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

Source study found

Story checked

The subtleties a spectrograph would miss – For Better Science (opens in a new tab)

forbetterscience.com · 2026-09-08

Short answerEvidenceSource

Short answer

Not supported

Not supported.

The study does not answer the story's main claims.

  • 7 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

Not supported

The study doesn't address any of the story's claims. We found the paper, but it doesn't report the details the story leads with.

  • 7 not covered
Open claim evidence
3
Source paper

Source layer

The 23 papers the story cites

Source study separated from background citations.

The research anchor for the report.

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 7Not coveredA series of machine-learning papers are presented as being built on nonsense, stolen, garbled, or fabricated datasets, with figures and results that the author says are meaningless.View evidenceHide evidence

Why this verdict

The supplied paper profile is only an abstract-level profile of the breast-cancer BCDNN paper, not the full set of papers criticized in the story. It does not provide evidence that datasets were stolen, garbled, fabricated, or that figures/results were meaningless. For the breast-cancer paper, the abstract only states that a Kaggle dataset with genomic and pathological features was used and reports cross-validation accuracy; the profile's limitations note missing dataset provenance details, but that is not enough to verify the broader allegation.

1

Study evidence

The BCDNN achieved a mean classification accuracy of 93.84% for benign versus malignant tumor classification during cross-validation on the Kaggle dataset.93.84% mean accuracy (cross-validation)

“develop and evaluate a BCDNN model that classifies tumors as benign or malignant using genomic and histopathological data.”
Claim 2 of 7Not coveredThe article says one Scientific Reports paper on smart EV charging used an AI-integrated blockchain framework trained on a barebones 2014-2015 EV-charging dataset, despite there being no blockchain component in that archive and questionable data provenance from Kaggle.View evidenceHide evidence

As stated2014-2015

Why this verdict

The supplied paper profile contains no abstract-level evidence about a Scientific Reports smart EV charging paper, blockchain, or a 2014–2015 EV-charging dataset. This claim cannot be checked against the provided breast-cancer paper profile.

Claim 3 of 7Not coveredThe article says a PLOS One paper on logistics and route optimization relied on a Kaggle dataset that was allegedly random numbers and teleported drivers and vehicles across implausible locations.View evidenceHide evidence

As stated12000 snapshots

Why this verdict

The supplied paper profile contains no evidence about a PLOS One logistics or route-optimization paper, a 12,000-snapshot Kaggle dataset, or alleged implausible vehicle movements. This claim is outside the provided profile.

Claim 4 of 7Not coveredThe article says a PLOS One paper on student mental health claimed to use 2,000 student records, but the spreadsheet was described as a synthetic simulation with 10,000 rows and only a subset labeled as students.View evidenceHide evidence

As stated2,000 student records

Why this verdict

The supplied paper profile contains no evidence about a PLOS One student mental-health paper, 2,000 student records, or a synthetic spreadsheet with 10,000 rows. This claim cannot be reconciled with the provided breast-cancer paper abstract profile.

Claim 5 of 7Not coveredThe article says epilepsy-seizure classification papers repeatedly reuse or distort the Bonn EEG dataset and related derivatives, while claiming very high accuracy despite severe provenance and methodology problems.View evidenceHide evidence

As stated100% accuracy

Why this verdict

The supplied paper profile contains no evidence about epilepsy EEG seizure-classification papers, the Bonn EEG dataset, reused derivatives, or claimed 100% accuracy. The claim is not checkable using this profile.

Claim 6 of 7Not coveredThe article says a Parkinson’s disease detection paper combined an Oxford PD voice dataset with the unrelated DAIC-WOZ corpus, which the author argues is not a sensible pairing for the stated task.View evidenceHide evidence

As statedTwo publicly available datasets

Why this verdict

The supplied paper profile contains no evidence about a Parkinson’s disease detection paper, the Oxford PD voice dataset, or DAIC-WOZ. This claim is outside the provided paper profile.

Claim 7 of 7Not coveredThe article says a breast-cancer paper claimed to fuse genomic and histopathological data, but the Kaggle dataset it used was actually the Wisconsin Breast Cancer Dataset and did not contain genomic data.View evidenceHide evidence

Why this verdict

The abstract-level profile supports that the breast-cancer paper claimed to use a publicly available Kaggle breast-cancer dataset encompassing genomic and pathological features and to fuse genomic and histopathological/pathological data. However, the abstract profile does not identify the exact Kaggle dataset or verify whether it was actually the Wisconsin Breast Cancer Dataset lacking genomic data. The story’s stronger dataset-mismatch allegation is therefore not verifiable at abstract depth.

1

Study evidence

The BCDNN achieved a mean classification accuracy of 93.84% for benign versus malignant tumor classification during cross-validation on the Kaggle dataset.93.84% mean accuracy (cross-validation)

“develop and evaluate a BCDNN model that classifies tumors as benign or malignant using genomic and histopathological data.”
Then look for missing context

Context layer

What the story left out

Important study details the story did not include.

  • The paper reports mean cross-validation accuracy of 93.84% for benign-vs-malignant classification.

    The story presentation does not mention the paper’s reported 93.84% cross-validation accuracy for the breast-cancer model.

    From secondary_data: ML model development (BCDNN)

  • The abstract describes preprocessing, feature selection, optimized hyperparameters, and MATLAB R2016 implementation before cross-validation.

    These methodological details from the paper profile are not reflected in the story presentation.

    From secondary_data: ML model development (BCDNN)

  • The paper states that code for the BCDNN study is available on GitHub.

    The story presentation does not mention the paper’s code-availability claim or the implementation artifact.

    From software/code release (GitHub)

2 things the story did carry across
  • The paper’s primary contribution is a BCDNN model for benign-vs-malignant breast tumor classification using fused genomic and histopathological/pathological features.
  • The abstract says the model used a publicly available Kaggle breast-cancer dataset described as containing genomic and pathological features.
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 dataDevelop and evaluate a deep-learning model (BCDNN) that classifies breast tumors as benign vs malignant by fusing genomic and histopathological (pathological) features.secondary data: ML model development (BCDNN)Expand

In plain English

Developed and evaluated a BCDNN that fuses genomic and histopathological features to classify breast tumors as benign versus malignant using a publicly available Kaggle breast cancer dataset. The dataset was pre-processed and feature-selected; the model was implemented in MATLAB R2016 with optimized hyperparameters and evaluated using cross-validation, yielding a reported mean accuracy of 93.84%. Code availability on GitHub is noted.

Key findings

  • The BCDNN achieved a mean classification accuracy of 93.84% for benign versus malignant tumor classification during cross-validation on the Kaggle dataset.93.84% mean accuracy (cross-validation)
“develop and evaluate a BCDNN model that classifies tumors as benign or malignant using genomic and histopathological data.”
What this piece can’t prove
  • No external or independent validation set is described; results are reported from cross-validation only.

3 further details could not be confirmed from the summary.

2otherProvide an implementation artifact (code) for the proposed BCDNN approach.software/code release (GitHub)Expand

In plain English

The paper's abstract states that the code implementing the proposed BCDNN model is publicly available on GitHub; the implementation is described as MATLAB R2016-based. The abstract provides an availability statement but does not include repository URL or metadata.

Key findings

  • The authors state that the study's code for the BCDNN approach is available on GitHub.
“The code for this study is available on GitHub .”
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.

Crossref, Europe PMC · 35 candidate papers

And 29 more candidates considered.