Kosmos Guidance Performance Evaluation Data Annotation

Key requirements

Who can take part, as listed by the sponsor. Only the study team can confirm whether you are eligible, so take this list to your doctor.

• Ages 18 to 80 years
• All sexes
• Healthy volunteers accepted
• Adult, Older adult

Study population: The study used retrospective datasets from 2 sources: ECHO-020 Kosmos Guidance, Grading, and Capture Performance Evaluation study, and ECHO-025 Kosmos Guidance Performance Evaluation study. Both datasets comprise fully de-identified, retrospective, echocardiographic clips acquired using the Kosmos diagnostic ultrasound system. Data collection for both studies was performed at EchoNous headquarters in Redmond, Washington. Source data was collected targeting a balanced representation of demographics and clinical characteristics (including gender and BMI) to ensure the generalizability of study findings.

Trial locations and contacts

1 site in 1 country.

About this study

Conditions

Heart, Ultrasound, Cardiac Ultrasound, Cardiac Disease

Being tested

Category 1: Corrective Guidance, Category 2: Incorrect Scanning Window, Category 3: Incorrect Probe Orientation, Category 4: Low Confidence Prediction, Auto-Capture Frame-Level Re-Annotation

Medical subject headings

Heart Diseases

What participants receive

What participants receive. In a randomised study each person is assigned to one group by chance.

• Category 1: Corrective GuidanceDevice

  When the system detects that the probe is near the correct cardiac window but requires a specific directional adjustment, it displays a Category 1 - Corrective Guidance message. Each case in this project consists of two clips: Clip A - Suboptimal View: captured after the sonographer has intentionally moved the probe to a prescribed off-target position; and Clip B - Guided View: captured after the sonographer followed the Category 1 probe motion cue displayed by the system. For each pair of clips, annotators will be asked when Comparing Clip A and Clip B whether following the Category 1 corrective guidance instruction produces a discernible improvement in overall image quality. They will provide an answer of Yes, No, or Unevaluable.

• Category 2: Incorrect Scanning WindowDevice

  When the system detects that the probe is positioned at a fundamentally incorrect anatomical location relative to the target cardiac view, it displays a Category 2 - Incorrect Scanning Window message. Each case in this project consists of two clips: Clip A - Suboptimal View: captured after the sonographer has intentionally a clearly incorrect anatomical location; and Clip B - Guided View: captured after the sonographer followed the Category 2 window-navigation cue and moved to the correct scanning window displayed by the system. For each pair of clips, annotators will be asked when Comparing Clip A and Clip B whether following the Category 2 guidance instruction produces a clip in the correct scanning window. They will provide an answer of Yes, No, or Unevaluable.

• Category 3: Incorrect Probe OrientationDevice

  When the system detects that the probe has been rotated 180° around its longitudinal axis, producing a laterally mirrored image, it displays a Category 3 - Incorrect Probe Orientation message. Each case in this project consists of two clips: Clip A - Suboptimal View: with the probe rotated 180° at the correct cardiac window, producing a mirrored image; and Clip B - Guided View: sonographer corrected the probe orientation in response to the Category 3 message displayed by the system. For each pair of clips, annotators will be asked when Comparing Clip A and Clip B whether following the Category 3 guidance instruction produced a clip that is oriented correctly. They will provide an answer of Yes, No, or Unevaluable.

• Category 4: Low Confidence PredictionDevice

  When the system's confidence in the current image is insufficient to issue a specific directional correction, it displays a Category 4 - Low Confidence Prediction message. Each case in this project consists of two clips: Clip A - Suboptimal View: with the probe rotated 180° at the correct cardiac window, producing a mirrored image; and Clip B - Guided View: sonographer corrected the probe orientation in response to the Category 3 message displayed by the system. For each pair of clips, annotators will be asked when Comparing Clip A and Clip B whether following the Category 3 guidance instruction produced a clip that is oriented correctly. They will provide an answer of Yes, No, or Unevaluable.

• Auto-Capture Frame-Level Re-AnnotationDevice

  The Kosmos Trio Auto-Capture feature automatically triggers clip recording when it detects that a diagnostic quality image has been achieved. Each case in this project is a single captured clip. No comparison clip is provided. The view (PLAX, A4C, or A2C) is indicated for each case. For each frame in this clip, annotators will be asked to assign a quality grade using the ACEP 1-5 image quality scale (where 1 represents a frame in which the relevant anatomy is not visualized or identifiable, and 5 represents a frame in which the image is sharp and the relevant anatomy is well visualized and present in the frame) to each frame.

What is being measured

The primary outcome is the main question the study is designed to answer.

1. Guidance Algorithm Accuracy

   Guidance algorithm accuracy will be evaluated using Positive Predictive Value (PPV), a calculation that compares the diagnostic quality between the suboptimal view following an indicated probe movement and the guided view following the Guidance correction of the probe location. PPV will be calculated for each cardiac view included in this study (PLAX, A4C, and A2C), across all four Guidance categories. PPV is calculated by dividing the number of True Positives (the number of trial pairs in which the post-action clip demonstrates that following the guidance instruction produced the intended improvement for the applicable guidance category) by the number of True Positives plus the number of False Positives (the number of trial pairs in which the post-action clip does not demonstrate the expected improvement for the applicable category). Success is defined as a PPV > 0.80 and 95% CI lower bound > 0.80 for each view. All three views must independently satisfy this criteria.

   Measured: Post data acquisition (2 months)

2. Performance of Auto-Capture

   The accuracy of Auto-Capture will be evaluated using Positive Predictive Value (PPV), a calculation that provides the proportion of clips captured using Auto-Capture that contain a sufficient proportion of diagnostic-quality frames to be considered clinically acceptable. PPV is calculated by dividing the number of True Positives (the number of clips with at least one full cardiac cycle of frames is determined to be diagnostic by expert reviewers, in accordance with the ACEP-based grading scale) by the number of True Positives plus the number of False Positives (the number of clips that do not have at least one full cardiac cycle of frames is determined to be diagnostic by expert reviewers, in accordance with the ACEP-based grading scale). Success is defined as a PPV > 0.80 and 95% CI lower bound > 0.80.

   Measured: Post data acquisition (2 months)

Timeline

From the start of the study to its completion; the mark is today. Estimated dates are the sponsor’s plan.

1. First posted18 Aug 2026
2. Study start5 Jun 2026
3. Primary completion30 Aug 2026 (estimated)
4. Study completion30 Sep 2026 (estimated)

Resources

• Official record on ClinicalTrials.govThe registry entry this page is built from, with its full change history.
• Other trials for HeartStudies of the same condition, open and under way.
• What the trial phases meanHow a treatment moves from first tests in people to approval.

More details

For doctors and researchers: how the study is designed, who runs it, and its registry IDs.