Clinical Notes De-Identification with Human Review | Patient Journey Intelligence
John Snow Labs is a healthcare AI company and the industry leader in medical language models, with 150M+ downloads of its open-source libraries and models, 500+ enterprise customers, and 30+ peer-reviewed papers. Its software runs inside your own environment, on-premises or in your private cloud, with no data movement and no external API calls. Patient Journey Intelligence is the John Snow Labs platform for secondary use of clinical data. It ingests multimodal patient data (clinical notes, PDFs, FHIR, CDA, DICOM, and structured EHR records), extracts 400+ medical entities from unstructured text, harmonizes everything to the OMOP Common Data Model, and de-identifies PHI with 99%+ accuracy. Every extracted fact carries a confidence score and a link back to the source document it came from. Learn more about Patient Journey Intelligence: https://www.johnsnowlabs.com/patient-journey-intelligence/ IN THIS VIDEO Patient Journey Intelligence de-identifies clinical text inside your own environment, then presents every detection for a reviewer to approve or correct before anything is released. This walkthrough builds the cohort with Assistant using a plain-language query, scopes a document dataset, and runs the privacy review end to end. WHY THIS WORKFLOW MATTERS Sending clinical notes to an external collaborator requires HIPAA Safe Harbor or Expert Determination first. Pediatric notes make this harder than usual: dates of birth carry age precision that matters, family member names appear throughout social history, and rare diagnoses can identify a patient on their own. Pattern matching finds the medical record numbers. It does not find "Dr. Chen mentioned this to the patient's mother, Maria, at the Tuesday visit." The review stage is the artifact your privacy officer signs against, and it is what separates a de-identification claim from a documented de-identification process. Settle which standard applies, Safe Harbor or Expert Determination, before the reviewer starts. The two lead to different review decisions. WHAT THE VIDEO SHOWS - Creating a cohort with Assistant from a plain-language query - Refining patient-selection criteria - Building a clinical-document dataset from the cohort - Running the de-identification job - Reviewing each detected identifier and approving or correcting it - Validating the de-identified outputs - Downloading the reviewed files WHO THIS IS FOR Anyone who has to move clinical text somewhere it cannot carry PHI: - Research data offices, honest brokers, and data stewards - Privacy officers and compliance teams approving a release - AI and ML teams assembling training or evaluation corpora - Analytics teams moving notes into a non-clinical enclave or a lower environment - Study teams sharing notes with an academic or commercial collaborator The destination changes the review, not the workflow. A model training corpus, a statistician's working set, a vendor evaluation sample, and a partner deliverable under a data use agreement all run the same job. What differs is the standard you review against. QUESTIONS THIS VIDEO ANSWERS - How do I de-identify clinical notes before sharing them with a collaborator? - How do I document that a reviewer approved each PHI detection? LINKS Tutorial: https://www.johnsnowlabs.com/pji/tutorials/scenarios/assistant-cohort-to-dataset-to-deidentification De-identification documentation: https://www.johnsnowlabs.com/pji/data_enrichment/de-identification Patient Journey Intelligence: https://www.johnsnowlabs.com/patient-journey-intelligence/ All patient data shown is synthetically generated for illustration. #DeIdentification #ClinicalNLP #DataPrivacy