Our Take
FDA clearance for a specific diagnostic task (six named conditions from EKG) is real; distribution through OpenEvidence multiplies adoption risk and benefit simultaneously, but the evidence of clinical accuracy is vendor-controlled.
Why it matters
Structural heart disease kills thousands annually and often goes undiagnosed at first contact. Automating EKG triage in a tool clinicians already use daily could shift when and where diagnosis begins, though real-world false-positive and false-negative rates will determine whether it saves time or creates downstream work.
Do this week
Operators: confirm with your EHR and OpenEvidence teams whether EchoNext will auto-populate in your institution's workflow before July so you can plan clinical governance review.
Pathway Labs receives FDA clearance for EchoNext
Pathway Labs, a spinout commercializing work from researchers at New York-Presbyterian Hospital and Columbia University, this month received FDA clearance for EchoNext, an AI model trained to detect six forms of structural heart disease from electrocardiogram images. The model identifies conditions including valve dysfunction (blocked or leaky valves) and chamber dysfunction (reduced pumping capacity).
The company plans a dual distribution strategy: direct licensing to hospitals and a partnership with OpenEvidence, a medical evidence search engine used by hundreds of thousands of clinicians. This second channel means the tool will be available inside a platform many doctors already consult during clinical decision-making, rather than as a separate application requiring navigation to a new interface.
Why clinicians care about EKG-based screening
Structural heart disease is frequently present but silent until acute decompensation occurs. Current workflow requires ordering echocardiography (ultrasound imaging) as a confirmatory test after clinical suspicion emerges. If EchoNext can flag candidates for further workup from EKG alone, it collapses the triage loop.
The FDA clearance pathway (510(k) category) signals the agency considered EchoNext substantially equivalent to predicate devices already in clinical use, not a novel category. That speeds deployment but does not tell practitioners the false-positive or false-negative rates in their specific patient populations. Vendor-published accuracy claims on training or test cohorts do not typically match real-world performance once the tool sees data drift from deployment sites.
OpenEvidence's distribution model matters operationally. Integrated tools see higher uptake than standalone applications, but integration also means the AI output appears in the same context as clinician queries and evidence summaries. If the model confidence scores or reasoning are opaque, clinicians may treat it as authoritative without scrutiny.
What to verify before deployment
Ask Pathway Labs for independent validation data or peer-reviewed publications on EchoNext's performance across different patient demographics, EKG quality standards, and institution sizes. FDA clearance confirms safety and intended use, not superiority over existing triage methods.
Coordinate with your clinical informatics and cardiology teams on governance before OpenEvidence pushes the integration. Confirm whether alerts will fire automatically or require clinician review before surfacing. Document decision thresholds for escalation to echocardiography so the downstream workflow does not create duplicate testing.
If your institution uses OpenEvidence as a reference tool, audit whether cardiologists have already incorporated EKG interpretation into that search context. Adding an AI layer to an existing workflow tool can either reduce friction or fragment decision-making depending on how the interface presents confidence and context.