AI-assisted rare disease triage
Zebreon helps general practitioners navigate the diagnostic odyssey. Structured clinical interviews, real-time phenotype matching, and cross-referenced genetic evidence, in one conversation.
6-year-old male, progressive muscle weakness in lower limbs, Gowers' sign positive, elevated CK at 15,000 U/L.
The combination of proximal weakness with markedly elevated CK and a positive Gowers' sign raises concern for a dystrophinopathy.
Capabilities
A structured clinical interview with interactive widgets (single-choice, multi-select, free text, and verification cards) so clinicians answer naturally, not in forms.
Aggregates and cross-references publicly available genetic and phenotypic knowledge bases to surface relevant candidate conditions from reported symptoms.
Extracts and maps symptoms to standardised clinical ontology terms in real-time, enabling precise semantic matching across genetic and phenotypic knowledge bases.
Designed to support facial phenotyping approaches, adding a visual dimension to help identify recognisable patterns associated with genetic conditions.
Built-in growth charts based on international reference standards and an interactive pedigree editor with standard file import/export. No separate software needed.
Produces structured triage summaries with candidate differentials, suggested next steps, and relevant specialist pathways, exportable as PDF or interoperable data formats.
Workflow
Enter basic demographics, presenting symptoms, family history, and any prior investigations. Upload photos, growth data, or existing reports.
Zebreon asks targeted follow-up questions based on emerging phenotype patterns, confirming or excluding clinical features with each turn.
Behind the scenes, standardised phenotype terms are extracted and queried across genetic and disease knowledge bases. A live evidence sidebar shows what the system is finding.
A structured report with ranked conditions, confidence levels, recommended tests, gene candidates, and referral pathways. Ready to share with specialists.
Data & Privacy
Patient names and identifiers are automatically stripped before any data reaches the language model. The AI only ever sees de-identified clinical information.
Facial photos and genetic data are processed locally. Only structured, de-identified phenotype terms are sent to cloud AI for clinical reasoning.
Every AI interaction is logged with session context, prompt versioning, and PII replacement counts, enabling full traceability for clinical governance.
Self-host the entire stack or use enterprise LLM endpoints with BAA/DPA agreements. You control where the data lives and who can access it.
Who is it for
You suspect something rare but aren't sure where to start. Zebreon turns your clinical observations into structured evidence and a clear next step.
Use the workup wizard to build comprehensive case files, manage pedigrees, and prepare structured referral packages for genetic testing labs.
Systematically phenotype patient cohorts with standardised ontology coding, export structured data, and integrate with existing bioinformatics pipelines.
Get in touch
We're in early development and looking for clinical partners and early adopters. Reach out and we'll get back to you.
Thanks for reaching out. We'll get back to you shortly.
Or reach us directly:
info@zebreon.com