DSUR AI writing tool — annual safety reports without the assembly tax.
Asthra is a DSUR AI writing tool that drafts ICH E2F-aligned Development Safety Update Reports from study-level safety data, exposure tabulations, the reference safety information, and the prior DSUR — with sentence-level provenance. Pharmacovigilance teams stay in Word, focused on signal evaluation instead of assembly.
Three aggregate safety formats. One engine.
Development Safety Update Report
Investigational product, ICH E2F. Annual report covering the development programme — replaces the FDA IND annual report. Drafted on this page.
Periodic Safety Update Report
Marketed product, ICH E2C(R2) format. Periodic post-authorization safety. See PSUR automation.
Periodic Benefit-Risk Evaluation Report
Marketed product, modern ICH E2C(R2) equivalent. Asthra writes either PSUR or PBRER from the same source set. See PSUR / PBRER.
What Asthra reads for a DSUR.
All structured and unstructured safety data across the development programme, ingested into a single closed corpus per reporting period.
Reference Safety Information
The current IB section listing expected adverse reactions. Drives listedness assessment throughout the DSUR.
SAE line listings
All serious adverse events across studies in the development programme — by SOC, PT, seriousness, expectedness, causality.
Exposure tabulations
Subjects exposed and subject-years per trial, cumulative and interval. Asthra reconciles denominators across studies and regions.
Prior DSUR
For continuity of cumulative tables, signal history, and benefit-risk narratives across reporting periods.
Investigator Brochure
Current and superseded versions — context for listedness and the evolving benefit-risk profile.
Study reports & CSRs
Completed or interim CSRs, abbreviated reports, and protocol-amendment summaries from the reporting period.
From data lock
to sign-off ready draft.
Load the reporting-period data (~1 hour)
Drop in line listings, exposure tables, RSI, prior DSUR, and any new study reports. Asthra reconciles dates, classifies sources, and confirms the data lock window.
Reconcile cumulative tables (~30 min)
Asthra builds the standard tabulations — cumulative exposure, SAEs by SOC/PT, listedness against the RSI — directly from line listings with cell-level traceability. Inconsistencies between studies are surfaced before drafting begins.
Draft narrative sections (~2 hours)
§3 Worldwide marketing status, §6 Cumulative exposure, §7 Data in line listings & summary tabulations, §16 Overall safety assessment, §17 Summary of important risks, §18 Conclusions. Each narrative is generated from approved sources only.
Safety scientist review
The PV team reviews in Word — chats with Asthra to refine the benefit-risk narrative, swap evidence, or pull additional cases. The reasoning behind every cumulative figure and every listedness call is auditable in the citation graph.
QPPV / sponsor sign-off
Final draft ships with the audit ledger and end-of-run QC report embedded. The sponsor signs off and submits per regional procedure — FDA (replacing the IND annual report), EMA, PMDA.
Where the cycle time goes — and where it comes back.
Manual DSUR drafting
- −4–8 weeks per DSUR cycle, longer for programmes with many active studies
- −Cumulative-table assembly across line listings: days of analyst time
- −Listedness checked against the RSI line by line
- −Benefit-risk narrative rewritten from scratch each reporting period
- −Cross-study consistency depends on writer memory
DSUR drafting with Asthra
- +Draft-ready in days; review begins as sections complete
- +Cumulative tables auto-generated from line listings with cell-level provenance
- +Listedness applied at draft time against the current RSI
- +Benefit-risk narrative reuses prior context, regenerated with current data
- +Cross-section consistency held by the citation graph
Common questions on AI-assisted DSUR writing.
What is a DSUR AI writing tool?+
A DSUR AI writing tool generates first drafts of the Development Safety Update Report — the annual safety report for an investigational medicinal product, structured per ICH E2F. Asthra ingests study-level safety data, exposure tabulations, reference safety information, and the prior DSUR, then drafts each section with sentence-level provenance back to its source.
How is a DSUR different from a PSUR?+
A DSUR (ICH E2F) covers a product still in clinical development — IND/CTA stage, before marketing authorization. A PSUR/PBRER (ICH E2C(R2)) covers a product after marketing authorization. They report different exposure denominators (trial subjects vs. patients in clinical use), pull from different safety datasets (clinical-trial line listings vs. post-marketing ICSRs + spontaneous reports), and serve different audiences. Asthra handles both — same engine, different module rules.
What inputs does Asthra need to draft a DSUR?+
The reference safety information (Investigator Brochure or equivalent), cumulative and interval safety data across all trials in the development programme, exposure tabulations (subjects, subject-years), SAE line listings, prior DSURs for continuity, and any updates to the benefit-risk profile. Asthra reconciles cumulative tables across studies and surfaces inconsistencies before drafting.
Which regulatory authorities accept DSURs drafted with Asthra?+
The DSUR format itself is harmonised under ICH E2F and is the accepted annual safety report across FDA (replacing the IND annual report), EMA, and PMDA. Asthra writes to the ICH E2F structure; the human-reviewed final document is what the sponsor submits. Asthra is a drafting tool, not a regulatory filer.
Can Asthra produce a DSUR overnight?+
For a single development programme with assembled inputs, yes — Asthra runs the drafting agent in walk-away mode, producing a first draft and QC report by the next morning. Cycle compression in pilot pharmacovigilance teams is roughly 4–8 weeks of manual assembly down to days.
How does Asthra prevent hallucinations in safety reporting?+
Closed-system retrieval (no open-internet fallback, no model-memory factual claims), every generated sentence bound to a retrieved passage, explicit gap flags when source data is missing or contradictory, and an end-of-run QC pass that checks cumulative-table arithmetic, listedness against the reference safety information, and cross-section consistency. The output ships with an embedded audit ledger.
Asthra writes across the regulatory document set.
PSUR / PBRER →
Marketed-product aggregate safety reports under ICH E2C(R2).
CSR automation →
ICH E3-aligned clinical study reports from protocol, SAP, and TFLs.
Investigator brochure →
IB drafts from nonclinical, clinical, and prior-IB content.
Protocol authoring →
ICH E6(R3)-aligned trial protocols from synopsis + IB.
Clinical evaluation reports →
MDR-aligned CERs from literature, equivalence, and PMS data.
CMC (Module 3) →
Module 3 narratives from BMRs, CoAs, methods, and stability data.
Pilot Asthra
on a real DSUR.
30-day pilot. Bring one development programme and one reporting period. We'll match Asthra's draft against your existing process — line listing to QPPV.