Trial Outcomes & Findings for Validation of an Artificial Intelligence-based Algorithm for Skeletal Age Assessment (NCT NCT03530098)

NCT ID: NCT03530098

Last Updated: 2021-06-09

Results Overview

Mean absolute difference between dictated final impressions (baseline measure by Radiologist) and the consensus determination of a panel of radiologists following review.

Recruitment status

COMPLETED

Study phase

NA

Target enrollment

1903 participants

Primary outcome timeframe

Up to 10 minutes to acquire the scan; up to 2 days to complete diagnosis review

Results posted on

2021-06-09

Participant Flow

Participant milestones

Participant milestones
Measure
Control (Without-AI)
Diagnosis by radiologists made according to current standard of care methods.
Experiment (With-AI)
Diagnosis by radiologists informed by "BoneAgeModel" Artificial Intelligence (AI) algorithm incorporated into normal radiologist workflows and considered as a factor in the clinical decision making process. The radiologists' diagnosis will be considered final. BoneAgeModel takes in a hand radiograph and gender, and outputs the skeletal (bone) age.
Overall Study
STARTED
939
964
Overall Study
Ground-truth Labeled
741
794
Overall Study
Primary Analysis Set
739
792
Overall Study
COMPLETED
939
964
Overall Study
NOT COMPLETED
0
0

Reasons for withdrawal

Withdrawal data not reported

Baseline Characteristics

Race and Ethnicity were not collected from any participant.

Baseline characteristics by cohort

Baseline characteristics by cohort
Measure
Control (Without-AI)
n=739 Participants
Diagnosis by radiologists made according to current standard of care methods.
Experiment (With-AI)
n=792 Participants
Diagnosis by radiologists informed by "BoneAgeModel" AI algorithm incorporated into normal radiologist workflows and considered as a factor in the clinical decision making process.
Total
n=1531 Participants
Total of all reporting groups
Age, Continuous
11.8 years
STANDARD_DEVIATION 3.6 • n=739 Participants
11.5 years
STANDARD_DEVIATION 3.6 • n=792 Participants
11.7 years
STANDARD_DEVIATION 3.6 • n=1531 Participants
Age, Customized
0-4 years
12 Participants
n=739 Participants
14 Participants
n=792 Participants
26 Participants
n=1531 Participants
Age, Customized
>4-8 years
110 Participants
n=739 Participants
131 Participants
n=792 Participants
241 Participants
n=1531 Participants
Age, Customized
>8-12 years
206 Participants
n=739 Participants
249 Participants
n=792 Participants
455 Participants
n=1531 Participants
Age, Customized
>12-16 years
331 Participants
n=739 Participants
338 Participants
n=792 Participants
669 Participants
n=1531 Participants
Age, Customized
>16-20 years
75 Participants
n=739 Participants
57 Participants
n=792 Participants
132 Participants
n=1531 Participants
Age, Customized
>20 years
5 Participants
n=739 Participants
3 Participants
n=792 Participants
8 Participants
n=1531 Participants
Sex: Female, Male
Female
338 Participants
n=739 Participants
359 Participants
n=792 Participants
697 Participants
n=1531 Participants
Sex: Female, Male
Male
401 Participants
n=739 Participants
433 Participants
n=792 Participants
834 Participants
n=1531 Participants
Race and Ethnicity Not Collected
0 Participants
Race and Ethnicity were not collected from any participant.
Region of Enrollment
United States
739 participants
n=739 Participants
792 participants
n=792 Participants
1531 participants
n=1531 Participants
Skeletal age final impression (mean)
11.6 years
STANDARD_DEVIATION 3.6 • n=739 Participants
11.4 years
STANDARD_DEVIATION 3.5 • n=792 Participants
11.5 years
STANDARD_DEVIATION 3.5 • n=1531 Participants
Skeletal age final impression (categorical)
0-4 years
21 Participants
n=739 Participants
17 Participants
n=792 Participants
38 Participants
n=1531 Participants
Skeletal age final impression (categorical)
>4-8 years
110 Participants
n=739 Participants
131 Participants
n=792 Participants
241 Participants
n=1531 Participants
Skeletal age final impression (categorical)
>8-12 years
191 Participants
n=739 Participants
223 Participants
n=792 Participants
414 Participants
n=1531 Participants
Skeletal age final impression (categorical)
>12-16 years
344 Participants
n=739 Participants
362 Participants
n=792 Participants
706 Participants
n=1531 Participants
Skeletal age final impression (categorical)
>16-20 years
73 Participants
n=739 Participants
59 Participants
n=792 Participants
132 Participants
n=1531 Participants
Skeletal age final impression (categorical)
>20 years
0 Participants
n=739 Participants
0 Participants
n=792 Participants
0 Participants
n=1531 Participants
Clinical histories
Endocrine
391 Participants
n=739 Participants
430 Participants
n=792 Participants
821 Participants
n=1531 Participants
Clinical histories
Orthopedic
82 Participants
n=739 Participants
63 Participants
n=792 Participants
145 Participants
n=1531 Participants
Clinical histories
Congenital/syndrome
9 Participants
n=739 Participants
12 Participants
n=792 Participants
21 Participants
n=1531 Participants
Clinical histories
Medical
16 Participants
n=739 Participants
17 Participants
n=792 Participants
33 Participants
n=1531 Participants
Clinical histories
Other
4 Participants
n=739 Participants
12 Participants
n=792 Participants
16 Participants
n=1531 Participants
Clinical histories
More than one category
20 Participants
n=739 Participants
17 Participants
n=792 Participants
37 Participants
n=1531 Participants
Clinical histories
Not available
217 Participants
n=739 Participants
241 Participants
n=792 Participants
458 Participants
n=1531 Participants

PRIMARY outcome

Timeframe: Up to 10 minutes to acquire the scan; up to 2 days to complete diagnosis review

Population: Primary analysis set: Participants with ground-truth labeled exam results and no bone deformity. Ground-truth labeled: exam was interpreted by a panel of 4 radiologists and their interpretations were averaged to determine a final label.

Mean absolute difference between dictated final impressions (baseline measure by Radiologist) and the consensus determination of a panel of radiologists following review.

Outcome measures

Outcome measures
Measure
Control (Without-AI)
n=739 Participants
Diagnosis by radiologists made according to current standard of care methods.
Experiment (With-AI)
n=792 Participants
Diagnosis by radiologists informed by "BoneAgeModel" AI algorithm incorporated into normal radiologist workflows and considered as a factor in the clinical decision making process.
Paired Difference of Skeletal Age Estimate
5.95 months
Interval 5.53 to 6.37
5.36 months
Interval 5.01 to 5.71

SECONDARY outcome

Timeframe: Up to approximately 4 minutes

Population: Primary analysis set: Participants with ground-truth labeled exam results and no bone deformity. Ground-truth labeled: exam was interpreted by a panel of 4 radiologists and their interpretations were averaged to determine a final label.

Amount of time taken by radiologists when using the BoneAgeModel as compared to when they are not.

Outcome measures

Outcome measures
Measure
Control (Without-AI)
n=739 Participants
Diagnosis by radiologists made according to current standard of care methods.
Experiment (With-AI)
n=792 Participants
Diagnosis by radiologists informed by "BoneAgeModel" AI algorithm incorporated into normal radiologist workflows and considered as a factor in the clinical decision making process.
Time for Diagnosis
142 seconds
Interval 80.0 to 248.0
102 seconds
Interval 59.0 to 196.0

Adverse Events

Control (Without-AI)

Serious events: 0 serious events
Other events: 0 other events
Deaths: 0 deaths

Experiment (With-AI)

Serious events: 0 serious events
Other events: 0 other events
Deaths: 0 deaths

Serious adverse events

Adverse event data not reported

Other adverse events

Adverse event data not reported

Additional Information

Safwan S Halabi, MD

Stanford University

Phone: (650) 721-2850

Results disclosure agreements

  • Principal investigator is a sponsor employee
  • Publication restrictions are in place