Predictive organ donor identification
17 people die every day waiting for an organ. The delay is not a care failure. It is an information failure. Vigil fixes it.
Potential donor · 24 y/o, Neuro ICU
How it works
Vigil continuously reads 38 clinical variables from the hospital EHR in real time via HL7 FHIR API. Neurological trajectory, hemodynamic instability, ventilator dependence, lab trends. The signal is already in the chart. Vigil reads it first.
Before any alert is sent, Vigil cross-references the patient's active diagnoses, cancer history, and comorbidities to automatically exclude non-viable organs. The OPO receives only actionable information.
The OPO receives a prioritized notification 24 to 48 hours before clinical declaration, with viable organs identified and procurement lead time already started. Family approach, logistics, and preservation all begin earlier.
Who it is for
Vigil gives your coordinators a 24 to 48 hour head start on every potential donor. Earlier identification means higher family consent rates, better procurement logistics, and stronger CMS performance metrics. The ROI is direct and quantifiable before you sign a contract.
A missed referral exposes your institution to CMS scrutiny and public performance reporting. Vigil eliminates that risk by catching declining patients before any manual process would. No new infrastructure required. Integrates with Epic and Cerner via existing FHIR endpoints.
Why it matters
The gap is not surgical. It is logistical. By the time a family is approached, the optimal preservation window for several organs has already closed.
Vigil does not change medicine. It changes the timing of information. That is enough to save lives.
AUROC. A 2023 Scientific Reports study validating this approach showed that a temporal machine learning model, using only routine data from the ICU record, identified potential organ donors with an AUROC of 0.966. The signal exists. We built the system to act on it. Sauthier N, Bouchakri R, Carrier FM, et al. Automated screening of potential organ donors using a temporal machine learning model. Sci Rep. 2023;13:8459. doi:10.1038/s41598-023-35270-w
Mission
Vigil exists to close the gap between the organs that are recoverable and the organs that are actually recovered. We believe predictive AI applied at the right moment in the ICU can add tens of thousands of life-years annually in the US alone.
Our goal is to become the standard of care for donor identification across every OPO and major ICU program in the country, and then internationally.
The technology
Vigil is built on a temporal machine learning architecture trained on MIMIC-IV, the largest publicly available ICU dataset with over 300,000 admissions from Beth Israel Deaconess Medical Center. The model monitors 38 structured clinical variables including GCS trajectory, ICP trends, vasopressor escalation, and ventilator parameters.
EHR integration is via HL7 FHIR API, the federal interoperability standard all major hospital systems including Epic and Cerner are required to support. This means no custom integration at each deployment.
The baseline model is XGBoost, moving to a temporal neural network architecture consistent with published literature achieving AUROC 0.966. Deployed on AWS HIPAA-eligible infrastructure with a React dashboard for OPO coordinators.
The team
CEO and Cofounder
Clinical Research Operations Coordinator at UCLA Health. MPH from USC Keck School of Medicine, BS in Neuroscience. First-author publication in PNAS. Manages 25+ active clinical studies in transplant and hepatology. Identified the donor identification gap through direct ICU observation and OPO interviews.
CTO and Cofounder
25 years building inside Epic, first at the University of Michigan and now at UCLA, where she is embedded in the EMR system leading AI initiatives. Led the technical integration for UCLA's acquisition of West Valley Hospital. Completing a master's in Healthcare Informatics and AI at UCLA.
Background
The US transplant system is the most advanced in the world and still loses 17 patients per day to waitlist mortality.
When a patient begins neurological decline in the ICU, the clinical team is focused on treatment. The pivot to donation evaluation happens only after brain death is formally declared, a process that can take 12 to 24 hours from the moment the trajectory becomes medically clear. By the time an OPO coordinator is notified, the optimal preservation window for the liver, heart, and lungs has often already narrowed.
57 OPOs across the US are federally evaluated by CMS on donor conversion rates. Since 2020 reform, underperforming OPOs face public scrutiny and risk losing their federal designation. Despite this pressure, OPOs still learn about a potential donor only when a referral is made. Existing tools such as InVita Healthcare Technologies' iReferral automate that referral, sending it once standard clinical triggers are met in the EHR. They do not predict the decline ahead of time.
Vigil adds the predictive layer this workflow is missing.
Typical limits once an organ is recovered. Hearts and lungs leave the least room, so recipient matching, transport and perfusion devices need to be lined up before recovery.
Request a demo
If you are an OPO coordinator, transplant program director, or hospital administrator, we want to talk to you.
Explore the interactive demo
as a hospital admin or an OPO user. Username Vigil,
password Vigi12026 (the fifth character is the
number one). All data is fictional.
Your request has been received. In the meantime, you can explore the interactive demo.