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

Explainable AI predicts how plant compounds may cross intestinal barriers (opens in a new tab)

medicalxpress.com · 2026-11-06

Short answerEvidenceSource

Short answer

Mostly supported

Mostly supported.

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

  • 4 supported
  • 1 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

Mostly supported

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

  • 4 supported
  • 1 not covered
Open claim evidence
3
Source paper

Source layer

The 2 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.

5 claims in this story

Showing all 5 claimsChoose a verdict to focus the list.

Claim 1 of 5Not coveredUsing experimentally measured values for 83 phytochemicals and 5,003 molecular descriptors, the researchers compared 10 regression algorithms and found the best performance in a stacking ensemble combining CatBoost, LightGBM, Gradient Boosting, and linear regression.View evidenceHide evidence

As statedup to about 0.96 for apparent permeability; AUC 0.96 for efflux classification

Why this verdict

The abstract-level profile supports the dataset size, 5,003 descriptors, comparison of 10 regression algorithms, and the stacking ensemble architecture using CatBoost, LightGBM, Gradient Boosting, and linear regression. However, the stated numerical performance values, including 'up to about 0.96' for Papp and 'AUC 0.96 for efflux classification,' are not reported in the abstract-level profile, which specifically notes that quantitative metrics and detailed validation procedures are not available at this depth.

1

Study evidence

Pearson correlation analysis revealed significant relationships among the three permeability-related indicators (Papp, TEER, ER).

“develop and validate quantitative structure-property relationship (QSPR) models for predicting permeability related bioavailability indicators including apparent permeability (Papp), trans-epithelial electrical resistance (TEER) and efflux ratio (ER) based on molecular descriptors (n = 5003) of 83 phytochemicals.”
Claim 2 of 5SupportedResearchers at Sejong University and the Korea Institute of Science and Technology developed explainable machine learning models that predict key intestinal absorption properties of phytochemicals directly from molecular structure before laboratory experiments are done.View evidenceHide evidence

Why this verdict

The abstract-level profile supports the core claim that the paper developed and validated QSPR/machine-learning models to predict Caco-2 permeability-related indicators for phytochemicals from molecular descriptors, with SHAP used for explainability. The 'before laboratory experiments' framing is best read as an in-silico prescreening implication; the abstract does not show prospective replacement of experiments, but the story caveats that the models are prescreening tools.

1

Study evidence

Pearson correlation analysis revealed significant relationships among the three permeability-related indicators (Papp, TEER, ER).

“develop and validate quantitative structure-property relationship (QSPR) models for predicting permeability related bioavailability indicators including apparent permeability (Papp), trans-epithelial electrical resistance (TEER) and efflux ratio (ER) based on molecular descriptors (n = 5003) of 83 phytochemicals.”
2

Study evidence

SHAP-based feature importance analysis identified molecular descriptor classes related to electronic, topological and branching properties as key contributors to model predictions for Papp, TEER and ER.

“SHapley Additive exPlanations (i.e. SHAP)-based feature importance analysis provided insight into key molecular descriptors associated with electronic, topological and branching.”
Claim 3 of 5SupportedThe study focused on three Caco-2 cell model indicators: apparent permeability (Papp), transepithelial electrical resistance (TEER), and efflux ratio (ER).View evidenceHide evidence

Why this verdict

The profile explicitly identifies apparent permeability (Papp), transepithelial electrical resistance (TEER), and efflux ratio (ER) as the three Caco-2 permeability-related indicators modeled in the study.

1

Study evidence

Pearson correlation analysis revealed significant relationships among the three permeability-related indicators (Papp, TEER, ER).

“develop and validate quantitative structure-property relationship (QSPR) models for predicting permeability related bioavailability indicators including apparent permeability (Papp), trans-epithelial electrical resistance (TEER) and efflux ratio (ER) based on molecular descriptors (n = 5003) of 83 phytochemicals.”
Claim 4 of 5SupportedThe article says TEER was harder to predict than the other endpoints, with moderate but more variable performance.View evidenceHide evidence

Why this verdict

The profile states that the stacking ensemble achieved strong predictive accuracy for Papp and ER but only moderate and highly variable performance for TEER, matching the story's claim that TEER was harder to predict and less robust.

1

Study evidence

Pearson correlation analysis revealed significant relationships among the three permeability-related indicators (Papp, TEER, ER).

“develop and validate quantitative structure-property relationship (QSPR) models for predicting permeability related bioavailability indicators including apparent permeability (Papp), trans-epithelial electrical resistance (TEER) and efflux ratio (ER) based on molecular descriptors (n = 5003) of 83 phytochemicals.”
Claim 5 of 5SupportedUsing SHAP, the authors identified electronic, topological, and branching-related descriptors as key determinants of the predictions, which the article presents as a way to explain why compounds are predicted to be better or worse absorbed.View evidenceHide evidence

Why this verdict

The profile supports that SHAP-based feature-importance analysis identified descriptor classes related to electronic, topological, and branching properties as influential for model predictions. Because the story frames these as determinants of the predictions rather than proven biological causes of absorption, the claim is supported at abstract depth.

1

Study evidence

SHAP-based feature importance analysis identified molecular descriptor classes related to electronic, topological and branching properties as key contributors to model predictions for Papp, TEER and ER.

“SHapley Additive exPlanations (i.e. SHAP)-based feature importance analysis provided insight into key molecular descriptors associated with electronic, topological and branching.”
Then look for missing context

Context layer

What the story left out

Important study details the story did not include.

  • Supporting analysis: Pearson correlation analysis found significant relationships among Papp, TEER, and ER in the dataset.

    The profile treats this as a distinct supporting analysis, but the story presentation does not mention the Pearson correlation analysis among the three indicators.

    From Exploratory Pearson correlation analysis

7 things the story did carry across
  • Primary contribution: development and validation of QSPR/machine-learning models to predict Caco-2 permeability-related indicators from molecular descriptors for phytochemicals.
  • Modeled endpoints: apparent permeability (Papp), transepithelial electrical resistance (TEER), and efflux ratio (ER).
  • Dataset and modeling workflow: 83 phytochemicals, 5,003 molecular descriptors, feature selection, and comparison of 10 regression algorithms.
  • Best-performing model architecture: stacking ensemble with CatBoost, LightGBM, and Gradient Boosting base learners and linear regression meta-model.
  • Performance pattern: strong predictive accuracy for Papp and ER, but moderate and highly variable performance for TEER.
  • SHAP interpretability analysis identifying influential descriptors associated with electronic, topological, and branching properties.
  • Limitation: dataset size was small, with 83 phytochemicals.
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 and validate QSPR (machine-learning) regression models to predict Caco-2 permeability-related indicators (Papp, TEER, efflux ratio) for phytochemicals from molecular descriptors.QSPR regression modeling (in silico)Expand

In plain English

The study developed and validated QSPR machine-learning regression models to predict Caco-2 permeability-related indicators (apparent permeability Papp, trans-epithelial electrical resistance TEER, and efflux ratio ER) from molecular descriptors for 83 phytochemicals (5,003 descriptors). A workflow combining feature selection, evaluation of 10 regression algorithms, and a stacking ensemble (CatBoost, LightGBM, Gradient Boosting base learners; linear regression meta-model) produced strong predictive accuracy for Papp and ER, and moderate but highly variable performance for TEER. SHAP was used to identify key molecular descriptors related to electronic, topological and branching properties.

Key findings

  • Pearson correlation analysis revealed significant relationships among the three permeability-related indicators (Papp, TEER, ER).
  • A stacking ensemble (CatBoost, LightGBM, Gradient Boosting base learners; linear regression meta-model) achieved strong predictive accuracy for Papp and ER, while TEER predictions were moderate and highly variable.
“develop and validate quantitative structure-property relationship (QSPR) models for predicting permeability related bioavailability indicators including apparent permeability (Papp), trans-epithelial electrical resistance (TEER) and efflux ratio (ER) based on molecular descriptors (n = 5003) of 83 phytochemicals.”
What this piece can’t prove
  • Study dataset size is 83 phytochemicals (as reported in abstract); abstract does not provide information on external validation or applicability domain.
  • TEER predictions were reported as moderate and highly variable, indicating weaker/less consistent model performance for that outcome.

1 further detail could not be confirmed from the summary.

2in silicoCharacterize relationships among the three bioavailability indicators (Papp, TEER, ER) within the compiled phytochemical dataset.Exploratory Pearson correlation analysisExpand

In plain English

The authors performed Pearson correlation coefficient analysis on apparent permeability (Papp), trans-epithelial electrical resistance (TEER), and efflux ratio (ER) across a compiled dataset of 83 phytochemicals and report significant relationships among the three indicators.

Key findings

  • Pearson correlation coefficient analysis indicated significant relationships among apparent permeability (Papp), trans-epithelial electrical resistance (TEER), and efflux ratio (ER) in the 83-phytochemical dataset.
“Pearson correlation coefficient analysis revealed significant relationships among the three indictors.”
What this piece can’t prove
  • Unclear whether assumptions of Pearson correlation (linearity, normality) were assessed or met.
  • No information on correction for multiple testing or sensitivity analyses.

2 further details could not be confirmed from the summary.

3in silicoProvide model interpretability/insight by identifying influential molecular descriptors using SHAP feature-importance analyses.in silico interpretability (SHAP) analysisExpand

In plain English

The study applied SHapley Additive exPlanations (SHAP) to QSPR models to identify molecular descriptors that most influence predicted apparent permeability (Papp), trans-epithelial electrical resistance (TEER) and efflux ratio (ER) for a set of 83 phytochemicals. SHAP-based feature importance highlighted descriptors related to electronic, topological and branching properties and was reported as providing interpretability that supports construction of robust, explainable QSPR models.

Key findings

  • SHAP-based feature importance analysis identified molecular descriptor classes related to electronic, topological and branching properties as key contributors to model predictions for Papp, TEER and ER.
  • The authors state that SHAP-based interpretability provided actionable insight and supports constructing robust, explainable QSPR models for bioavailability prediction in phytochemicals.
“SHapley Additive exPlanations (i.e. SHAP)-based feature importance analysis provided insight into key molecular descriptors associated with electronic, topological and branching.”
What this piece can’t prove
  • Abstract does not list the individual molecular descriptors identified or quantitative SHAP values.
  • Interpretability findings are contextualized by model predictive performance; TEER models had moderate and highly variable accuracy, which may limit confidence in SHAP attributions for TEER.

1 further detail 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 · 38 candidate papers

SelectedOpen access

Molecular descriptor driven QSPR modeling of Papp, TEER and Efflux Ratio from Caco-2 cells using machine learning for various phytochemicals.

Journal of the Science of Food and Agriculture · 2026 · PubMed, Europe PMC, Crossref

Tessa (opens in a new tab)Source (opens in a new tab)
Candidate

Comparison of deep learning models for 3D food printing monitoring system

Journal of Food Engineering · 2026 · Crossref

Source (opens in a new tab)
Candidate

Enhanced rice bran extracts attenuate STZ-induced β-cell injury and diabetes-related metabolic disturbances

Journal of Agriculture and Food Research · 2026 · Crossref

Source (opens in a new tab)

And 32 more candidates considered.