Artificial Intelligence Detects Cardiac Amyloidosis Before Clinical Diagnosis

Q: What is the clinical evidence from this study?

Study design: Cohort. Population: Amyloidosis (including cardiac amyloidosis) (n=396). Intervention: Externally validated open-source AI echocardiography model vs. Timing of clinical diagnosis of cardiac amyloidosis. Primary outcome: Positive AI detection prior to clinical diagnosis of cardiac amyloidosis.

Q: What does this research mean for the field?

Artificial intelligence applied to pre-diagnostic echocardiograms can identify cardiac amyloidosis in over 40% of patients an average of 273 days prior to clinical diagnosis. Novelty: ClaimNovelty.NOVEL_FINDING. Consensus alignment: ConsensusAlignment.NEUTRAL.

Victoria Yuan; Saisanjana Kalagara; Hirotaka Ieki; Ankeet S. Bhatt; Andrew P. Ambrosy; Aly Elezaby; Lily Kagan Stern; Paul P. Cheng; David Ouyang

doi:10.1016/j.cardfail.2026.06.030

November 1, 2026Journal of Cardiac FailureOpen Access

Artificial Intelligence Detects Cardiac Amyloidosis Before Clinical Diagnosis

View Full Paper

Ask AI

Bookmark

Share

Key result

AI detects cardiac amyloidosis in ~41% of patients prior to clinical diagnosis on historical echo.

n=396

Why the study?

The diagnosis of cardiac amyloidosis is often delayed due to phenotypic mimickers, yet timely detection is imperative because novel disease-modifying treatments have a greater clinical impact earlier in the disease course.

Does an externally validated AI model applied to historical echocardiography detect cardiac amyloidosis prior to clinical diagnosis in patients with amyloidosis?

Key Points

To determine whether an artificial intelligence model applied to historical echocardiograms can identify cardiac amyloidosis prior to formal clinical diagnosis.
Retrospective review of 396 patients with amyloidosis evaluated at Cedars-Sinai Medical Center (2006–2024), of whom 304 had echocardiograms prior to diagnosis and 221 had confirmed cardiac involvement.
Pre-diagnostic echocardiograms were analyzed using an externally validated, open-source artificial intelligence model to calculate the probability of cardiac amyloidosis.
Artificial intelligence detection scores were positive prior to clinical diagnosis in 90 of 221 patients (40.7%) with cardiac amyloidosis and 100 of 304 all-comers (32.9%).
In patients with cardiac amyloidosis, artificial intelligence was positive a mean of 273 days before diagnosis, detecting disease in 30 patients (31.9%) at 6 months and 22 patients (31.4%) at least 1 year prior.
In the all-comer cohort, predictions were positive a mean of 260 days before diagnosis, identifying 34 patients (23.3%) at 6 months and 25 patients (22.3%) at least 1 year prior.

Study Design

Type

Cohort (n=396)

Multicenter

No

Structured PICO

Does an externally validated AI model applied to historical echocardiography detect cardiac amyloidosis prior to clinical diagnosis in patients with amyloidosis?

P

Population

396 patients with a clinical diagnosis of amyloidosis (mean age 72.2 years, 25.0% female) at Cedars-Sinai Medical Hospital between 2006 and 2024, of whom 304 had prior echocardiography studies and 221 had documented cardiac involvement.

E

Exposure

Externally validated, open-source artificial intelligence (AI) deep learning model applied to historical echocardiography studies performed prior to clinical diagnosis to predict the probability of cardiac amyloidosis.

O

Outcome

Incidence of echocardiography imaging prior to clinical diagnosis and rate and lead time of positive AI detection of cardiac amyloidosis prior to clinical diagnosis.surrogate

Ask about this paper

Answers are grounded in this paper and cite their sources.

Implication

An externally validated AI echocardiography model identified cardiac amyloidosis in 40.7% of patients with cardiac involvement a mean of 273 days before their clinical diagnosis, demonstrating potential to accelerate diagnostic testing and earlier initiation of disease-modifying therapy.

Abstract

Cardiac amyloidosis (CA) causes contractile dysfunction through the gradual deposition of amyloid fibrils within the myocardium. However, the diagnosis is often delayed given phenotypic mimickers.1 The approval of novel disease-modifying treatments offers an avenue to stabilize disease progression, with a greater clinical impact seen earlier in the course.2,3 Timely detection is imperative to connect patients with disease-specific management. Echocardiography is the most common form of cardiac imaging.

Discussion

No takes yet. Share an insight, caveat, or question.

Cite This Study

Yuan et al. (2026) conducted a cohort in Amyloidosis (including cardiac amyloidosis) (n=396). Externally validated open-source AI echocardiography model vs. Timing of clinical diagnosis of cardiac amyloidosis was evaluated on Positive AI detection prior to clinical diagnosis of cardiac amyloidosis. AI applied to historical echocardiography detected cardiac amyloidosis in 40.7% of patients prior to clinical diagnosis, identifying disease a mean of 273 days earlier.

synapsesocial.com/papers/6af8f7b3016f830a1d01745c https://doi.org/10.1016/j.cardfail.2026.06.030

View Full Paper

Ask AI

Bookmark

Share

Artificial Intelligence Detects Cardiac Amyloidosis Before Clinical Diagnosis

Key Points

Study Design

Structured PICO

Ask about this paper

Implication

Abstract

Discussion

Cite This Study

Artificial Intelligence Detects Cardiac Amyloidosis Before Clinical Diagnosis

Key Points

Study Design

Structured PICO

Ask about this paper

Implication

Abstract

Discussion

Cite This Study

Keep reading

Keep reading