an SD card, leaving the remaining 24 patients analyzed. That accounting is important when considering how widely the endpoint estimate might apply, but the record does not provide enough information to evaluate the effect of excluded or missing data. It does not describe whether data were missing for particular measurements or whether the missingness could be related to performance.
The posted safety counts are limited descriptive information and should not be treated as a complete assessment of device safety. The study record lists 0 serious adverse events, 0 other adverse events, and 0 deaths, but does not describe the event ascertainment procedures or observation period. Overall, the registry record documents the study question and selected design features but does not provide the outcome estimates needed to judge measurement agreement, subgroup performance, comfort, or accelerometer functionality.
13. Why This Trial Matters Statistically
Device evaluation requires more than showing that a sensor produces readings. A measurement method is useful only to the extent that its readings are sufficiently close to an appropriate reference for the intended purpose. This study’s primary endpoint addresses that issue directly by specifying mean absolute error against an FDA-approved SpO2 sensor and by naming Bland-Altmann limits of agreement. These are complementary ideas: average error describes the typical size of discrepancies, while agreement limits describe the spread of differences.
The endpoint description also identifies skin tone as a factor in assessing SpO2 function. Statistically, this matters because an overall average can conceal variation across characteristics relevant to measurement performance. However, the record does not report how skin tone was assessed or any results by skin tone. The stated analytic consideration is not itself evidence about subgroup performance.
The study is also a reminder to distinguish a reference measurement from a randomized control arm. The FDA-approved sensor is used as a comparator for readings, while the registry describes a single-group study with everyone in the same arm. That structure is appropriate to understand as method comparison rather than as a treatment-effect comparison. It does not answer questions about clinical outcomes that are not measured or reported.
Reporting the analysis population is another important statistical feature. The record states that 2 participants had corrupted data files and 1 patient had the device applied without an SD card, leaving 24 patients analyzed for the primary endpoint. Such accounting helps readers understand which observations contributed to the stated endpoint. Yet, without numerical results, details of missing-data handling, or a fuller description of repeated readings, the limits of inference remain substantial.
The registry record contains endpoint definitions and safety counts, but no numerical primary result or posted analyses. That distinction is important: a clearly described statistical plan or endpoint does not substitute for estimates and their uncertainty. The study’s statistical importance lies in the measurement-agreement question it poses and in the need for transparent results that show both average error and the distribution of differences.
14. 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 most consequential limitation of the registry record is that it does not report numerical primary endpoint results. The mean absolute error, its 95% Confidence Interval, and the Bland-Altmann limits of agreement are not reported in the registry record. Consequently, the magnitude, precision, direction, and spread of measurement differences cannot be assessed from the posted information.
The record describes the study as single-center and single-group, with allocation NA and masking NONE. The comparison with an FDA-approved sensor is a measurement reference rather than a randomized control arm. The record therefore does not support treatment-effect conclusions. The enrollment is 27, while the primary endpoint description states that 2 participants had corrupted data files and 1 patient had the device applied without an SD card; the remaining 24 patients were analyzed. The reasons for these exclusions are stated, but the potential effect of excluded data is not evaluated in the registry record.
Important implementation details are also absent. The record does not describe the exact timing and number of readings, the approach to repeated measurements within participants, the method for calculating the confidence interval or agreement limits, or how missing readings were handled. The approach to skin-tone measurement and analysis is not reported. Although the study purpose includes accelerometer functionality, no separate accelerometer outcome result is reported. The comfort questionnaire is listed as a secondary outcome, but its instrument, scoring, and results are not reported.
The registry lists zero serious adverse events, zero other adverse events, and zero deaths, but does not report safety collection procedures or an observation period. The sample-size assumptions describe a planning illustration using a hypothetical two-arm framework; the registry does not report the study’s own sample-size justification. Taken together, the missing results and limited methods detail prevent a complete assessment of measurement performance, missing-data effects, subgroup behavior, and safety reporting.
15. Related Tutorials
16. Related Statistical Calculators
17. Sources
- ClinicalTrials.gov: Functionality of the Pulse Oximeter and Accelerometer Portions of a Novel Wearable Device — NCT05197790.
- Registry record last updated 2026-03-17; status last verified 2026-03.
- Results first posted 2026-03-17.
- 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.
Continue through the Clinical Biostats statistical pathway
Use this trial's endpoint as a starting point for deeper study of effect measures, confidence intervals, and sample-size planning.
18. Record Summary
NCT05197790 is a completed, single-center, single-group study of the functionality of the pulse oximeter and accelerometer portions of a novel wearable device. The registry lists enrollment of 27, allocation NA, masking NONE, and a primary completion date of 2024-12-21. The primary endpoint is mean absolute error between the Dove device and an FDA-approved SpO2 sensor, measured in percentage points over 15 hours; the description also names skin tone as a factor and Bland-Altmann analysis for limits of agreement.
The primary endpoint population description states that 2 participants had corrupted data files and 1 patient had the device applied without an SD card; the remaining 24 patients were analyzed. A questionnaire about device comfort is listed as a secondary outcome. The registry record does not report numerical primary endpoint results, comfort results, skin-tone-specific results, or a separate accelerometer result. It lists 0 serious adverse events, 0 other adverse events, and 0 deaths in the primary arm. The available record defines the study question and some design features, but does not provide the outcome estimates needed to assess measurement agreement.