Analysis & ranking
PHASE 2 — Evidence and Impact Analysis
Article 1 — FIBOM-AI (Donzel et al.) | PMID 42242261
Prediction of bone marrow fibrosis from CBC in MPN
| Dimension | Score | Rationale |
|---|---|---|
| Scientific Novelty | 8 | First validated ML model predicting BM fibrosis grade from CBC alone across 18 international centres; XGBoost with 27 CBC parameters is technically well-executed |
| Clinical Relevance | 9 | BM biopsy is invasive, painful, and rate-limiting in MPN management; a rule-out tool with 98.6% prospective sensitivity directly changes clinical workflow |
| Population Reach | 7 | MPN prevalence ~3/10,000; significant within haematology but not a mass-population finding; unmet need is real and well-defined |
| Implementation Speed | 8 | CBC is universally available; model is XGBoost (non-proprietary); prospective validation already done across Canada + Europe; regulatory/EHR integration is the main remaining barrier |
| Evidence Strength | 9 | Multicentre (18 centres), retrospective development + prospective validation (n=493), peer-reviewed in Lancet Haematol; rigorous design with two-mode prediction |
Key quantitative result: AUC 0.92 (external validation); prospective rule-out sensitivity 98.6%; overall prospective accuracy 85.2%
External validation: Yes — 7-centre external retrospective + 5-centre prospective including Canadian site
Main limitation: Abstract-only access; full feature importance, model calibration curves, and failure-mode analysis not reviewable; single disease context (MPN only)
Equity implications: CBC is accessible globally; if deployed as open-source or embedded in existing LIS, this could benefit MPN patients in resource-limited settings who currently lack access to haematopathology for biopsy. Risk: if commercialised as proprietary software, could create access disparity.
Evidence Maturity: ✅ Validated (confirmed — multicentre prospective validation complete)
Article 2 — Intratumoral bicarbonate + PD-1 blockade in HCC (Wang et al.) | PMID 42243329
Tislelizumab + NaHCO₃ in hepatocellular carcinoma
| Dimension | Score | Rationale |
|---|---|---|
| Scientific Novelty | 9 | Intratumoral bicarbonate as an immunotherapy adjuvant is conceptually original; mechanistic link to cGAS-STING via mitochondrial DNA release in response to pH change is novel |
| Clinical Relevance | 6 | Extraordinary ORR but n=30, single-centre, single-arm, open-label — too early to claim practice change; benchmark comparison to historical ICB ORR ~15–20% is compelling but uncontrolled |
| Population Reach | 7 | HCC is the 3rd leading cause of cancer death globally; advanced HCC with limited ICB response is a major unmet need affecting hundreds of thousands annually |
| Implementation Speed | 4 | Sodium bicarbonate is cheap and accessible, but intratumoral delivery requires interventional radiology infrastructure; Phase 2/3 RCT required before adoption; likely 5–8 years minimum |
| Evidence Strength | 4 | Prospective registered study (ChiCTR2100053537) is a strength, but n=30, single-arm, single-centre, no control arm, abstract only, medium classification confidence; cannot rule out selection bias |
Key quantitative result: ORR 93.3% (CR 53.3%, PR 40%); median PFS 31 months; OS not reached
External validation: None — single institution, no independent replication
Main limitation: No control arm makes it impossible to isolate contribution of NaHCO₃ vs. Tislelizumab alone; patient selection criteria and baseline characteristics not reviewable; single-centre China design limits generalisability
Equity implications: If confirmed, sodium bicarbonate's near-zero cost could democratise immunotherapy augmentation in LMIC settings where expensive combination regimens are inaccessible. Risk: interventional radiology access required for intratumoral delivery.
Evidence Maturity: ✅ Exploratory (confirmed — requires RCT confirmation; extraordinary signal warrants expedited Phase 2/3)
Article 3 — ctDNA MRD meta-analysis in TNBC (Barjij et al.) | PMID 42243561
Post-neoadjuvant ctDNA MRD in triple-negative breast cancer
| Dimension | Score | Rationale |
|---|---|---|
| Scientific Novelty | 6 | ctDNA MRD in TNBC is an active and growing field; this is the first meta-analysis in the residual disease subgroup specifically, which adds value, but individual studies are not new |
| Clinical Relevance | 8 | HR 4.63 with I²=0% in the highest-risk TNBC subgroup (residual disease) is immediately actionable for treatment escalation decisions (capecitabine, olaparib, pembrolizumab eligibility); this is a gap in current guidelines |
| Population Reach | 7 | TNBC represents ~15% of breast cancer; ~350,000 new TNBC diagnoses/year globally; residual disease subgroup is substantial and has worst prognosis |
| Implementation Speed | 6 | ctDNA testing infrastructure exists in major cancer centres but is not universally standardised; assay heterogeneity across studies is a barrier to immediate guideline integration |
| Evidence Strength | 7 | Systematic review + meta-analysis with QUIPS risk-of-bias assessment; I²=0% indicates homogeneity; key limitation is only 4 studies in the primary pooled estimate despite 22 studies reviewed |
Key quantitative result: Pooled HR 4.63 (95% CI 3.07–6.98); I²=0%
External validation: Meta-analytic synthesis — not a single-study validation; represents pooled evidence from existing prospective data
Main limitation: Only 4 studies contributed to the quantitative synthesis; OS synthesis not possible; assay heterogeneity across platforms limits standardisation conclusions; searched to January 2026
Equity implications: ctDNA testing is expensive and not universally reimbursed; TNBC disproportionately affects younger Black women globally — those with highest need for better risk stratification may have lowest access to ctDNA testing.
Evidence Maturity: ✅ Validated (confirmed — meta-analytic synthesis of prospective data; I²=0% supports consistency; limited by pool size)
Article 4 — BrcaDetect breast US AI (Wu et al.) | PMID 42243600
Interpretable deep learning for breast cancer on ultrasound
| Dimension | Score | Rationale |
|---|---|---|
| Scientific Novelty | 6 | Interpretable multimodal US AI is a competitive space; Grad-CAM + Shapley integration with BI-RADS and demographics adds incremental novelty over existing models |
| Clinical Relevance | 6 | Reader study showing 6% radiologist accuracy improvement is meaningful; interpretability features reduce adoption barrier; but China-only retrospective design limits immediate transfer |
| Population Reach | 8 | Breast cancer affects 2.3 million women/year globally; US is primary modality in many LMICs; broad potential reach if validated cross-population |
| Implementation Speed | 5 | Retrospective validation; no prospective clinical trial; FDA/CE regulatory pathway needed; cross-population validation required; 3–6 years realistic |
| Evidence Strength | 6 | Multicenter (5 hospitals) + external validation + reader study; but retrospective, China-only, biopsy or 3-year follow-up for ground truth (not uniform), abstract only |
Key quantitative result: AUC 0.989/0.851/0.826 (train/internal/external); radiologist + AI accuracy 0.977 vs 0.919 unassisted (p<0.001)
External validation: Yes — single external cohort (649 patients); cross-national validation absent
Main limitation: Non-Chinese population generalisability unproven; training AUC (0.989) substantially exceeds external validation (0.826) suggesting some overfitting; retrospective design
Equity implications: Potential benefit for LMIC settings where radiologist expertise is limited, but requires Chinese-specific training data recalibration for other populations; could exacerbate diagnostic disparities if not re-validated in non-Asian cohorts.
Evidence Maturity: Revised to Exploratory (consistent with classification — multicenter but single-country retrospective)
Article 5 — Methylation ctDNA in metastatic BC on CDK4/6i (Elliott et al.) | PMID 42243127
mTF ctDNA monitoring during CDK4/6 inhibitor therapy
| Dimension | Score | Rationale |
|---|---|---|
| Scientific Novelty | 8 | Tissue-agnostic methylation-based tumour fraction (mTF) as a serial monitoring tool in CDK4/6i is methodologically innovative; 5.8-month lead time for molecular progression is a clinically significant new observation |
| Clinical Relevance | 7 | HR 0.17 for ctDNA clearance is dramatic; 5.8-month molecular lead on clinical progression has direct implications for treatment switching; limited by n=57 and COI |
| Population Reach | 7 | ER+/HER2- metastatic BC is the most common metastatic BC subtype; CDK4/6i is now standard of care for hundreds of thousands of patients globally |
| Implementation Speed | 4 | Tissue-agnostic mTF assay requires commercial platform (Guardant involvement noted); single-institution data; prospective trial needed before clinical uptake; 5–8 years |
| Evidence Strength | 5 | Prospective serial sampling is a strength; n=57 and single-institution design are major weaknesses; commercial COI (Guardant Health co-authors); abstract only |
Key quantitative result: HR 0.17 (95% CI 0.07–0.41, p<0.0001) for ctDNA clearance; molecular progression precedes clinical by median 5.8 months
External validation: None
Main limitation: n=57 single institution; Guardant Health commercial conflict of interest; abstract only reviewed; no pre-specified decision algorithm for treatment change
Equity implications: Commercial assay platform may limit access in public healthcare systems and LMIC settings; if validated, could prevent unnecessary continuation of ineffective CDK4/6i
Evidence Maturity: ✅ Exploratory (confirmed)
Article 6 — SNP-NIPT cytogenetic validation (Uchida et al.) | PMID 42243550
SNP-based NIPT validated by neonatal FISH in 4,466 pregnancies
| Dimension | Score | Rationale |
|---|---|---|
| Scientific Novelty | 4 | SNP-NIPT performance is well-studied; cytogenetic neonatal confirmation is methodologically valuable but the technology is not new |
| Clinical Relevance | 6 | T21 sensitivity of 80% (lower than expected) is an important clinical limitation that could alter counselling; specificity 99.9% is reassuring; directly informs prenatal practice |
| Population Reach | 7 | NIPT is offered to millions of pregnancies annually worldwide; high-risk population here but findings inform screening policy broadly |
| Implementation Speed | 7 | SNP-NIPT already in clinical use; this validates performance rather than introducing new technology; clinically implementable now with updated counselling |
| Evidence Strength | 7 | Prospective clinical validation (n=4,466) with neonatal FISH confirmation is rigorous and uncommon; limited by single centre |
Key quantitative result: Sensitivity T21 80%, T18 100%, T13 94.6%; specificity 99.9% all; PPV 89.1%, NPV 99.9%
External validation: Single-centre Keio University — no multicentre replication
Main limitation: Single centre; high-risk referral population limits applicability to general screening populations; 2013–2022 data may not reflect current platform versions
Equity implications: NIPT primarily available in high-income countries and urban centres in LMIC; the lower T21 sensitivity finding is particularly important for counselling in settings without easy access to confirmatory invasive testing.
Evidence Maturity: ✅ Validated (confirmed — large prospective cytogenetic confirmation study)
Article 7 — CCHS survival registry (Bokov et al.) | PMID 42242766
Survival in congenital central hypoventilation syndrome
| Dimension | Score | Rationale |
|---|---|---|
| Scientific Novelty | 6 | Largest European CCHS registry; quantification of HSCR as a mortality driver (HR 6.8) is new at this scale; fills a critical evidence gap in an ultra-rare disease |
| Clinical Relevance | 7 | Directly informs risk stratification, family counselling, and monitoring intensity for CCHS+HSCR subgroup; actionable for rare disease clinicians |
| Population Reach | 3 | Ultra-rare (~1/148,000–200,000 births); population reach is minimal in absolute terms but maximal relative to the available patient population |
| Implementation Speed | 7 | No new intervention required — findings are immediately applicable to counselling and surveillance protocols in existing CCHS specialist centres |
| Evidence Strength | 7 | Registry-based, genetically confirmed, 240 patients (large for CCHS), 2012–2021; HR 6.8 for HSCR is compelling; limited by registry start-date left-truncation |
Key quantitative result: 14% overall mortality; 25-year survival 89% (isolated CCHS) vs 26% (CCHS+HSCR); HR 6.8 (95% CI 2.2–21.1)
External validation: European consortium — multi-country registry is itself a form of external validation within this disease context
Main limitation: Registry enrollment from 2012 — deaths before this date missed; limited ascertainment of milder phenotypes; abstract only
Equity implications: CCHS is globally distributed but specialist centres concentrated in Western Europe and North America; findings may not account for mortality differences in LMIC where ventilator access is limited.
Evidence Maturity: ✅ Validated (confirmed — largest consortium dataset, genetically confirmed cohort)
Article 8 — WSI AI for cutaneous vasculitis (Luo et al.) | PMID 42243228
Weakly-supervised deep learning for vasculitis on whole-slide images
| Dimension | Score | Rationale |
|---|---|---|
| Scientific Novelty | 6 | Weakly-supervised learning on diagnostic report labels (no pixel annotation) applied to vasculitis is a useful methodological contribution; disease-specific application is novel |
| Clinical Relevance | 5 | Cutaneous vasculitis vs mimicker distinction is clinically important; high AUC is impressive but no reader study to confirm real-world improvement over expert pathologists |
| Population Reach | 4 | Cutaneous vasculitis is uncommon; pathology bottleneck exists but patient population is relatively small |
| Implementation Speed | 4 | Retrospective proof-of-concept; no reader study; China-only; regulatory pathway needed; 4–7 years |
| Evidence Strength | 5 | Two-centre development/validation; AUC 98.39% is strong; weaknesses are retrospective design, no external cross-national validation, no comparison to pathologist performance |
Key quantitative result: AUC 98.39% multi-classification (cutaneous vasculitis vs 3 mimicker categories)
External validation: Two centres (both Chinese institutions)
Main limitation: No reader study; China-only; retrospective; no prospective clinical integration; AUC on training-adjacent data may be optimistic
Evidence Maturity: ✅ Exploratory (confirmed)
Article 9 — 225Ac-DOTATATE dosimetry, ACTION-1 (Sgouros et al.) | PMID 42242867
Alpha-emitting DOTATATE dosimetry in GEP-NETs
| Dimension | Score | Rationale |
|---|---|---|
| Scientific Novelty | 7 | First published dosimetry data from pivotal Phase 3 trial of alpha-emitting DOTATATE; bismuth-213 retention characterisation in tumors is technically novel |
| Clinical Relevance | 5 | Dosimetry substudy only (n=9); efficacy data pending; informs rational dosing but no patient outcome data yet; important for the field |
| Population Reach | 4 | GEP-NETs refractory to 177Lu: rare subpopulation (~5,000–10,000 patients in trial-eligible pool in US+EU) but severe unmet need |
| Implementation Speed | 4 | Phase 3 trial ongoing; dosimetry feasibility confirmed but efficacy, safety, and regulatory approval still needed; 3–5 years |
| Evidence Strength | 5 | n=9 dosimetry substudy from Phase 1b; technically rigorous SPECT/CT methodology; small n limits generalisability of dosing conclusions |
Key quantitative result: Kidney absorbed dose ~22.3 Gy; tumor ADCRBE₅ range 488–8775 mGy/MBq; favourable tumor-to-normal tissue ratio
External validation: None (substudy only)
Main limitation: n=9; no efficacy endpoint; no comparison arm; commercial trial (BMS/RayzeBio)
Evidence Maturity: ✅ Exploratory (confirmed — dosimetry feasibility only)
Article 10 — hiPSC-CM cardiac NAMs at FDA (Simpson et al.) | PMID 42243606
In vitro hiPSC-cardiomyocyte assays predict QT risk comparable to animal studies
| Dimension | Score | Rationale |
|---|---|---|
| Scientific Novelty | 7 | FDA regulatory-evidence analysis directly comparing hiPSC-CM to animal study concordance across IND applications is novel and policy-significant |
| Clinical Relevance | 5 | Indirect clinical relevance — affects drug development pipeline safety and may accelerate safer drugs to trial; does not directly change patient care today |
| Population Reach | 6 | Affects all drug candidates in cardiac safety testing; downstream benefit to all patients receiving new medications |
| Implementation Speed | 6 | FDA authorship creates direct policy pathway; concordance data supports guideline update; regulatory lag likely 2–4 years |
| Evidence Strength | 5 | Small pilot (n>20 INDs, only 5 QT+ drugs); retrospective IND analysis; positive predictive value estimates unreliable with n=5 QT+ cases |
Key quantitative result: hiPSC-CM concordance 0.71–0.82; animal study concordance 0.78; hiPSC-CM overall accuracy 0.83
External validation: Internal FDA dataset — not independently replicated
Main limitation: Only 5 QT-prolonging drugs; PPV estimates statistically underpowered; abstract only
Evidence Maturity: ✅ Exploratory (confirmed — pilot; regulatory signal present)
Article 11 — SAMe + breast cancer outcomes (Gao et al.) | PMID 42243831
SAMe hepatoprotection associated with worse breast cancer outcomes
| Dimension | Score | Rationale |
|---|---|---|
| Scientific Novelty | 8 | SAMe as a potentially chemoresistance-promoting agent via m6A RNA methylation (METTL3/METTL14) is conceptually novel; this is an unsolicited finding that challenges routine clinical practice |
| Clinical Relevance | 6 | SAMe is widely used as hepatoprotectant in Asia during chemotherapy; if confirmed, this would mandate immediate prescribing review for millions of patients; currently retrospective + PSM only |
| Population Reach | 6 | Practice predominantly in Asia (China, Japan, Korea); breast cancer is the most common cancer in women globally; practice-relevant to a large but geographically concentrated cohort |
| Implementation Speed | 5 | PSM retrospective study; mechanism is exploratory; prospective RCT needed; however, if signal holds, prescribing review is fast to implement; 3–5 years for guideline change |
| Evidence Strength | 5 | Propensity score matching (1:2) in n=1013 is a methodological strength; single-centre retrospective design, medium classification confidence, abstract only |
Key quantitative result: Prolonged SAMe (≥14 days) independently associated with shorter OS and DFS after PSM (specific HRs not reported in abstract)
External validation: None — single institution
Main limitation: Retrospective; single institution; cannot exclude residual confounding (SAMe prescribed to sicker patients?); specific HR/survival curves not reviewable from abstract
Evidence Maturity: ✅ Exploratory (confirmed — hypothesis-generating; mechanistically interesting)
Article 12 — GREGoR rare disease data model (Heavner et al.) | PMID 42239344 ⚠️ Preprint
| Dimension | Score | Rationale |
|---|---|---|
| Scientific Novelty | 6 | Interoperable multi-omic rare disease data model at scale (n=12,292) is a significant infrastructure contribution; adoption by other consortia indicates impact |
| Clinical Relevance | 4 | Infrastructure paper — indirect clinical relevance through enabling future diagnostic discoveries; no direct patient care change |
| Population Reach | 5 | Rare diseases collectively affect 300M+ people globally; this infrastructure could accelerate diagnosis for many |
| Implementation Speed | 4 | Infrastructure already deployed and being adopted; but translating to patient diagnosis requires further research pipeline; 3–7 years |
| Evidence Strength | 5 | Preprint cap applies (cannot exceed 7); methods paper with large participant base; peer review pending (Evidence Strength capped at 7 for preprints; scored 5 given methods-only design) |
Key quantitative result: 12,292 participants, 5,029 families; model adopted by other rare disease consortia
External validation: Adoption by other consortia is a form of external endorsement; full peer review pending
Main limitation: Preprint; infrastructure paper — no clinical outcomes reported; impact depends on downstream research use
Evidence Maturity: ✅ Exploratory (confirmed — preprint infrastructure paper)
Articles 13–26 — Abbreviated Phase 2 Scores
| # | PMID | Title (short) | Novelty | Clinical Rel. | Pop. Reach | Impl. Speed | Evidence Str. | Evidence Maturity |
|---|---|---|---|---|---|---|---|---|
| 13 | 42243281 | CircPTPN22-CARM1 in ALK+ ALCL | 7 | 3 | 2 | 2 | 4 | Exploratory |
| 14 | 42243643 | Pediatric leukemia Afro-Caribbean | 5 | 5 | 3 | 5 | 5 | Exploratory |
| 15 | 42243010 | CV complications in MPN (review) | 3 | 5 | 5 | 5 | 4 | Exploratory |
| 16 | 42243742 | RDW/albumin ratio in MASLD mortality | 5 | 5 | 7 | 4 | 5 | Exploratory |
| 17 | 42243686 | Mitochondrial DNA in STEMI | 6 | 4 | 6 | 3 | 5 | Exploratory |
| 18 | 42243754 | TIM-3 in cholangiocarcinoma | 5 | 4 | 3 | 3 | 4 | Exploratory |
| 19 | 42242566 | NIV in FOP children (n=3) | 6 | 6 | 1 | 5 | 2 | Exploratory |
| 20 | 42243352 | Fairness-aware ECG AI (DA-GAT-v2) | 7 | 4 | 7 | 4 | 4 | Exploratory |
| 21 | 42243532 | PLD1/PLD2 TME immunosuppression | 6 | 2 | 4 | 2 | 4 | Exploratory |
| 22 | 42243682 | Lung microbiota and tumor immunity | 5 | 3 | 5 | 3 | 4 | Exploratory |
| 23 | 42243411 | SEM symptom burden in cancer survivors | 4 | 4 | 7 | 4 | 6 | Exploratory |
| 24 | 42243704 | Uncontrolled SBP in Sri Lanka elderly | 4 | 5 | 6 | 6 | 5 | Exploratory |
| 25 | 42243732 | CA125/HE4/ROMA in advanced EOC | 3 | 5 | 5 | 4 | 4 | Exploratory |
| 26 | 42243551 | Hepatic steatosis in pediatric obesity | 4 | 4 | 5 | 5 | 4 | Exploratory |
PHASE 3 — Ranking
Conflict Check
No directly conflicting findings across articles in this batch. Complementary tension worth noting: Articles 1 (FIBOM-AI) and 7 (CCHS registry) both address management of rare/specialist conditions through data-driven tools but operate in entirely different disease spaces. Articles 3 and 5 both address ctDNA in breast cancer but are complementary — Article 3 addresses TNBC MRD post-neoadjuvant; Article 5 addresses ER+ metastatic monitoring. No contradictions identified.
Ranked Impact Table
Weighting: Clinical Relevance 30% | Population Reach 25% | Scientific Novelty 20% | Implementation Speed 15% | Evidence Strength 10%
| Rank | Article | Flag | Triage Score | Novelty | Clin. Rel. | Pop. Reach | Impl. Speed | Evid. Str. | Impact Score |
|---|---|---|---|---|---|---|---|---|---|
| 1 | FIBOM-AI: CBC predicts BM fibrosis in MPN (Art. 1, PMID 42242261) | 🟢 | 9 | 8 | 9 | 7 | 8 | 9 | 8.35 |
| 2 | ctDNA MRD meta-analysis in TNBC (Art. 3, PMID 42243561) | 🔴 | 8 | 6 | 8 | 7 | 6 | 7 | 7.10 |
| 3 | Intratumoral NaHCO₃ + PD-1 in HCC (Art. 2, PMID 42243329) | 🟠 | 8 | 9 | 6 | 7 | 4 | 4 | 6.45 |
| 4 | Methylation ctDNA monitoring in mBC/CDK4/6i (Art. 5, PMID 42243127) | ⚪ | 7 | 8 | 7 | 7 | 4 | 5 | 6.45 |
| 5 | CCHS survival registry (Art. 7, PMID 42242766) | 🟡 | 7 | 6 | 7 | 3 | 7 | 7 | 6.00 |
| 6 | BrcaDetect breast US AI (Art. 4, PMID 42243600) | 🔴 | 7 | 6 | 6 | 8 | 5 | 6 | 6.35 |
| 7 | SAMe + breast cancer outcomes (Art. 11, PMID 42243831) | ⚪ | 6 | 8 | 6 | 6 | 5 | 5 | 6.05 |
| 8 | SNP-NIPT cytogenetic validation (Art. 6, PMID 42243550) | ⬜ | 7 | 4 | 6 | 7 | 7 | 7 | 6.00 |
| 9 | hiPSC-CM cardiac NAMs at FDA (Art. 10, PMID 42243606) | 🟢 | 6 | 7 | 5 | 6 | 6 | 5 | 5.70 |
| 10 | 225Ac-DOTATATE dosimetry, ACTION-1 (Art. 9, PMID 42242867) | ⚪ | 6 | 7 | 5 | 4 | 4 | 5 | 5.10 |
| 11 | Fairness-aware ECG AI (DA-GAT-v2) (Art. 20, PMID 42243352) | ⬜ | 5 | 7 | 4 | 7 | 4 | 4 | 5.25 |
| 12 | RAR/albumin ratio in MASLD mortality (Art. 16, PMID 42243742) | ⬜ | 5 | 5 | 5 | 7 | 4 | 5 | 5.30 |
| 13 | WSI AI for cutaneous vasculitis (Art. 8, PMID 42243228) | ⬜ | 6 | 6 | 5 | 4 | 4 | 5 | 4.90 |
| 14 | Uncontrolled SBP in Sri Lanka elderly (Art. 24, PMID 42243704) | 🟡 | 5 | 4 | 5 | 6 | 6 | 5 | 5.15 |
| 15 | GREGoR rare disease multi-omic resource (Art. 12, PMID 42239344) | 🟡 | 6 | 6 | 4 | 5 | 4 | 5 | 4.75 |
| 16 | Pediatric leukemia Afro-Caribbean (Art. 14, PMID 42243643) | 🟡 | 5 | 5 | 5 | 3 | 5 | 5 | 4.60 |
| 17 | CV complications in MPN (review) (Art. 15, PMID 42243010) | ⬜ | 5 | 3 | 5 | 5 | 5 | 4 | 4.50 |
| 18 | Mitochondrial DNA biomarker in STEMI (Art. 17, PMID 42243686) | ⚪ | 5 | 6 | 4 | 6 | 3 | 5 | 4.75 |
| 19 | CircPTPN22-CARM1 in ALK+ ALCL (Art. 13, PMID 42243281) | ⚪ | 5 | 7 | 3 | 2 | 2 | 4 | 3.65 |
| 20 | TIM-3 in cholangiocarcinoma (Art. 18, PMID 42243754) | ⬜ | 5 | 5 | 4 | 3 | 3 | 4 | 3.85 |
| 21 | SEM symptom burden, cancer survivors (Art. 23, PMID 42243411) | ⬜ | 5 | 4 | 4 | 7 | 4 | 6 | 4.80 |
| 22 | PLD1/PLD2 TME immunosuppression (Art. 21, PMID 42243532) | ⚪ | 5 | 6 | 2 | 4 | 2 | 4 | 3.50 |
| 23 | CA125/HE4/ROMA in advanced EOC (Art. 25, PMID 42243732) | ⬜ | 5 | 3 | 5 | 5 | 4 | 4 | 4.35 |
| 24 | Lung microbiota and tumor immunity (Art. 22, PMID 42243682) | ⬜ | 5 | 5 | 3 | 5 | 3 | 4 | 3.95 |
| 25 | NIV in FOP children (n=3) (Art. 19, PMID 42242566) | 🟡 | 5 | 6 | 6 | 1 | 5 | 2 | 3.90 |
| 26 | Hepatic steatosis in pediatric obesity (Art. 26, PMID 42243551) | ⬜ | 4 | 4 | 4 | 5 | 5 | 4 | 4.30 |
Rank Justification Highlights
#1 — FIBOM-AI 🟢 (Impact Score 8.35) This is the standout article of the batch. Published in Lancet Haematology, FIBOM-AI achieves what every diagnostic AI study aspires to but few deliver: genuine multicentre prospective validation (including a Canadian external site) with a clinically meaningful endpoint — replacing or deferring an invasive, painful bone marrow biopsy. The 98.6% rule-out sensitivity in prospective real-world use is not a holdout-set number; it is a live clinical deployment result. CBC is the most universally available blood test on earth. Implementation barriers are low: the XGBoost framework is non-proprietary, and the inputs (27 CBC parameters + age) are already collected for every MPN patient at every visit. The main remaining step is EHR integration and regulatory classification as a clinical decision support tool — feasible within 2–3 years.
Why it matters: For MPN patients, bone marrow biopsy determines disease stage and treatment eligibility, but it is painful, resource-intensive, and not always accessible. A CBC-based tool with near-perfect sensitivity for ruling out high-grade fibrosis could safely defer thousands of unnecessary biopsies annually, reduce patient burden, and enable more frequent low-cost disease monitoring.
#2 — ctDNA MRD meta-analysis in TNBC 🔴 (Impact Score 7.10) An HR of 4.63 with I²=0% from a pooled meta-analysis is unusually clean. The residual-disease TNBC subgroup is precisely the clinical scenario where oncologists face the hardest decisions — these are patients who have failed neoadjuvant therapy and face choices about escalation (capecitabine, olaparib, pembrolizumab). A validated ctDNA MRD biomarker in this context would directly guide those decisions. The limitation — only 4 studies in the quantitative pool — is real but is contextualised by the I² result suggesting no heterogeneity. This is a near-term actionable finding for specialist centres.
#3 — Intratumoral NaHCO₃ + PD-1 in HCC 🟠 (Impact Score 6.45) Ranked third despite the most extraordinary efficacy signal in the batch (ORR 93.3% vs ~15–20% historical benchmark) because the evidence level is rated cautiously: n=30, single-arm, single-centre, no control. The cGAS-STING mechanistic story is internally coherent and supported by preclinical data, but this is hypothesis-generating clinical evidence at this stage. It is ranked #3 rather than higher precisely because its extraordinary signal-to-sample-size ratio demands rather than justifies immediate practice adoption. Watch for Phase 2/3 RCT registration.
#4 — Methylation ctDNA in mBC/CDK4/6i ⚪ (Impact Score 6.45) Tied with Article 3 on raw score but ranked 4th on Clinical Relevance tie-break (7 vs 7) then Evidence Strength (5 vs 4 — n=57 and COI reduce this). The 5.8-month molecular lead time on clinical progression is a compelling translational finding that, if replicated, would support mTF-guided treatment switching algorithms in the most common metastatic breast cancer subtype. Commercial conflict of interest and small n are the primary cautions.