NCT07756398

Brief Summary

Structural heart disease (SHD), including valvular heart disease and left ventricular systolic dysfunction, is common in older adults and often goes undetected until symptoms or complications such as heart failure develop. Current detection relies on opportunistic clinical evaluation, including cardiac auscultation, which has limited sensitivity and is inconsistently performed, and echocardiography, which is accurate but resource-intensive. The IMPROVE trial (IMportance of Point-of-caRe ultrasOund for Early Detection of Valvular and Cardiac DiseasEs) evaluates whether an artificial intelligence-guided point-of-care ultrasound (POCUS) screening strategy, embedded in routine primary care and geriatrics ambulatory visits, increases detection of clinically significant SHD compared with usual care. Provider clusters at three health systems will be randomized 1:1 to the AI-POCUS screening strategy or usual care, and approximately 1,440 patients aged 65 years and older will be enrolled over 24 months. In the screening arm, trained clinic staff will acquire cardiac images using an AI-guided handheld ultrasound device; images will be analyzed by FDA-cleared AI software under core laboratory supervision, and patients with a positive screen will be referred to their primary care provider for consideration of a confirmatory transthoracic echocardiogram. The primary outcome is a new diagnosis of SHD (moderate or greater valvular disease, or left ventricular ejection fraction of 50% or less) confirmed by echocardiography within 90 days of the index visit. Secondary outcomes include recall rate, downstream testing and costs, cardiology or cardiac surgery referrals, and major adverse cardiac events over two years. The trial uses a type 1 hybrid effectiveness-implementation design and will concurrently assess the acceptability, feasibility, and determinants of AI-POCUS adoption in ambulatory practice.

Trial Health

63
Monitor

Trial Health Score

Automated assessment based on enrollment pace, timeline, and geographic reach

Enrollment
1,440

participants targeted

Target at P75+ for not_applicable

Timeline
28mo left

Started Aug 2026

Typical duration for not_applicable

Geographic Reach
1 country

3 active sites

Status
not yet recruiting

Health score is calculated from publicly available data and should be used for screening purposes only.

Trial Relationships

Click on a node to explore related trials.

Study Timeline

Key milestones and dates

Study Progress1%
Aug 2026Dec 2028

Study Start

First participant enrolled

August 1, 2026

Completed
3 days until next milestone

First Submitted

Initial submission to the registry

August 4, 2026

Completed
6 days until next milestone

First Posted

Study publicly available on registry

August 10, 2026

Completed
1.8 years until next milestone

Primary Completion

Last participant's last visit for primary outcome

June 1, 2028

Expected
6 months until next milestone

Study Completion

Last participant's last visit for all outcomes

December 1, 2028

Last Updated

August 10, 2026

Status Verified

August 1, 2026

Enrollment Period

1.8 years

First QC Date

August 4, 2026

Last Update Submit

August 4, 2026

Conditions

Keywords

Point-of-care ultrasoundPOCUSArtificial IntelligenceAI-guided imagingEchocardiographyScreeningEarly detectionHeart FailureOlder AdultsGeriatricsPrimary CareCluster randomized TrialImplementation Science

Outcome Measures

Primary Outcomes (1)

  • New diagnosis of structural heart disease (≥ moderate valvular heart disease or left ventricular systolic dysfunction)

    New diagnosis of ≥ moderate aortic or mitral valvular heart disease or left ventricular systolic dysfunction (ejection fraction ≤50%), confirmed by standard transthoracic echocardiography performed within 90 days of the index clinic visit. Assessed using the disease detection rate (true positives/total number of patients in the study).

    90 days from index clinic visit

Secondary Outcomes (3)

  • Recall Rate

    90 days from index clinic visit

  • Referrals to cardiology or cardiac surgery

    2 years from index clinic visit

  • Major adverse cardiac events (MACE)

    2 years from index clinic visit

Other Outcomes (3)

  • Downstream Processes of Care

    90 days from index clinic visit

  • AI-POCUS Adoption

    6 months post-implementation

  • Acceptability and Feasibility of AI-POCUS

    Through study completion, estimated 2 years

Study Arms (2)

AI-POCUS Screening

EXPERIMENTAL

Trained clinic staff perform AI-guided point-of-care ultrasound using the Kosmos Torso-One device on eligible patients during routine primary care or geriatrics visits. Patients with ≥ moderate valvular heart disease or EF ≤50% are referred for confirmatory echocardiography within 90 days.

Device: Kosmos Torso-One AI-guided Point-of-Care Ultrasound (EchoNous, Inc.)

Usual Care

NO INTERVENTION

Eligible patients receive standard provider-driven care. Echocardiography referral occurs per routine clinical indications such as auscultated murmur or patient-reported symptoms.

Interventions

A single AI-guided handheld POCUS examination performed by trained clinic staff (medical or certified nursing assistants) during a routine primary care or geriatrics visit. Acquisition is standardized, including parasternal long- and short-axis and apical 3- and 4-chamber views, with 2D imaging, color Doppler, M-mode of the mitral valve, and continuous-wave Doppler of the aortic and mitral valves. Images are transmitted to a central echocardiographic core laboratory, which screens for adequacy; adequate studies are analyzed by pre-specified AI algorithms based on 2017 ASE guidelines, blinded to arm assignment. A positive screen is defined as moderate or greater valvular heart disease or left ventricular ejection fraction of 50% or less. Positive results are communicated to the participant's primary care provider with a recommendation to consider confirmatory TTE; ordering remains at provider discretion, and the study team facilitates completion within 90 days of index visit.

AI-POCUS Screening

Eligibility Criteria

Age65 Years - 85 Years
Sexall
Healthy VolunteersNo
Age GroupsOlder Adult (65+)

You may qualify if:

  • Age 65-85 years Presenting for a new or established visit at a primary care or geriatrics clinic at UT Southwestern Medical Center or Parkland Hospital Seen by a provider cluster enrolled in the study No prior known aortic or mitral valvular heart disease No prior known left ventricular systolic dysfunction (EF ≤50%) No prior echocardiography demonstrating structural heart disease

You may not qualify if:

  • Age younger than 65 or older than 85 years Prior known diagnosis of aortic or mitral valvular heart disease Prior known left ventricular systolic dysfunction Prior echocardiography showing structural heart disease Unable or unwilling to provide informed consent

Contact the study team to confirm eligibility.

Sponsors & Collaborators

Study Sites (3)

Samuel & Jean Frankel Cardiovascular Center

Ann Arbor, Michigan, 48109, United States

Location

Parkland Health

Dallas, Texas, 75390, United States

Location

University of Texas Southwestern Medical Center

Dallas, Texas, 75390, United States

Location

MeSH Terms

Conditions

Aortic Valve StenosisMitral Valve InsufficiencyVentricular Dysfunction, LeftHeart Failure

Condition Hierarchy (Ancestors)

Aortic Valve DiseaseHeart Valve DiseasesHeart DiseasesCardiovascular DiseasesVentricular Outflow ObstructionVentricular Dysfunction

Central Study Contacts

Dharam J Kumbhani, MD, SM, MRCP, FACC, FAHA, FSCA

CONTACT

Ritika Dhruve, MBBS, MPH

CONTACT

Study Design

Study Type
interventional
Phase
not applicable
Allocation
RANDOMIZED
Masking
SINGLE
Who Masked
OUTCOMES ASSESSOR
Masking Details
The echocardiographic core laboratory analysts are blinded to participant arm assignment (AI-POCUS vs. usual care) when interpreting all transthoracic echocardiograms and POCUS images.
Purpose
SCREENING
Intervention Model
PARALLEL
Model Details: Stratified cluster-randomized trial. 36 provider clusters (physicians and advanced practice providers) at 3 health systems are randomized 1:1 to AI-guided POCUS screening or usual care. Each health system serves as a randomization stratum contributing 12 clusters (6 per arm). Patients receive the strategy assigned to their provider, averaging 40 patients per cluster (1,440 total). Type 1 hybrid effectiveness-implementation design; primary analysis is intention-to-treat.
Sponsor Type
OTHER
Responsible Party
PRINCIPAL INVESTIGATOR
PI Title
Professor of Medicine

Study Record Dates

First Submitted

August 4, 2026

First Posted

August 10, 2026

Study Start

August 1, 2026

Primary Completion (Estimated)

June 1, 2028

Study Completion (Estimated)

December 1, 2028

Last Updated

August 10, 2026

Record last verified: 2026-08

Data Sharing

IPD Sharing
Will not share

Individual participant data will not be shared at this time. The study involves vulnerable older adult populations across two clinical sites, and data sharing requires additional privacy and governance review. Data sharing may be considered upon study completion pending IRB approval and execution of appropriate data use agreements.

Locations