A global value dossier drafted section by section, every sentence cited and checked against the label
A global value dossier drafted section by section, with every sentence linked to its source and checked against the label.
See one case, screen by screen ↓A GVD is the evidence base every affiliate builds on: the NICE submission, the German benefit dossier, the US payer dossier, the Canadian and Australian reviews. It has to be ready before the first of them is due — and every sentence in it has to stand up in front of an HTA reviewer.
Most of the writing time does not go on writing. It goes on finding the right table in a 1,184-page study report, copying effect sizes and intervals by hand, checking that a timepoint sits in the testing hierarchy, that an indirect comparison is not described as superiority, and that nothing outside the expected label slips into the core text. Then the week-52 report arrives, and every sentence that cited the week-36 cut has to be found again.
Estimated split for one dossier section, by hand and with the solution.
Six specialist agents index the evidence, draft the sections side by side, check every claim and keep the message map current; the dossier lead decides and reviewers sign.
Talmirex in moderate-to-severe hidradenitis suppurativa: GVD v0.6, v1.0 due Fri Nov 20, seven markets waiting to adapt it. Here is section 6 — clinical value — screen by screen, in the working solution.
Dana opens the solution to the value story — message map v3, five pillars, 15 messages, each coloured by the strength of its evidence — and the clock: 44 days to v1.0, 3 of 8 sections signed, 61 of ~74 pages drafted. Top of her list: three findings in section 6, and a new report that changes one sentence.
“Week-52 report arrived — 1 sentence cites the older week-36 cut.”
The agents have already done the first pass: the Evidence Librarian pulled 27 passages from 6 sources, the Clinical Value Writer drafted 6,920 words against template v4, and the Claim & Label Checker checked 112 claims. 109 are clear. Click any citation and the source passage opens on the right — LUMEN-1 study report, section 11.4.2, page 655, highlighted.
“Evidence level: Randomised controlled trial. Sources are cited, never edited.”
The draft says patients felt meaningful skin-pain relief within 2 weeks. The checker shows why that will not hold: week 2 (24.1% vs 17.9%, nominal p = 0.08) sat outside the testing hierarchy. The first tested timepoint was week 4 — 38.6% vs 21.0%, p < 0.001. It proposes the week-4 wording; Dana uses it.
“The source supports pain relief from week 4, not week 2.”
One sentence still cites the LUMEN-X week-36 cut (74.0%); the week-52 report that arrived this morning says 71.3%, n = 512. Dana updates it. Another reports adolescents aged 12–17 — but the draft core data sheet says adults only. That one goes to a person: Dana and Omar Haddad, legal & regulatory, move it to the payer-only appendix.
The checker cannot change text. It proposes; a person decides — and outside-label content always goes to a person.
Each decision is recorded with who made it and when. The adolescent data now sits in Appendix B · Payer-only information, with the statement that it concerns a use outside the expected labelling and is provided only to payers and formulary committees. 112 of 112 claims are clear, and the section turns Ready to sign.
Keeping text as written is allowed — with a reason, which the reviewer sees before signing.
Dr. Maya Chen signs with her password and the meaning of her signature: “Approved — scientifically accurate and balanced”. Her name, the date and time and that meaning are bound to section version 0.6 and recorded in the audit trail. Seven markets that adapt section 6 are notified; the dossier stands at 4 of 8 sections signed.
“§6 signed by Dr. Maya Chen · 7 markets notified.”
In comparative effectiveness the draft says Talmirex was more effective than anti-TNF therapy. The network meta-analysis gives an odds ratio of 1.21 with a 95% credible interval of 0.88–1.67 — it crosses 1. The suggested wording: comparable to anti-IL-17A, numerically higher than anti-TNF, a difference that was not conclusive. Sam Patel, HEOR, signs this section.
“An indirect comparison whose interval crosses 1 supports ‘comparable’, never ‘better’.”
The message map draws a line from every message to the passages behind it, and rates the evidence: strong, moderate, low or unsupported. “Fewer surgeries and hospital visits” has no source — surgery was not a trial endpoint — so it is marked unsupported and raised as gap G-1, owned by Sam Patel and kept out of v1.0 until supported.
The dashboard tracks pages drafted against the plan, days to a reviewed section, what the Claim & Label Checker found across all 390 claims, time per section against the team’s by-hand estimate, the strength of evidence under each pillar and the weakest messages — and which reviewer still has to sign what.
The same section, seen by the five people who carry it — what their week looked like, and what it looks like now.
Six specialist agents index the evidence, draft the sections side by side, check every claim and keep the message map current; the dossier lead decides and reviewers sign.
Reads every source in the evidence library, classifies it by study, design and data cut, and indexes each passage with page and table number so a sentence can cite it.
Drafts sections 2–4 — disease background, burden of illness, current management and unmet need — preferring systematic reviews and meta-analyses over single studies.
Drafts sections 5–7 — product profile, clinical value and comparative effectiveness — reporting endpoint, timepoint, arm sizes, effect and uncertainty exactly as the study reports state them.
Drafts section 8 — cost-effectiveness and budget impact — from the model technical reports: structure, perspective, time horizon, comparators, base case, sensitivity analyses and assumptions.
Checks every sentence against the passage it cites and the draft core data sheet — numbers, timepoints, comparative wording, label fit, superseded data cuts — and raises a finding with the evidence and a suggested rewording.
Keeps the value message map current: groups messages under the five value pillars, links each to its supporting passages and rates the strength of that evidence.
Decides findings; section owners e-sign. The agents propose; a named person decides.
Ask about the evidence, a claim or a section — or tell it what to change. Answers point to the passages they rest on.
What does the evidence say about pain relief?
The evidence supports meaningful skin-pain relief (NRS30) from week 4: 38.6% versus 21.0%, tested, p < 0.001. Week 2 (24.1% versus 17.9%, p = 0.08) sat outside the testing hierarchy and did not reach significance. By week 16 it was 42.1% versus 22.6%.
Which claims are still unsupported?
Two claims lack the evidence they need: G-1, no evidence that Talmirex reduces HS surgery or hospital admissions, and G-4, utility values mapped from DLQI that are not validated for HS — both owned by Sam Patel. G-1 is the one to watch: surgery was not a trial endpoint and the model does not include it.
What does Germany need from the dossier?
Katrin Vogel adapts sections 2, 3, 6 and 7 into the benefit dossier due at launch (Sep 2027). She asked for patient-relevant endpoints and the comparison in the Hurley III subgroup, which the network meta-analysis does not cover yet — gap G-5.
Add a rule: label every indirect comparison as indirect evidence
Done. Every sentence that cites the network meta-analysis now carries “(indirect comparison)” and is rated moderate at most. The rule is on in Settings and on the audit trail.
13 screens from the working solution, on its sample data. Pick one to see it large.
Template, style, claim rules and who signs what are settings — changed in one place, applied to every section.
| Setting | Default | Choose from |
|---|---|---|
| Dossier template | GVD template v4 | GVD template v3 · Payer dossier (AMCP format) |
| Citation style in the export | Vancouver | Vancouver · AMA · Harvard |
| Writing style | Scientific — HTA reviewers | Scientific — HTA reviewers · Plain — affiliate summaries |
| Claim rules | 6 rules on | Three locked on — citation, exact numbers, outside-label content · three you can switch off · more added by asking the assistant |
| Sections 2–6 · medical review | Dr. Maya Chen | Dr. Maya Chen · Omar Haddad · Sam Patel |
| Label and payer-only content | Omar Haddad | Omar Haddad · Dr. Maya Chen |
| Sections 7–8 · HEOR | Sam Patel | Sam Patel · Dana Okafor |
| Value message map approval | Dana Okafor | Dana Okafor · Sam Patel |
Every figure is labelled: a target the solution is built to, an estimate, a typical published result, or a proven one.
“demo” = seen in the working solution, on its sample dossier data · “target” = the design goal, measured in the live solution · “estimated” = our estimate · “typical” = published figures (ISPOR Europe 2025 posters on AI-assisted value dossiers — 70–80% time saving). Talmirex, the LUMEN trials and every source shown are the working solution’s sample data. People, companies and products named on this page are fictional — characters and sample data in the working solution.
A global value dossier (GVD) is the core evidence document a market access team builds for a medicine: disease background and burden, current management and unmet need, the product profile, clinical and comparative value, and economic value. Affiliates adapt it into their own HTA and payer submissions.
The Evidence Librarian indexes your study reports, publications, SLR tables, guidelines and models. Three section writers draft against your GVD template with a citation on every sentence, the Claim & Label Checker checks each claim against its source and the draft core data sheet, and the Value Message Mapper keeps the message map current. Your writers decide the findings and reviewers sign.
Yes. Every sentence is checked against the draft core data sheet. Content outside the expected label — for example adolescent data when the label is adults only — is raised as a finding and decided by the legal & regulatory reviewer, who can move it to a payer-only appendix with its statement or remove it.
The Evidence Librarian marks the older report superseded and finds every sentence that cites it. The Claim & Label Checker raises each one with the old and new figures side by side and a suggested update for a person to accept.
The claim rules stop it: an indirect comparison whose interval crosses 1 supports “comparable”, never “better”. In the working solution, “more effective than anti-TNF therapy” is flagged against an odds ratio of 1.21 (0.88–1.67) and reworded.
Yes. The checker cannot change text; it proposes and a person decides. Keeping a sentence as written needs a reason the reviewer sees before signing, re-running the agents keeps every decision a person made, and each section is e-signed by a named reviewer.
The design goal is a reviewed section in 2.1 days, where the team plans about 9 by hand. Published studies of AI-assisted value dossiers report 70–80% less drafting time than a writer working alone.
The Agentic Solution Engine builds and deploys it from your requirements — your GVD template, claim rules, sign-off roles and evidence library — and it goes live once every quality gate has passed. We will walk you through it on your own evidence first.
Every rebate invoice line checked against the contract, and every dispute backed by its clause.
1-min filmPayer ContractsEvery payer contract and amendment becomes cited terms in force, with every deadline on a clock.
1-min filmCoverage MonitorEvery payer policy change caught within a day of posting, and checked against what the plan contracted.
1-min filmWe’ll draft a section of your global value dossier from your own study reports and SLR tables.