Source check: Source checked. Not independent scientific confirmation.
Finding: Using MIMIC-IV for development and two external ICU databases, XGBoost predicted sepsis-associated acute kidney injury with AUC 0.72 internally and 0.82/0.84 externally; differences across datasets and the retrospective design limit clinical interpretation.
Technology: Six machine-learning algorithms were optimized with grid search and five-fold cross-validation; XGBoost was selected and interpreted with SHAP.
Publication: Peer-reviewed paper · Test context: retrospective or benchmark. Catalog categories, not an evidence grade.
- AI contribution
- Six machine-learning algorithms were optimized with grid search and five-fold cross-validation; XGBoost was selected and interpreted with SHAP.
- Test and result
- The retrospective model used MIMIC-IV for development and eICU-CRD and BB-ICU for external assessment. XGBoost AUC was 0.72 internally and 0.82 and 0.84 externally; calibration and decision-curve analyses were also reported.
- Limitation / still unknown
- Performance and AKI incidence differed across data sources. Retrospective prediction does not establish causal benefit, prospective performance or clinical utility of the web calculator.
Evidence sources and annotation context
Evidence context (descriptions follow the cited sources; test type is a catalog classification):
Evidence sources: https://link.springer.com/article/10.1186/s12911-026-03620-z
Source checked · editorial details
The source page was checked against this entry. Independent confirmation is not assessed by this label.
Source type: paper. Paper: not linked in this catalog; this does not mean no paper exists.
Review date: 22 Nov 2026. Review scope: Primary result-bearing material for the stated method, result and evaluation, with bibliographic version information..
No follow-up recorded; this is not a claim that no follow-up exists. Follow-up review scope and date are not recorded. No source-supported relationships recorded.
Editorial assessment: Impact 2/5 · Validation 2/5 · Novelty 3/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.
Source check: Source checked. Not independent scientific confirmation.
Finding: LASSO-based selection and network prioritization nominated five periodontitis-associated candidates. A source-reported macrophage comparison found ATF3 and CXCR4 higher, and TXN, CBX3 and MBP lower, at RNA and protein levels. This record represents those expression observations, without adopting a clinical diagnostic test or a causal disease mechanism.
Technology: LASSO feature selection
Publication: Peer-reviewed paper · Test context: direct test. Catalog categories, not an evidence grade.
- AI contribution
- LASSO selects22 expression candidates, followed by protein-network prioritization of five that undergo a new expression check. AI credit is limited to nomination, not downstream causal interpretation.
- Test and result
- Actual computational selection sections and nonactionable expression Results were read. The source reports concordant RNA/protein directions in an inflammatory macrophage comparison.
- Limitation / still unknown
- Operational culture/assay protocols and supplementary replicate details were not inspected. Archived discovery and validation arrays contain repeated gingival sites per patient, with healthy sites within periodontitis patients and possible cohort overlap; another series includes peri-implantitis and a further RNA-sequencing series includes gingivitis. Clinical discrimination, independent patient validation, novel macrophage taxonomy and therapeutic/causal claims are outside the represented result. The reported model C-index is not adopted as an AUC.
Evidence sources and annotation context
Evidence context (descriptions follow the cited sources; test type is a catalog classification):
Evidence sources: https://pmc.ncbi.nlm.nih.gov/articles/PMC12615752/ · https://doi.org/10.1016/j.identj.2025.103983
Source checked · editorial details
The source page was checked against this entry. Independent confirmation is not assessed by this label.
Source type: paper. Paper: not linked in this catalog; this does not mean no paper exists.
Review date: 21 Nov 2026. Review scope: Actual relevant primary computational/clinical Methods and high-level Results read; no operational biological protocols inspected.
No follow-up recorded; this is not a claim that no follow-up exists. Follow-up review scope and date are not recorded. No source-supported relationships recorded.
Editorial assessment: Impact 1/5 · Validation 3/5 · Novelty 2/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.
Source check: Source checked. Not independent scientific confirmation.
Finding: LASSO selected 19 expression candidates in archived stroke samples. Four candidates—SCO2, ARF1, APP and NFE2L2—were reported as elevated in both additional datasets, while the full sets of significant genes differed. This is a retrospective expression association, not validation of a 19-gene diagnostic test or a cell-death mechanism.
Technology: LASSO
Publication: Peer-reviewed paper · Test context: retrospective or benchmark. Catalog categories, not an evidence grade.
- AI contribution
- A fitted LASSO classifier selected 19 genes from 30 prefiltered candidates, determining the panel examined in the additional datasets.
- Test and result
- The discovery dataset contains 69 stroke specimens sampled at three post-event times plus 23 control specimens. Additional datasets included 39 stroke/24 control records and 10 stroke/5 control specimens; the latter also has repeated time points.
- Limitation / still unknown
- Repeated specimens are not independent patients. The discovery AUC of 1 is an apparent fit, not external diagnostic performance; additional analyses compare individual gene expression and ROC curves rather than a frozen combined panel. Screening, time-point and batch effects limit inference. The genes came from a previously defined pathway list, and no new causal role, biological subtype or clinical utility is established.
Evidence sources and annotation context
Evidence context (descriptions follow the cited sources; test type is a catalog classification):
Evidence sources: https://pmc.ncbi.nlm.nih.gov/articles/PMC11408831/
Source checked · editorial details
The source page was checked against this entry. Independent confirmation is not assessed by this label.
Source type: paper. Paper: not linked in this catalog; this does not mean no paper exists.
Review date: 21 Nov 2026. Review scope: Primary data/LASSO Methods and external expression Results; primary GEO study metadata; no operational laboratory procedures reviewed.
No follow-up recorded; this is not a claim that no follow-up exists. Follow-up review scope and date are not recorded. No source-supported relationships recorded.
Editorial assessment: Impact 1/5 · Validation 2/5 · Novelty 1/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.
Source check: Source checked. Not independent scientific confirmation.
Finding: A retrospective gastric-cancer analysis combined CT features and learned pathology-slide features to classify stage groups. The combined model reached test AUC .837 in 60 patients, but its test advantage over the component models was not statistically established. Surgical slides limit claims of preoperative use.
Technology: ResNet-18 features, LASSO, SVM and multilayer perceptron
Publication: Peer-reviewed paper · Test context: retrospective or benchmark. Catalog categories, not an evidence grade.
- AI contribution
- Learned classifiers and representations evaluate a defined clinical staging or recurrence target; conventional clinicopathologic associations are not separately counted as AI discoveries.
- Test and result
- Two hundred surgically resected cases were split 140/60 by patient. ResNet-18 learned tumor-region features from pathology patches; selected pathology and CT signatures were combined. Component test AUCs were .797 for CT and .792 for pathology; combined AUC was .837. Original Results explicitly state no significant test superiority over either component despite a training difference.
- Limitation / still unknown
- This is one retrospective institution and a small random internal test. Methods do not clearly establish training-only fitting of every normalization, correlation filter and LASSO step. The best model families/signatures were selected after comparisons; untouched external confirmation is absent. Pathological slides came from surgical cases, with no demonstrated preoperative-biopsy deployment. Nonsignificant differences do not establish equivalence, and a nonsignificant goodness-of-fit test does not prove perfect calibration. Decision curves do not measure clinical benefit.
Evidence sources and annotation context
Evidence context (descriptions follow the cited sources; test type is a catalog classification):
Evidence sources: https://pmc.ncbi.nlm.nih.gov/articles/PMC10788295/
Source checked · editorial details
The source page was checked against this entry. Independent confirmation is not assessed by this label.
Source type: paper. Paper: linked separately.
Review date: 21 Nov 2026. Review scope: Original patient cohort, computational Methods and represented Results actually read.
No follow-up recorded; this is not a claim that no follow-up exists. Follow-up review scope and date are not recorded. No source-supported relationships recorded.
Editorial assessment: Impact 1/5 · Validation 2/5 · Novelty 2/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.