Tools


CIPHER-EHR©


From trial protocol to eligible patients and real-world evidence, with a clinician in the loop.

About


CIPHER-EHR turns a clinical trial protocol into a verified, explainable patient cohort in your electronic health records. Use it to find who is eligible for a trial today, or to emulate what the trial would have shown.

Matching trial eligibility criteria to EHR data is slow, manual and hard to check. CIPHER-EHR automates that work, while a clinician stays in charge of every decision.

You upload a protocol and AI agents do the groundwork:

  • They extract each criterion word for word and split compound criteria into single, testable pieces.

  • They find the right diagnosis, lab and medication codes.

  • They turn each criterion into a structured definition, which the platform compiles into read-only queries against your EHR.

Every definition is checked before anyone relies on it. It is tested against synthetic edge-case patients and rebuilt by a second, independent builder for comparison. A reviewer adjudicates any differences, and each judgement call is laid out with patient counts for clinician sign-off.

CIPHER-EHR is honest about missing data. Each patient is marked met, unknown or not met for every criterion, so gaps in the record become follow-up checks instead of silent exclusions.

Two uses, one platform

  • Recruitment: eligible and possibly eligible patients, screening worksheets that show the evidence, refresh alerts and outcome rates for sample-size planning.

  • Target-trial emulation: time zero at treatment start, eligibility checked at that moment, outcomes followed over time, propensity-weighted comparisons, balance checks and sensitivity analysis.

Privacy by design: CIPHER-EHR can use locally deployed LLMs and patient-level data never reach the AI model, and nothing is written back to the source EHR.

Copyright: Cardiovascular Data Science (CarDS) Lab, Yale School of Medicine, 2022