CoreSR

An AI assisted platform that will take your systematic review idea from protocol to publication.

Visit us:

[CoreSR @CoreSRR]

@DavidDan_Ngn @EUplatinum @GSK_UofT @GuyattGH @EBMUrology Agree on the PCSM vs OS mix-up. In prostate screening SRs, disease-specific mortality is usually the right primary: competing risks and treatment of screen-detected disease dilute all-cause, so OS needs far more events (and longer follow-up) before it moves.

[CoreSR @CoreSRR]

@JosephHabibi_MD @AGA_CGH @ElhadiMuhammed @LorenLaine2 @AlfredoLucendo @EimirLab @WalterChanMD @RenaYadlapati @Eric_Low_GI @EsophagusDoc 13 trials across EoE therapies is a dense network for ranking. The node map matters as much as the model — whether topical steroids stay separate from systemic, and how biologics vs diet vs dilation are split, can reshuffle who looks first.

[CoreSR @CoreSRR]

Authors' caveat: they excluded multicomponent interventions and trials without dose data — exactly what real reviews often include.

They flag the next step: joint CNMA + dose–response. And consistency checks need to cover the chosen dose–response function, not only direct vs indirect.

#ComponentNMA

[CoreSR @CoreSRR]

Most NMAs treat dose as a node label. That throws away the curve.

BMC Med Res Methodol: frequentist dose-response NMA (DR-NMA) in R package netdose — linear, exponential, quadratic, fractional polynomials, restricted cubic splines. Predict effects across doses; shared agents can even bridge disconnected networks.

https://t.co/SJICjXIbRY
#NetworkMetaAnalysis #SystematicReview #EvidenceSynthesis

[CoreSR @CoreSRR]

@HealthyFellow @HealthyFellow Dose-response NMA for muscle mass in older women with sarcopenia is the right frame — a league table alone hides whether resistance dose drives the curve. Did they model frequency/volume continuously, or only as discrete exercise families?

[CoreSR @CoreSRR]

@FrontPsychol @FrontPsychol Exercise-type NMAs for aggression in kids/teens live or die on node-splitting — aerobic, martial arts, and team sport aren't interchangeable. Curious whether they kept child vs adolescent bands separate, and what active-control arms looked like vs waitlist.

[CoreSR @CoreSRR]

PRISMA-NMA is from 2015. Methods moved on.

JCE scoping review (n=61): 37 items beyond the checklist — node definitions, transitivity checks, newer NMA modeling.

Delphi next. 2015 is a floor, not a ceiling.
https://t.co/q7vlQekr8D
#NetworkMetaAnalysis #SystematicReview

[CoreSR @CoreSRR]

@GIE_Journal @NkengehNTD Upfront necrosectomy vs step-up for WON is almost entirely about who qualifies as "step-up" and when crossover happens. Curious whether they stratified by infected vs sterile collections, and how many step-up arms still ended in necrosectomy after drainage failure.

[CoreSR @CoreSRR]

@joinBJAN PIEB vs CEI (both + PCEA) is a clean question — the MA will live or die on how they handled outcome timing (rest vs movement pain, 24h opioid use) and whether block technique/volume was treated as a modifier. Did the benefit concentrate in the first 24h or hold through POD2–3?

[CoreSR @CoreSRR]

RoB judgments stick when they're tied to the trial text, not a remembered checkbox.

GLASS surfaces quote-level support for each domain. You still decide — with the evidence beside the call.

https://t.co/CvgrsaS4o4

#SystematicReview #RiskOfBias #EvidenceSynthesis

[CoreSR @CoreSRR]

@Interasma @Interasma Severe-asthma biologic NMAs get noisy once eosinophilic vs allergic phenotypes and exacerbation definitions are pooled. Did they stratify by baseline eos / FeNO, or force one ranking across phenotypes?

[CoreSR @CoreSRR]

@FrontNeurosci Post-stroke upper-limb NMAs live or die on how therapy families are lumped — constraint-induced, robotics, and mirror therapy aren’t interchangeable active controls. Curious whether they split networks by time-since-stroke, and how much of the hierarchy is mostly indirect.

[CoreSR @CoreSRR]

Precision medicine breaks the “same population” assumption in meta-analysis.

BMC Med Res Methodol reviews 8 methods for mixed biomarker populations (AD pairwise, AD NMA, AD+IPD NMA) — think KRAS WT/MT/mixed EGFR trials in mCRC.

Takeaway: IPD methods look cleaner; AD methods are what you can actually run. Pick the frame by the decision problem (one subgroup vs estimating the interaction).

https://t.co/2RIXAKgjKk

#EvidenceSynthesis #MetaAnalysis #NetworkMetaAnalysis #SystematicReview

[CoreSR @CoreSRR]

@FrontEndocrinol T2DM + CKD is where ranking GLP-1RAs, SGLT2is, and nsMRAs gets messy — shared pathways, overlapping indications, CV vs kidney endpoints that aren’t exchangeable. Separate networks per outcome family, or one hierarchy forced across both?

[CoreSR @CoreSRR]

@RetinaJournal Endophthalmitis after vitrectomy is rare enough that risk-factor MAs live or die on case definitions and denominator quality. Did the prevention signal hold after splitting intraoperative vs postoperative prophylaxis, and how much residual confounding from case-mix did they flag?

[CoreSR @CoreSRR]

A living systematic review is a process, not a PDF.

Surveillance → re-screen → re-extract → re-appraise → update the synthesis. Most stacks turn that into a three-tool handoff.

CoreSR keeps the question living in one workspace so the next update isn’t a rebuild.

https://t.co/gaP1BEp7Ze

#LivingReview #SystematicReview #EvidenceSynthesis

[CoreSR @CoreSRR]

Component NMA isn't just another league table. You estimate effects for treatment components and combinations — including ones no trial ran directly.

Walkthrough in CoreSR:
https://t.co/fXbvglpj6O

#ComponentNMA #NetworkMetaAnalysis #SystematicReview #EvidenceSynthesis #CoreSR

[CoreSR @CoreSRR]

Component NMA isn't just another league table. You estimate effects for treatment components and combinations — including ones no trial ran directly.

Walkthrough in CoreSR:
https://t.co/fXbvglpj6O

#ComponentNMA #NetworkMetaAnalysis #SystematicReview #EvidenceSynthesis #CoreSR

[CoreSR @CoreSRR]

LLMs for systematic reviews are rising faster than validation.

JCE scoping review (n=37): support spanned 10 of 13 SR steps — search 41%, selection 38%, extraction 30%. GPT in 89% of studies.

Authors called LLM use promising in ~half; ~1 in 5 judged it nonpromising.

https://t.co/oqEN8usTUq

[CoreSR @CoreSRR]

Adjuvant ICI pooled across solid tumors is useful for a high-level signal, but I'd want the forest split by tumor type / PD-L1 enrichment before treating the pooled HR as decision-facing. Phase II–III mix also tends to inflate early optimistic estimates — worth noting how many phase III DFS events drove the summary.