Skip to main content
Tessa NewsLink
Paste a health news link, or browse
Latest checks

Welcome to the Tessa News Link Beta! We are actively improving this experience.

Share Feedback

Source study found

Story checked

Doctors take minutes to measure heart function. AI takes seconds (opens in a new tab)

news-medical.net · 2026-11-06

Short answerEvidenceSource

Short answer

Mixed

Mixed.

The claims we could check match the study, but some claims were not covered by the evidence reviewed.

  • 2 supported
  • 2 not covered

Checked against the study summary. The full text wasn't available, so some details couldn't be settled either way.

Notice: Analysis is currently derived from study abstracts. Full-text paper analysis capabilities are currently in development.

Share this check

Help improve NewsLink

Tell us what worked or what needs attention in this check.

Follow the evidence trail
1
2

NewsLink checks it

Mixed

Every claim we could check holds up. Two of four claims match the study. This overall rating is based only on the claims we could check. Two claims the study doesn't address.

  • 2 supported
  • 2 not covered
Open claim evidence
3
Then inspect each claim

Evidence layer

Claim by claim

Each claim gets a verdict. Expand it to see the evidence directly below.

4 claims in this story

Showing all 4 claimsChoose a verdict to focus the list.

Claim 1 of 4Not coveredThe AI algorithm processed cardiac MRI scans in approximately 3–4 seconds per patient, compared with a median of 13 minutes per patient for clinician segmentation in an earlier study.View evidenceHide evidence

As stated3–4 seconds per patient versus 13 minutes per patient

Why this verdict

The abstract-level profile does not report AI processing time, clinician segmentation time, or the earlier-study comparison of approximately 3–4 seconds versus 13 minutes per patient. This may require full-text or external-source verification.

Claim 2 of 4Not coveredThe researchers cautioned that AI measurements were not error-free, with segmentation errors in 6 of 600 randomly selected patients, and that prospective studies are needed to establish clinical utility, cost-effectiveness, and impact on patient care.View evidenceHide evidence

As stated6 of 600 patients; 1% error rate

Why this verdict

The abstract-level profile supports the general caveat that the work was retrospective and that prospective evaluation is needed, but it does not report segmentation errors in 6 of 600 randomly selected patients, an exact 1% error rate, or the full stated list of prospective endpoints such as cost-effectiveness and impact on patient care. At abstract depth, the detailed error-rate portion is not verifiable.

1

Study evidence

AI-derived ejection fraction yielded better overall model fit and discrimination for predicting all-cause mortality than clinician-derived ejection fraction in the full cohort.AIC 57,936 (AI) vs 58,097 (clinician); ΔAIC = 161 favoring AI. C index 0.669 (AI) vs 0.637 (clinician); ΔC = 0.033 (95% CI 0.026–0.041). Likelihood-ratio tests: adding AI to clinician χ2 = 168 (p < 0.0001); adding clinician to AI χ2 = 7.5 (p = 0.023).

“In this retrospective, multicentre, prognostic study, we studied a cohort of 27 382 consecutive patients who underwent cardiac MRI (CMR) in the UK and the USA.”
Claim 3 of 4SupportedAI-derived measurements of left-ventricular ejection fraction from cardiac MRI scans predicted all-cause mortality slightly better than clinician-derived measurements, according to a retrospective study published in The Lancet Digital Health.View evidenceHide evidence

As statedslightly better

Why this verdict

The abstract-level profile supports the story’s lead claim: this was a retrospective multicentre prognostic cohort study of cardiac MRI patients comparing AI-derived versus clinician-derived left-ventricular ejection fraction for all-cause mortality prediction. The profile describes AI-derived EF as having marginally/slightly better model fit and discrimination, including a small age/sex-adjusted C-index difference.

1

Study evidence

AI-derived ejection fraction yielded better overall model fit and discrimination for predicting all-cause mortality than clinician-derived ejection fraction in the full cohort.AIC 57,936 (AI) vs 58,097 (clinician); ΔAIC = 161 favoring AI. C index 0.669 (AI) vs 0.637 (clinician); ΔC = 0.033 (95% CI 0.026–0.041). Likelihood-ratio tests: adding AI to clinician χ2 = 168 (p < 0.0001); adding clinician to AI χ2 = 7.5 (p = 0.023).

“In this retrospective, multicentre, prognostic study, we studied a cohort of 27 382 consecutive patients who underwent cardiac MRI (CMR) in the UK and the USA.”
Claim 4 of 4SupportedThe retrospective study included 27,382 patients who underwent cardiac MRI at two centers in the UK and USA, and during a median follow-up of 5.6 years, 13% died.View evidenceHide evidence

As stated27,382 patients; 5.6 years; 13% died

Why this verdict

The profile reports 27,382 consecutive cardiac MRI patients from Barts Heart Centre and UPMC, corresponding to UK and USA centers/systems, with median follow-up of 5.6 years and 3,458 deaths, reported as 13%.

1

Study evidence

AI-derived ejection fraction yielded better overall model fit and discrimination for predicting all-cause mortality than clinician-derived ejection fraction in the full cohort.AIC 57,936 (AI) vs 58,097 (clinician); ΔAIC = 161 favoring AI. C index 0.669 (AI) vs 0.637 (clinician); ΔC = 0.033 (95% CI 0.026–0.041). Likelihood-ratio tests: adding AI to clinician χ2 = 168 (p < 0.0001); adding clinician to AI χ2 = 7.5 (p = 0.023).

“In this retrospective, multicentre, prognostic study, we studied a cohort of 27 382 consecutive patients who underwent cardiac MRI (CMR) in the UK and the USA.”
2

Study evidence

Adding AI-derived ejection fraction to a model already containing clinician-derived ejection fraction produced a large, statistically significant improvement in prognostic model fit for all-cause mortality (likelihood-ratio χ2=168, p<0·0001).χ2=168, p<0·0001

“…the incremental value of AI over clinician segmentation using likelihood-ratio tests of nested models.”
Then look for missing context

Context layer

What the story left out

Important study details the story did not include.

  • Incremental prognostic value analysis: adding AI EF to clinician EF improved model fit substantially, while adding clinician EF to AI EF produced only slight additional improvement.

    The profile treats nested Cox-model incremental-value testing as a separate material analytic element, but the presented story claims do not mention the likelihood-ratio tests or the asymmetry in incremental value.

    From Nested Cox models; likelihood-ratio tests

  • Nested-model limitations: full covariate specification, proportional-hazards checks, handling of missing data, site effects/external validation, and some calibration/model-validity details are not reported at abstract depth.

    The story notes that risk calibration was not assessed, but it does not mention several other interpretation-relevant limitations identified in the profile for the nested Cox analyses, including incomplete model specification, proportional-hazards checks, missing-data handling, site effects, and external validation details.

    From Nested Cox models; likelihood-ratio tests

4 things the story did carry across
  • Retrospective multicentre prognostic cohort design comparing AI-derived versus clinician-derived cardiac MRI left-ventricular ejection fraction for all-cause mortality prediction.
  • Cohort size, setting, follow-up, and deaths: 27,382 cardiac MRI patients from Barts Heart Centre and UPMC/UK-USA systems; median follow-up 5.6 years; 3,458 deaths, 13%.
  • Main finding: AI-derived ejection fraction had marginally better mortality-prediction model fit and discrimination than clinician-derived ejection fraction, especially small after age/sex adjustment.
  • Retrospective nature and need for prospective validation before clinical utility is established.
Then read the study layer

Study layer

Study at a glance

Scan the study first. Expand only the parts you want to inspect.

Pieces of work

2

Evidence read

study summary

Lead result

secondary data

1Lead resultsecondary dataAssess whether deep-learning AI-derived LV segmentation/ejection fraction from cardiac MRI predicts all-cause mortality better than clinician-derived segmentation/ejection fraction in a large retrospective multicentre cohort.Retrospective multicentre prognostic cohortExpand

In plain English

Retrospective multicentre prognostic cohort study (27,382 consecutive CMR patients from Barts Heart Centre and UPMC) comparing deep-learning AI-derived versus clinician-derived left-ventricular ejection fraction for prediction of all-cause mortality using Cox proportional hazards models; AI-derived ejection fraction showed marginally better model fit and discrimination than clinician-derived ejection fraction.

Key findings

  • AI-derived ejection fraction yielded better overall model fit and discrimination for predicting all-cause mortality than clinician-derived ejection fraction in the full cohort.AIC 57,936 (AI) vs 58,097 (clinician); ΔAIC = 161 favoring AI. C index 0.669 (AI) vs 0.637 (clinician); ΔC = 0.033 (95% CI 0.026–0.041). Likelihood-ratio tests: adding AI to clinician χ2 = 168 (p < 0.0001); adding clinician to AI χ2 = 7.5 (p = 0.023).
  • After adjustment for age and sex, AI-derived ejection fraction still showed modestly better model fit and slightly higher discrimination than clinician-derived ejection fraction.AIC 55,684 (AI) vs 55,742 (clinician); ΔAIC = 59 favoring AI. C index 0.781 (AI) vs 0.777 (clinician); ΔC = 0.004 (95% CI 0.002–0.006).
“In this retrospective, multicentre, prognostic study, we studied a cohort of 27 382 consecutive patients who underwent cardiac MRI (CMR) in the UK and the USA.”
What this piece can’t prove
  • Retrospective study design (authors state findings should be investigated in prospective studies).

1 further detail could not be confirmed from the summary.

2secondary dataQuantify the incremental prognostic value of AI-derived ejection fraction beyond clinician-derived ejection fraction (and vice versa) using nested Cox models and likelihood-ratio testing.Nested Cox models; likelihood-ratio testsExpand

In plain English

Using nested Cox proportional-hazards models and likelihood-ratio tests on a retrospective multicentre cohort (n=27,382; median follow-up 5.6 years), the study quantified the incremental prognostic value of AI-derived left-ventricular ejection fraction (EF) versus clinician-derived EF for prediction of all-cause mortality. Adding AI EF to a model containing clinician EF produced a large, statistically significant improvement in model fit (χ2=168, p<0·0001). Adding clinician EF to a model containing AI EF produced a much smaller but nominally significant improvement (χ2=7·5, p=0·023).

Key findings

  • Adding AI-derived ejection fraction to a model already containing clinician-derived ejection fraction produced a large, statistically significant improvement in prognostic model fit for all-cause mortality (likelihood-ratio χ2=168, p<0·0001).χ2=168, p<0·0001
  • Adding clinician-derived ejection fraction to a model already containing AI-derived ejection fraction produced a much smaller improvement in model fit (likelihood-ratio χ2=7.5, p=0·023), indicating only slight incremental prognostic value beyond AI EF in these models.χ2=7.5, p=0.023
“…the incremental value of AI over clinician segmentation using likelihood-ratio tests of nested models.”
What this piece can’t prove

3 further details could not be confirmed from the summary.

Finally, the search trail

Method layer

NewsLink found the paper. Tessa takes you deeper.

NewsLink checks the story. Tessa is where you inspect the paper, authors, evidence, and research context.

Papers considered

The selected paper, plus nearby candidates.

PubMed, Europe PMC, Crossref · 15 candidate papers

Candidate

Validation of an Artificial Intelligence Left Ventricular Ejection Fraction Quantification Software with Cardiac Magnetic Resonance Imaging in Consecutive Non-Selected Patients

2020 · Crossref

Source (opens in a new tab)

And 9 more candidates considered.