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

Source study found

Story checked

AI model detects signs of faulty DNA repair in ovarian cancer using tissue images (opens in a new tab)

medicalxpress.com · 2026-10-21

Short answerEvidenceSource

Short answer

Mostly not supported

Mostly not supported.

One claim goes further than the study. 4 other points were not covered by the paper.

  • 2 supported
  • 1 overstated
  • 4 not covered

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

Share this check

Follow the evidence trail
1
2

NewsLink checks it

Mostly not supported

One claim overstates the study. Two of seven check out. Four claims the study doesn't address.

  • 2 supported
  • 1 overstated
  • 4 not covered
Open claim evidence
3

The source study

HRDPath: an explainable multi-model deep learning architecture for predicting homologous recombination deficiency from histopathology images

Nature Communications · 2026
Open source record (opens in a new tab)
Then inspect each claim

Evidence layer

Claim by claim

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

7 claims in this story

Showing all 7 claimsChoose a verdict to focus the list.

Claim 1 of 7OverstatedThe model could accurately recognize tumors that contained broken DNA repair mechanisms, and this could enable clinicians to choose treatments that exploit this flaw.View evidenceHide evidence

Why this verdict

The profile supports predictive performance for HRD status from images, but the story frames this as enabling clinicians to choose treatments. At abstract depth, the paper supports treatment-stratification potential, not demonstrated clinical treatment selection or improved treatment outcomes, so the causal/clinical-utility framing is stronger than the evidence.

1

Study evidence

HRDPath, an explainable multi-model deep learning architecture, was developed to predict HRD status from whole-slide histopathology images at the patient level without additional molecular testing.

“We develop HRDPath, a patient-level deep learning architecture that predicts HRD status from whole slide images”
2

Study evidence

HRDPath validated across three cohorts and achieves an AUC of 0.846 for predicting HRD status from whole-slide images.AUC = 0.846

“HRDPath is validated across three cohorts of female patients and achieves an AUC of 0.846”
Claim 2 of 7Not coveredThe study focused on high-grade serous tubo-ovarian carcinoma (HGSOC), where homologous recombination deficiency (HRD) is an important biomarker because it can arise from BRCA1/BRCA2-related defects and help guide treatment choices such as platinum-based chemotherapy and PARP inhibitors.View evidenceHide evidence

Why this verdict

The profile supports that the study concerns HGSOC and that HRD is an important biomarker for treatment decisions. However, the abstract-level profile does not provide the specific BRCA1/BRCA2 background or name platinum chemotherapy and PARP inhibitors, so those details are not verifiable at this evidence depth.

1

Study evidence

HRDPath, an explainable multi-model deep learning architecture, was developed to predict HRD status from whole-slide histopathology images at the patient level without additional molecular testing.

“We develop HRDPath, a patient-level deep learning architecture that predicts HRD status from whole slide images”
2

Study evidence

HRDPath validated across three cohorts and achieves an AUC of 0.846 for predicting HRD status from whole-slide images.AUC = 0.846

“HRDPath is validated across three cohorts of female patients and achieves an AUC of 0.846”
Claim 3 of 7Not coveredHRDPath was trained on high-resolution histopathology images from patients whose HRD status was already known through genomic analysis, rather than looking at DNA directly.View evidenceHide evidence

Why this verdict

The profile supports use of whole-slide histopathology images and prediction without additional molecular testing. But the abstract-level profile explicitly does not specify the HRD reference standard, so the claim that known HRD status came through genomic analysis is not verifiable at this depth.

1

Study evidence

HRDPath, an explainable multi-model deep learning architecture, was developed to predict HRD status from whole-slide histopathology images at the patient level without additional molecular testing.

“We develop HRDPath, a patient-level deep learning architecture that predicts HRD status from whole slide images”
2

Study evidence

HRDPath validated across three cohorts and achieves an AUC of 0.846 for predicting HRD status from whole-slide images.AUC = 0.846

“HRDPath is validated across three cohorts of female patients and achieves an AUC of 0.846”
Claim 4 of 7Not coveredThe researchers identified subtle tissue-appearance differences associated with HRD, including rounder cell nuclei and more densely packed cancer cells in some tumor regions.View evidenceHide evidence

Why this verdict

The profile supports that interpretability analyses identified cellular- and tissue-level morphological/pathological changes associated with HRD-positive cancers. It does not, at abstract depth, list concrete features such as rounder nuclei or densely packed cancer cells, so those examples are not verifiable here.

1

Study evidence

The interpretability pipeline reveals morphological and pathological features associated with HRD-positive HGSOC and is reported to enhance model trustworthiness.

“Our interpretability pipeline enhances model trustworthiness and reveals important clinical and biological insights”
Claim 5 of 7Not coveredThe article says further validation in larger patient populations is needed before the approach can be reliably used in practice.View evidenceHide evidence

Why this verdict

The caution is broadly consistent with the profile’s abstract-level limitations, such as missing cohort sizes, uncertainty estimates, threshold-selection details, and validation specifics. However, the abstract-level profile does not directly state that larger patient-population validation is required before practice use, so the claim is not verifiable at this depth.

1

Study evidence

HRDPath validated across three cohorts and achieves an AUC of 0.846 for predicting HRD status from whole-slide images.AUC = 0.846

“HRDPath is validated across three cohorts of female patients and achieves an AUC of 0.846”
Claim 6 of 7SupportedAn AI model created by researchers at King's can detect signs of faulty DNA repair in ovarian cancer cells by analyzing tissue images, potentially leading to a low-cost, widely accessible diagnostic tool in the future.View evidenceHide evidence

Why this verdict

The abstract-level profile supports that HRDPath predicts HRD status from HGSOC whole-slide histopathology images without additional molecular testing, and frames it as an accessible/scalable future digital biomarker. The headline is hedged with “potentially” and does not materially outrun the paper profile.

1

Study evidence

HRDPath, an explainable multi-model deep learning architecture, was developed to predict HRD status from whole-slide histopathology images at the patient level without additional molecular testing.

“We develop HRDPath, a patient-level deep learning architecture that predicts HRD status from whole slide images”
2

Study evidence

HRDPath validated across three cohorts and achieves an AUC of 0.846 for predicting HRD status from whole-slide images.AUC = 0.846

“HRDPath is validated across three cohorts of female patients and achieves an AUC of 0.846”
Claim 7 of 7SupportedThe model correctly identified approximately 91% of HRD-positive cases when several images from the same patient were considered.View evidenceHide evidence

As statedapproximately 91%

Why this verdict

The profile reports patient-level prediction using multiple slides per patient and sensitivity reaching about 0.90. The story’s approximately 91% correct identification of HRD-positive cases is consistent with that sensitivity-level performance, though the profile’s abstract-level figure is closer to 90%.

1

Study evidence

HRDPath, an explainable multi-model deep learning architecture, was developed to predict HRD status from whole-slide histopathology images at the patient level without additional molecular testing.

“We develop HRDPath, a patient-level deep learning architecture that predicts HRD status from whole slide images”
2

Study evidence

HRDPath validated across three cohorts and achieves an AUC of 0.846 for predicting HRD status from whole-slide images.AUC = 0.846

“HRDPath is validated across three cohorts of female patients and achieves an AUC of 0.846”
Then look for missing context

Context layer

What the story left out

Important study details the story did not include.

  • The paper validates HRDPath across three cohorts and reports AUC = 0.846, with sensitivity and specificity reaching about 0.90.

    The story reflects the sensitivity-like result for HRD-positive cases, but it does not report the AUC, specificity, or the three-cohort validation framing. The complete performance context is therefore not fully reflected.

    From Retrospective multi-cohort model validation using secondary histopathology data

4 things the story did carry across
  • HRDPath is a patient-level deep learning model developed to predict HRD status from HGSOC whole-slide histopathology images without additional molecular testing.
  • The model incorporates multiple slides per patient for patient-level aggregation.
  • HRD is presented as an important biomarker for treatment decisions in HGSOC, motivating an image-based digital biomarker because genomics-based HRD testing can be costly and time-consuming.
  • The interpretability pipeline identifies morphological and pathological features associated with HRD-positive cancers at cellular and tissue levels.
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

3

Evidence read

study summary

Lead result

in silico

1Lead resultin silicoDevelop HRDPath, a patient-level multi-model deep learning architecture that predicts homologous recombination deficiency (HRD) status from HGSOC whole-slide histopathology images without additional molecular testing.patient-level multi-model deep learning for WSI-based HRD classificationExpand

In plain English

The authors develop HRDPath, an explainable patient-level multi-model deep learning system that predicts homologous recombination deficiency (HRD) status directly from whole-slide histopathology images of high-grade serous tubo-ovarian carcinoma (HGSOC) without requiring additional molecular testing. In abstract-level reporting, HRDPath is validated across three patient cohorts and reported to achieve an AUC of 0.846 and both sensitivity and specificity reaching 0.90; the approach incorporates aggregation across multiple slides per patient, targeted optimization strategies, and an interpretability pipeline that highlights associated cellular and tissue-level morphological changes in HRD-positive cancers.

Key findings

  • HRDPath, an explainable multi-model deep learning architecture, was developed to predict HRD status from whole-slide histopathology images at the patient level without additional molecular testing.
  • On abstract-level validation across three cohorts, HRDPath achieved an area under the ROC curve (AUC) of 0.846 for predicting HRD.AUC 0.846
“We develop HRDPath, a patient-level deep learning architecture that predicts HRD status from whole slide images”
What this piece can’t prove
  • Summary is based solely on the abstract; the abstract does not report dataset sizes, cohort characteristics, or inclusion/exclusion criteria.

4 further details could not be confirmed from the summary.

2secondary dataValidate HRDPath performance and generalizability across three independent cohorts of female HGSOC patients, reporting discrimination (AUC) and operating-point metrics (sensitivity/specificity).Retrospective multi-cohort model validation using secondary histopathology dataExpand

In plain English

Validation of HRDPath, a patient-level deep learning model that predicts homologous recombination deficiency (HRD) from whole-slide histopathology images, across three independent cohorts of female high-grade serous tubo-ovarian carcinoma (HGSOC) patients. Reported discrimination: AUC = 0.846. Reported operating-point performance: sensitivity and specificity reaching ~0.90. Validation emphasizes multi-slide per patient input and targeted optimization to strengthen clinical utility.

Key findings

  • HRDPath validated across three cohorts and achieves an AUC of 0.846 for predicting HRD status from whole-slide images.AUC = 0.846
  • Model operating-point performance reported with both sensitivity and specificity reaching approximately 0.90.sensitivity ≈ 0.90; specificity ≈ 0.90
“HRDPath is validated across three cohorts of female patients and achieves an AUC of 0.846”
What this piece can’t prove

3 further details could not be confirmed from the summary.

3in silicoProvide model explainability/interpretability analyses that identify morphological and pathological features associated with HRD-positive cancers to enhance trust and yield clinicopathologic insight.Expand

In plain English

The paper presents an interpretability/explainability pipeline applied to the HRDPath whole-slide-image classifier that (according to the abstract) identifies image regions and morphologies driving HRD predictions, yielding cellular- and tissue-level morphological and pathological insights and enhancing model trustworthiness.

Key findings

  • The interpretability pipeline reveals morphological and pathological features associated with HRD-positive HGSOC and is reported to enhance model trustworthiness.
“Our interpretability pipeline enhances model trustworthiness and reveals important clinical and biological insights”
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.

Open the paper in Tessa

HRDPath: an explainable multi-model deep learning architecture for predicting homologous recombination deficiency from histopathology images

Nature Communications · 2026

Source record (opens in a new tab)

Why this one

Near certain

NewsLink found the paper. Tessa is where you inspect it deeply.

Papers considered

The selected paper, plus nearby candidates.

Crossref, Europe PMC, PubMed · 41 candidate papers

Selected

HRDPath: an explainable multi-model deep learning architecture for predicting homologous recombination deficiency from histopathology images

Nature Communications · 2026 · Crossref

Source (opens in a new tab)

And 35 more candidates considered.