Launching SynthesisBench: an open screening benchmark across 85 systematic reviews, 2,327 records and 26 systems. Top Youden's J: TITAN-SR v6 0.879, Claude Opus 5.5 0.831, GLASS-SR 0.827. Run by CoreSR. synthero.ai/screening

New CoreSR models: TITAN-SR v6 screens, GLASS v3 appraises with quotes from the paper, ROB BERT v6 fills in RoB 2 while a second model checks it. On 60 held-out reviews, TITAN kept 932 of 933 includes and screened out 78% of the rest. youtu.be/uTbnVkTiOnE

Which AI should screen 10,000 abstracts? Chatbots write one word at a time. BERT reads both ways and gives a score you can tune for high recall. TITAN-SR kept 99.8% of included studies in 22 later Cochrane reviews. sciencedirect.com/science/ar… coresr.ai #SystematicReview #Epi

coreSR now runs inside Claude and ChatGPT I connected the coreSR MCP server and gave Claude one sentence: build a search on sildenafil for pulmonary arterial hypertension, run it, and screen it with TITAN at the strictest setting, randomized trials only. Twelve minutes later, Claude had searched PubMed and Europe PMC, run TITAN across 1,022 unique records, and written a report that reads like a methods section. It also checked the screen against landmark trials. Two were missing. Claude traced the gap to the search strategy, ran a corrected supplement, and recovered both. That division of labour is the point. Validated engines do the science. Claude does the legwork. You make the final calls. The full demo runs under three minutes. Connect at coresr.ai/mcp. #SystematicReview #EvidenceSynthesis #MetaAnalysis #ClaudeAI

Our TITAN-SR validation paper is out in Journal of Clinical Epidemiology. Biomedical BERT ensemble for title/abstract screening — trained on ~20k reviews — beat ASReview and commercial LLM chatbots. doi.org/10.1016/j.jclinepi.2… #SystematicReview #EvidenceSynthesis

CoreSR now offers a free tier, forever, for the full manual systematic review workflow. No AI, no review limit, no expiration date. Screening, data extraction, risk-of-bias assessment, and meta-analysis run on the platform at no cost, for as many reviews as your team needs. Most platforms gate the basics behind a trial that expires or a seat that costs money from day one. We built CoreSR around AI-assisted screening and GRADE certainty automation on network meta-analysis, and those stay paid features. But a reviewer, a student, or a society shouldn't need a credit card to run a modern systematic review. The manual tools stay free, permanently. See the platform in action: lnkd.in/gSYP4KFg Try it at coresr.ai. hashtag#systematic review, #coresr.ai

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: youtube.com/watch?v=hW73KzeX… #ComponentNMA #NetworkMetaAnalysis #SystematicReview #EvidenceSynthesis #CoreSR

Typical SR stack: one tool for screening, another for RoB, R for the analysis, GRADEpro for certainty, a doc for the living update. Every handoff is a chance for PICO drift — mismatched outcomes, certainty that doesn't match the model you just ran. One project from question to published review. coresr.ai

How long would it take you to complete a risk of bias assessment using RoB 2.0 on 12 studies with over 72 assessments across multiple outcomes? With CoreSR? About 5 minutes. youtu.be/AgHtmx3YSOk

Most NMAs take a team and the better part of a year. In this video, the analysis finishes in minutes.CoreSR runs on a dataset built to mirror the 2026 BMJ obesity pharmacotherapy NMA (Nong et al.): 266 trials, 539 arms, 24 outcomes, covariates and RoB intact. Trial structure is real; outcomes are simulated and calibrated so a refitted RE model reproduces the published network http://estimates. It ingests the file, fits every network, and returns effects with CrIs, rankings, heterogeneity, and GRADE-ready SoF tables across all 24 outcomes.Arithmetic was never the bottleneck. Preparation was. Remove those hours and a small team does what only a large one could before.#coreSR #MetaAnalysis #NMA coresr.ai @GuyattGH youtube.com/watch?v=wBh_GNqx…

CoreSR performs a full network meta-analysis with GRADE assessment in under 10 minutes. youtube.com/watch?v=Q-ldHQeO… #NetworkMetaAnalysis #GRADE

Publication type classifiers for biomedical literature often degrade under distribution shift. This preprint evaluates knowledge-guided perturbations to strengthen robustness in study design indexing. arxiv.org/abs/2605.11502v2 #GRADE #AIinMedicine