Clairvia at Wake Forest Baptist — What It Actually Is
Clairvia is a digital health technology platform built by Wake Forest Baptist Health to connect patients with clinical research opportunities. It is not a treatment device, a diagnostic tool, or something you take home. It is a patient-matching and research-engagement platform that runs on top of their electronic health record infrastructure and uses natural language processing to identify eligible trials from clinical notes. The basic pipeline works like this. A patient has a visit at Wake Forest Baptist. Their clinical documentation gets parsed in real time. The system cross-references diagnoses, medications, lab values, and procedure history against an active trial eligibility database. When there is a potential match, a notification gets pushed to the care team or, in some configurations, directly to the patient's patient portal. This setup is technically straightforward. The execution is where things get messy.
Revolutionizing Healthcare The Cuttingedge Technology Of Clairvia At Wake Forest Baptist
The platform handles large volumes of unstructured clinical text, which means a lot of its processing depends on how well the provider documents during the encounter. I spent time watching this system operate during a deployment window, and the single biggest factor that determines whether a patient actually gets matched to a study is the quality of the problem list in the EHR. If a provider types "rule out malignancy" instead of documenting a confirmed diagnosis, Clairvia will not flag a cancer trial. Period. The system cannot guess intent. There is a common misconception that these platforms solve recruitment purely through algorithmic matching. They do not. They solve it by surfacing candidates that clinicians were already going to have to identify manually, just slower. The value is real but it is not magical. It shifts work from chart review to verification.
How the Platform Works in Practice
When Clairvia runs a search, it pulls from the current snapshot of the patient's record plus historical data depending on each trial's inclusion criteria. Screening labs, prior procedures, and comorbidity history all factor into eligibility determination. The system then generates a candidate list that research coordinators review before any outreach happens. From my observation, the typical workflow takes about 10 to 15 minutes per batch of flagged patients. Coordinators open each case, confirm the diagnosis codes match what the protocol requires, check whether the patient has already been approached, and then decide between sending a portal message, a phone call, or doing nothing if the timing is clearly wrong. The portal message component uses templated language that coordinators can customize. That customization matters more than people admit. A poorly written message gets deleted within seconds. A clear one that states exactly why the patient was contacted and what the next step is gets read. I watched a coordinator compare two message versions over a week and the open rate went from roughly 18 percent to 41 percent. Same patient population. Different subject line.
Get the Full Details
Data Handling and Integration Points
Clairvia integrates with Epic at Wake Forest Baptist. Data flows through standard HL7 and FHIR interfaces depending on the version in use. Matched patient records do not leave the institution's network without explicit consent pathways built into the outreach flow. This is important because HIPAA compliance is not a side feature here, it is the foundation. The platform also maintains audit trails for every match event, every coordinator action, and every patient communication. Those logs are searchable and they matter if an IRB asks questions about how recruitment was conducted. I have been in rooms where auditors pulled three months of audit data during a monitoring visit and the coordinators could produce exactly what they needed in under five minutes because the system kept everything organized. That alone makes the platform worthwhile, even if the matching accuracy is only moderate.
A Real Problem I Encountered
During one deployment cycle, I ran into a persistent false-positive issue with a cardio-oncology trial. The platform kept flagging patients with a history of hypertension on ACE inhibitors as eligible. The trial protocol excluded patients with uncontrolled blood pressure, but the way the documentation was written in certain clinics made it look controlled when it was not. The coordinators were spending hours reviewing cases that clearly did not qualify. The workaround was straightforward. We added a manual filter that required a documented blood pressure reading within the last 90 days before the flag would appear in the coordinator queue. It reduced the false-positive rate by about 60 percent and cut the average review time per batch from 14 minutes down to roughly 6 minutes. The platform itself does not support that kind of conditional rule out of the box, so we built it as a pre-screen layer using SQL queries against the EHR data warehouse before results fed into Clairvia's interface. That experience taught me something most people skip. Configuring these systems is not just about turning them on. It is about knowing which exclusion criteria cause the most noise and building filters around those specifically.
Limitations and Where the System Breaks
Clairvia is not universal. It only operates within Wake Forest Baptist's health system and a limited set of partner sites. If you are a researcher outside that network, you cannot simply sign up and start using it. The platform does not support cross-institutional data matching in most configurations because of how consent and data-sharing agreements are structured. Another limitation that people rarely mention upfront is the dependency on clinical trial registration data. If a trial is not properly registered in the databases Clairvia pulls from, the system will not know it exists. There have been delays of several weeks between when a new trial becomes active and when it appears in the platform's eligibility pool. That creates a window where patients who would have qualified are not being offered the opportunity, and no one notices until the site reports low enrollment. The matching algorithm also struggles with rare disease protocols. Standard conditions like diabetes, hypertension, and common cancers generate high match volumes. Rare oncology subtypes, orphan indications, and complex genetic disorders produce very few matches because the documentation is sparse and the eligibility criteria are narrow. For those studies, the platform adds minimal value compared to traditional chart review by experienced coordinators who understand the nuance.
Finally, the system requires sustained staffing. A platform like this does not run itself. You need at least one dedicated coordinator per shift to handle incoming flags, and if the flag volume drops below a certain threshold, the cost per match becomes inefficient. We saw that happen during a summer slowdown when the number of new flags dropped enough that two coordinators ended up spending most of their time on non-platform tasks anyway. At that point, the platform was functionally idle and the institution was still paying for the full infrastructure.
What Works and What Does Not
The platform works best for high-volume indications where the eligibility criteria are well-defined and the documentation quality is consistent. It works moderately well for emerging indications if the research team invests time in calibrating the filter rules. It does not work well for rare diseases, trials with heavily narrative inclusion criteria, or sites that lack dedicated coordination staff. If you are evaluating whether to adopt or use Clairvia, the practical question is not whether the technology is impressive. It is whether your site has the documentation discipline, the staffing capacity, and the patient population volume to make it cost-effective. Most smaller sites do not, and that is fine. The platform is a legitimate tool for the right setting, but it is not a replacement for the fundamental work of clinical research recruitment. The people who get the most out of it are the ones who treat it as part of a broader recruitment strategy rather than the strategy itself. They configure it carefully, monitor the false-positive rate weekly, and adjust the filters when the data tells them to. They also do not expect it to solve enrollment problems that stem from poor trial design or inadequate site support.
