PRODUCT UPDATES · MAY 15, 2026 · 4 MIN READ
Comorbidity-aware attribution, and live PRO submission tracking
The grading workflow now reads a patient's pre-existing conditions before it proposes how an adverse event relates to treatment. The provider patient list shows when each trial patient last submitted a PRO, and check-in safety triggers follow the latest answer.
Burna AI Team
Comorbidity-aware attribution means the grading workflow reads the conditions a patient arrived with before it proposes whether an adverse event is related to treatment, to one of those conditions, or to both. In week 20 of 2026 (11 to 17 May), Burna AI added patient comorbidities to adverse event grading, put each trial patient's latest PRO submission time on the provider patient list, and made check-in safety triggers follow a patient's latest answer. The work serves the research coordinators and investigators who attribute adverse events and follow trial participants between visits.
Comorbidity-aware attribution in adverse event grading
Attribution is hard to hold steady. In the placebo arms of two phase III trials, when the same adverse event was reported in the same patient at more than one visit, its attribution category changed at least once 31% and 36% of the time (Hillman et al., Journal of Clinical Oncology, 2010). Context is one reason: numbness in the feet means one thing in a patient with no neuropathy history and another in a patient with ten years of diabetic neuropathy. That context lives in the chart, and this week it moved into the grading flow.
What shipped:
- A comorbidity list on every patient. Each entry records where it came from: extracted from the note on the first grading run, entered by a clinician, or imported from the EHR.
- Comorbidities beside the note. Clinicians view, add and remove comorbidities on the encounter, next to the clinical note they are grading.
- Comorbidities in the reasoning. The grading workflow passes the list to the step that reviews each proposed grade, and an adverse event can carry a comorbidity note naming the pre-existing condition that bears on it. The note appears on the adverse event card the clinician reviews.
- An audit entry for every change. Each edit to a patient's comorbidity list writes an audit log entry, and the workflow logs counts and sources, never clinical text.
The investigator decides attribution. The comorbidity note puts the relevant history in front of them at the moment they decide. The method the platform applies is set out in per-drug attribution with WHO-UMC and Kramer.
How a comorbidity changes the attribution question
Three clinical scenarios shaped the tests for this release:
| Pre-existing condition | Adverse event under review | The question the comorbidity note raises |
|---|---|---|
| Diabetic neuropathy | Peripheral sensory neuropathy | Is this new, or a worsening of what the patient arrived with? |
| COPD | Dyspnea | Does the breathlessness reflect an exacerbation, the study drug, or both? |
| Chronic kidney disease | Creatinine increased | Does the value depart from this patient's own baseline? |
Grading still follows the CTCAE criteria for the term, some of which are written relative to baseline. The comorbidity bears on attribution, and on whether the event is new or worse than the patient's starting point. For how that starting point defines what counts as treatment-emergent, see treatment-emergent adverse events and the baseline a patient arrives with. The short version sits in the FAQ on comorbidity-aware grading.
PRO submission tracking on the provider patient list
Coordinators following trial participants between visits need to know who has gone quiet. The provider patient list now shows each patient's latest patient-reported outcome submission as a relative time, such as "Last PRO: 3 hours ago", read from the submission itself.
- One request returns the latest submission time for up to 100 patients at once, and the query is scoped to the organization.
- In the participant app, the home screen refreshes enrollment state each time it comes back into view, and a check-in token that has already been used hides the check-in card.
The roster becomes the triage view. A coordinator sees which participants are behind on the PRO schedule the protocol sets and follows up from the list, without opening each record.
Since then: in July 2026 the web provider app was retired in favor of the platform at app.burna.ai, and in August 2026 the trial participant app became Healthnotes, the participant app sponsors deploy through sites.
Check-in safety triggers that follow the latest answer
The participant check-in asks safety questions. A "yes" to a yes-or-no safety question, or a temperature out of range, raises a safety trigger in the check-in's safety summary. Each question now holds at most one trigger, and the trigger clears when a later answer is negative: a "no" on the same question, or a temperature back in the normal range.
The safety summary a coordinator reviews reflects what the patient reported last, with one entry per question.
Trial terminology with inline definitions, in seven languages
More than 100 clinical terms across the trial management screens, including CTCAE grades, SAE reporting, biomarkers, trial phases and enrollment stages, now carry a definition on hover, marked with a dotted underline. Labels and definitions ship in seven languages: English, Spanish, French, German, Portuguese, Arabic and Chinese.
A coordinator new to the acronyms, or working in a second language, reads the definition where the term appears. The rest of the provider interface followed in the same seven languages the next week.
Workflow status that holds through a page refresh
During a long grading run, the workflow tabs now show the run's real state, in progress or complete, the moment the page reloads. The status is stored on the workflow record itself, so a tab reads it directly instead of briefly showing idle while live data reloads.
Also shipped in week 20
- Fuller example notes. The synthetic oncology notes used to try grading were rewritten as complete structured notes, with history, examination, labs, assessment and plan, and the vital signs, lab values and functional impact a grade rests on.
- A product FAQ for burna.ai. The FAQ was rewritten around the platform, with self-contained answers that AI search tools can cite: 129 questions in 28 categories at launch.
- A public site separate from sign-in. The public site stopped loading the product's authentication and gained a sitemap, an RSS feed and a plain-text llms.txt for AI search tools.
- Engineering. An end-to-end test suite for the participant app runs against an isolated, seeded environment, and both API services read their trusted origins from one shared list.
AI suggests, clinicians decide. Human-in-the-loop, always. To see comorbidity-aware attribution on cases like your own, book a 15-minute conversation.