Yarrow is a clinically-trained AI nurse triage system designed specifically for oncology care. It synthesizes patient-reported symptoms, treatment history, and real-time vitals to stratify risk, reduce unnecessary ED visits, and surface the patients who need you now.
For outpatient oncology clinics, the challenge isn't only clinical outcomes — it's the operational reality of staying available to patients around the clock while protecting a nursing workforce that is already stretched thin. Yarrow was built for that reality.
Outpatient oncology patients generate a high volume of after-hours calls — many driven by anxiety, uncertainty about symptoms, or a lack of accessible guidance. Studies show that a significant proportion of these calls do not require clinical escalation, yet each one demands a nurse's time and judgment.
Poirier et al., Clinical Journal of Oncology Nursing, 2009 · Stacey et al., Supportive Care in Cancer, 2016Oncology nursing shortages are accelerating. Reactive triage — driven by phone volume rather than clinical priority — forces experienced nurses into a triage bottleneck rather than direct care. AI-assisted triage has demonstrated the potential to reduce non-urgent nurse call burden by up to 40%, freeing capacity for the calls that matter most.
Stricker et al., Annals of Oncology, 2022 · Bruinooge et al., JOP, 2018Even in well-resourced clinics, telephone triage is highly variable. Escalation decisions depend on the individual nurse's experience, shift fatigue, and access to the patient's full treatment context — factors that introduce avoidable risk and inconsistency across a patient population that cannot tolerate missed signals.
Poirier et al., Clinical Journal of Oncology Nursing, 2009 · Donaldson et al., Cancer Nursing, 2020When genuine emergencies — febrile neutropenia, sepsis, acute toxicity — are missed or delayed in triage, the downstream consequences are severe. Timely structured escalation is associated with a 20–30% reduction in 30-day mortality, and structured triage tools consistently outperform ad hoc assessment.
Carmona-Bayonas et al., Journal of Clinical Oncology, 2020 · Mayer et al., Cancer, 2011Yarrow operates as an always-on clinical intelligence layer — accessible via phone, patient portal, or app — guiding patients through structured symptom assessment and surfacing real-time risk to your care team.
Patient reports symptoms via app, web portal, SMS, or integrated phone system — 24 hours a day, 7 days a week.
Hygia conducts a dynamic, conversational intake guided by oncology-specific clinical protocols and NCI CTCAE grading criteria.
The model cross-references symptom data against the patient's active regimen, prior AEs, and real-time EMR vitals to generate a risk score.
Low-risk cases receive self-care guidance. Moderate cases are queued for nurse callback. High-risk cases trigger immediate clinical alert — automatically.
Every interaction is automatically logged in the patient's chart with structured clinical notes, reducing documentation burden on your nursing staff.
Yarrow connects natively to your existing EMR infrastructure via HL7 FHIR R4 and proprietary APIs — no rip-and-replace, no parallel workflows. It reads, contextualises, and writes back to the patient record in real time.
Yarrow connects to over 50,000 health systems, clinics, labs, and imaging centers across North America via the CommonWell Health Alliance and Carequality frameworks — pulling longitudinal patient records regardless of where prior care occurred.
Nationwide coverage through Carequality and CommonWell frameworks, ensuring patient records follow the patient — not the institution.
Admission, discharge, transfer events and lab results are pushed to Yarrow automatically, keeping triage context current without manual lookup.
End-to-end encryption, role-based access controls, and full audit logging. No patient data is used for model retraining without explicit institutional and patient consent.
Yarrow's foundation model was developed on de-identified oncology triage records and validated symptom grading datasets. Upon deployment, it is fine-tuned on your clinic's own protocols, formulary, and patient population.
Pre-trained on large-scale de-identified oncology nursing triage interactions, NCI CTCAE grading corpora, and clinical decision support literature encompassing over a decade of symptom escalation data.
At onboarding, Yarrow ingests your institution's specific triage protocols, on-call escalation pathways, formulary, and active treatment regimens — grounding every assessment in your standard of care.
Using retrospective triage outcomes from your own patient population, Yarrow fine-tunes its risk stratification thresholds to reflect the specific demographics, comorbidity burden, and regimen mix of your clinic.
With clinician consent, every resolved triage episode feeds a reinforcement loop — improving calibration over time and flagging protocol drift for clinical review.
Generic AI triage tools apply one-size-fits-all thresholds to a patient population defined by complexity and heterogeneity. Yarrow is the first oncology triage system built to reflect the way your team actually works.
We're onboarding a limited cohort of cancer centers ahead of our full launch. Reach out to arrange a tailored demonstration or discuss a pilot partnership.