Loading...
Burna reads the clinical note, surfaces the events, cites the source sentence, and quotes the CTCAE criterion verbatim. You review, modify if needed, sign. Ninety seconds per event.
SMART on FHIR app inside Epic, Oracle Health, and regional EHRs. 21 CFR Part 11 e-signature trail. AI suggests, clinicians decide.
Per adverse event, with the source sentence, CTCAE criterion, and attribution path attached.
Clinical notes supported in English, French, German, Spanish, Portuguese, Italian, Japanese, Korean, and Arabic at launch.
Every grade ships with its source sentence, its CTCAE criterion, and its named-algorithm attribution path. If the system cannot cite it, it does not suggest it.
In Phase 1 oncology, three drugs and four cycles in, attribution is the hardest call I make as PI. When Burna shows the suggested grade with the WHO-UMC and Kramer paths laid out and cited, I am reviewing a defensible chain, not rebuilding the reasoning from scratch.
Dr. Usman ShahMedical Director, GI Oncology and Phase 1 Trials, Atlantic Health
The work the coordinator trained for is the work she gets back.
A coordinator reads the progress note. The CTCAE PDF is open in another tab. She scrolls through 837+ criteria to find “neutropenia.” She reads the grade definitions. Grade 1 is ANC less than the lower limit of normal to 1,500 per cubic millimeter. Grade 2 is 1,500 to 1,000. Grade 3 is 1,000 to 500. Grade 4 is below 500. Grade 5 is death.
She returns to the note to find the ANC. The note says “low neutrophil count,” not a number. She opens the labs tab. She finds ANC of 780. Grade 3.
She documents the grade, the date, the attribution (the chemo, the antibiotic, the underlying disease?), the severity, the seriousness, the expectedness, the action taken, the outcome. She writes the narrative. She closes the encounter.
Seventeen minutes. One adverse event. Eight more patients are on her list today. Each with two to four events.
The coordinator opens the patient encounter in the EHR. Burna launches as a SMART on FHIR app inside Epic or Oracle Health. No new login, no second window.
Burna has read the note, the labs, the medication list, the prior cycles, and the regimen protocol. It surfaces the adverse events with suggested grades. “Neutropenia, Grade 3. ANC 780 on 2026-04-12. CTCAE v5.0 criterion: Grade 3 equals less than 1,000 to 500 per cubic millimeter. Source: labs tab, CBC, 2026-04-12 08:17.”
The coordinator clicks “show me why.” The criterion, the source sentence, and the attribution reasoning (WHO-UMC Probable to FOLFIRINOX, Possible to the trial drug) are visible. No black box. She agrees. She clicks sign. The grade flows into Medidata. The narrative is drafted. She moves to the next event.
Two minutes. One adverse event.
ANC 780 on 2026-04-12 08:17 · CTCAE v5.0 criterion
WHO-UMC Probable → FOLFIRINOX
TODAY 09:14
Grade 2MedDRA · 10052015NEW
Fever with hypotension responsive to fluids
Low-flow oxygen support
VISIT TRAJECTORY
CITATION PATH
ATTRIBUTION PATH
When the documentation arc compresses, your cosignatures are faster, your SAE reviews are sharper, and your attestations move with the rhythm of the day.
SAE Reviews
When a serious adverse event arrives at your desk, the attribution logic is visible. You review the clinical judgment, with the data wrangling already complete.
DLT Committees
Every dose-limiting toxicity carries its named-algorithm reasoning. Discussions move to clinical nuance and patient context.
Your Day
When the documentation arc compresses, your cosignatures are faster, your patient safety narratives are better written, and your attestations move with the rhythm of the day.
Sponsor Queries
When Medidata raises a data clarification, the source sentence and the reasoning are already in the record. The audit trail is built.
Four architectural commitments. Said on every page of the site, executed in every line of the product.
Every suggestion has a source sentence highlighted in the note, a CTCAE criterion quoted verbatim, and a named-algorithm attribution path (WHO-UMC or Kramer). If the system cannot cite it, it does not suggest it. Architectural constraint, not runtime filter.
The engine is a twelve-agent cascading constraint pipeline, citation-bound by design. Two patents filed.
AI suggests. Clinicians decide. Said on every page of the site, executed in every line of the product.
Burna will not ship a feature that removes the clinician from the loop.
When two graders disagree, the adjudication view shows every grader’s reasoning. Multi-event grading with explicit consensus detection. Optional blind grading for research settings.
Every correction leaves a trail your monitor can read.
The advisors are oncologists. The product team has watched coordinators work. The 42 regimen profiles were built from oncology protocols.
Not from a general-purpose medical knowledge graph. Specialist tool. Specialist standard.
The shape of an AE workflow with Burna at the desk. Every step preserves the chain of evidence.
About 2 minutes
Open encounter. Burna surfaces the event with citation. Review the source text. Approve or modify. Sign. Two minutes.
Research
Burna flags a case for multi-grader review. Each grader works blind to the other. The adjudication view shows both chains of reasoning. Consensus auto-detected; an adjudication committee resolves disagreements. Chain of evidence preserved end to end.
About 10 minutes
Burna flags seriousness criteria met. Attribution chain drawn. Narrative draft generated with named-algorithm reasoning. The MD reviews. The PI signs. Auto-submitted to the EDC and the sponsor. Ten minutes end to end.
A unit at Gustave Roussy, the Royal Marsden, Peter MacCallum, Samsung Medical Center, King Hussein Cancer Center, or Aga Khan University Hospital is a first-class audience.
Monday morning. The coordinator opens her queue inside the EHR. Six adverse events landed over the weekend across four trials. Each one arrives with the source sentence highlighted, a suggested CTCAE grade, and the WHO-UMC or Kramer attribution path already worked out. She agrees on five and signs them in under fifteen minutes. She modifies one (the nausea is Grade 2, not Grade 1); her reasoning is logged with the edit, and the calibration improves quietly behind her work.
At ten she meets the new coordinator who started six weeks ago. They review three charts together. The new coordinator is grading her first cycles independently. Her inconsistency rate has been falling steadily since week one because the citation is on every suggestion and the reasoning is legible. The senior coordinator’s review takes five minutes per chart, not twenty. The bench is deeper than it was a quarter ago.
By noon the SAE that came in on Friday is ready for the PI. The attribution chain is drawn. The narrative is drafted with the named-algorithm reasoning. The PI reads it, makes one edit, signs. The packet ships to the sponsor before lunch. Last quarter this would have been a Thursday task.
At three the sponsor monitor for the Phase 1 trial logs in. She finds the source sentence on the grade she would have queried. She closes the review without opening a query. The coordinator never sees it. The audit trail is already built.
She leaves at five. The grading is done.
“She agrees on five. She modifies one. She signs.”
The seven we hear most often.
No. The grading workflow compresses from 15 to 20 minutes per event to under 3. Paired-grader timing studies are designed to publish this measurement, not to leave it as a vendor claim.
You catch it, you correct it, the correction is logged with your reasoning, and the modification feeds back into your site’s tuning under explicit governance review. The platform improves through the work, and your unit’s institutional knowledge is preserved instead of lost between coordinator turnovers.
No. Burna is a research and safety documentation tool. It does not prescribe, modify dose, or trigger care-pathway alerts. Burna is narrow on purpose.
Yes. SMART on FHIR is the integration layer. Burna runs inside Epic, Oracle Health (Cerner), Athena, Allscripts, MEDITECH, and regional EHRs. If your EHR supports FHIR R4, Burna runs inside it.
Burna reads clinical text in nine languages at launch. The coordinator UI is localized in six. International sites are a first-class audience.
Not without explicit opt-in and governance review. Patient identifiers never leave your network in on-premises deployments and never leave your jurisdiction in regional cloud deployments. The grading model only sees de-identified clinical text.
Yes, and the mechanism is architectural, not aspirational. Across three Phase 1 oncology trials (Pronker et al., British Journal of Clinical Pharmacology, 2011), 85.7% of data clarification queries on adverse event forms were confirmation requests, not data corrections; 71.9% resulted in no change to the underlying data. Burna attaches the source sentence, the CTCAE criterion, and the named-algorithm attribution path at the moment of grading. The query the monitor would have asked is answered before it is opened.
Most coordinators and PIs who try Burna ask their Clinical Trials Director the same week: when can we pilot this. We make that easy.
Forward this page, request a demo, or submit your Director’s contact and we will set up the conversation.
Pilot outcome terms, restated for the Director who opens the forwarded link: in success, your unit gets citation-bound CTCAE grading inside the EHR, the labor recovery the platform was scoped to produce, and a peer-reviewed manuscript with your unit as primary authors. If the pre-registered endpoints are not met at study end, the contract terminates with no further fee, and your institution retains the deployment, the playbook, and the manuscript. Either way, your team publishes.
Subscribe to the quarterly safety and data quality briefing. Validation readouts, design partner cohort updates, and conference timing only.
No tracking beacons, no marketing email cadence.
The current validation roadmap, pre-registered endpoints, target journals, and publication timeline. One download, no calendar.