探花视频

John Snow Labs logo


  • Finding the Patients Your Data is Missing
  • Clinical De-identification at Scale: Removing PHI from Text, PDFs, and Medical Images with Regulatory-Grade Accuracy
  • Lessons Learned from Harnessing Healthcare Generative AI in a Hybrid Multi-Cloud Environment
  • Unlocking Clinical Data as an Asset
  • Healthcare NLP: State-of-the-art Medical Language Models
  • slide
  • The Free No-Code Platform NLP Lab
  • slide
  • slide
  • Generative AI in Healthcare
  • Medical Chatbot
  • slide

Regulatory-Grade Healthcare AI. Deployed Inside Your Environment.

About John Snow Labs

builds healthcare AI that meets the accuracy, privacy, and compliance standards clinical and government environments demand, and publishes the benchmarks to prove it.

Our software is used by more than 500 healthcare organizations worldwide, including Providence Health, Mayo Clinic, the US Department of Veterans Affairs, UCLA Health, and the Department of Defense. It processes clinical notes, DICOM images, FHIR resources, and structured EHR data with no external data sharing - everything runs inside your infrastructure.

Why organizations choose John Snow Labs:

  • #1 on 12 of 13 clinical AI benchmarks - outperforming GPT-5.4, Gemini, and Claude across medical NLP tasks
  • 99%+ de-identification accuracy - validated at 2 billion patient notes, with zero re-identifications in adversarial red team testing
  • On-premises deployment - HIPAA, GDPR, and FedRAMP-compatible; data never leaves your environment
  • 3,000+ pretrained clinical models - covering entity recognition, relation extraction, coding, summarization, and more
  • 30+ peer-reviewed publications - in Nature Machine Intelligence, JMIR, ACL, AAAI, and AMIA
  • AI governance certified - conforms to 200+ laws, regulations, and industry standards worldwide

John Snow Labs software is available through 探花视频 and its reseller partners on GSA, NASA SEWP V, ITES-SW2, and NCPA contract vehicles.

Peer-Reviewed Research and Benchmarks

John Snow Labs publishes its accuracy claims as peer-reviewed research. Key published results:

De-identification

  • 96% F1-score for PHI detection vs. 91% for Azure, 83% for AWS, 79% for GPT-4o-  independently evaluated on 48 clinical documents annotated by medical experts (Text2Story Workshop, ECIR 2025)
  • First system externally validated at 2 billion patient notes with zero re-identifications in adversarial red team testing; surpasses triple manual review by domain experts - in collaboration with Providence Health (ResearchGate, 2024)
  • 50% fewer errors than AWS Medical Comprehend, 475% fewer than Azure, 575% fewer than Google Cloud Healthcare API (ML4H 2023)
  • 100% success rates for DICOM tag retention, date shifting, and UID consistency; 99.9%+ text processing accuracy on medical images - evaluated on the MIDI-B dataset (ICHI 2026)

Clinical NLP

  • SOTA on 7 of 8 major biomedical NER benchmarks without BERT-style contextual embeddings (Software Impacts, Elsevier, 2022)
  • Outperforms AWS Comprehend Medical by 8.9% and Google Cloud Healthcare API by 6.7% on clinical NER (Software Impacts, Elsevier, 2022)
  • NER precision up to 0.989 for clinical procedures; assertion model accuracy 0.889 - validated across 138,250 clinical notes from 5,000 patients at Mount Sinai Hospital (JMIR AI, 2025)
  • Assertion detection accuracy 0.962 vs. GPT-4o 0.901; 23.4% improvement on hypothetical assertions (Text2Story Workshop, ECIR 2025)
  • New SOTA F1 scores on 5 of 7 relation extraction benchmarks (Software Impacts, Elsevier, 2022)

Medical LLMs

  • Domain-specific 8B and 70B models outperform GPT-4o by 45% to 92% on factuality, clinical relevance, and conciseness - blind evaluation by practicing medical doctors (CLEVER, JMIR AI, 2025)
  • Red-team safety evaluation: 690 clinically grounded scenarios across 9 domains and 150+ subcategories; top performers scored above 0.97 with low variance (ECIR 2026)

Clinical Data Extraction

  • Cancer registry abstraction time reduced from 120 minutes to under 2 minutes per case at regulatory-grade accuracy - Winner, RWE Catalyst Challenge (PHUSE US Connect 2026)
  • Adding unstructured NLP data to structured EHR doubled suicidality and self-harm event detection across 17 million clinical notes from 109,076 patients - events invisible in structured fields alone (PHUSE/FDA Computational Science Symposium, 2025)
  • HPO code extraction: nearly 100% coverage vs. below 60% for GPT-4; 100 documents processed in ~15 seconds on CPU vs. ~1,000 seconds for API-based LLMs (King's Health Partners / Wilson Aarhus Symposium, 2026)
  • SDoH-NER outperforms GPT-4 in precision for homelessness/housing and marital status detection (Artificial Intelligence in Health, 2024)

Published in: Nature Machine Intelligence, JMIR AI, ACL, AAAI, AMIA, Software Impacts (Elsevier), Pharmacoepidemiology & Drug Safety, Journal of Personalized Medicine, and more.

See all peer-reviewed papers !

Awards and Recognition

John Snow Labs has received since 2017, including:

2026

  • Frost & Sullivan Customer Value Leadership Award — Medical LLMs
  • Winner, RWE Catalyst Challenge — PHUSE US Connect 2026

2025

  • ACQ5 Global Award — Best LLM of the Year (Healthcare)
  • Corporate America Today — Best Medical Application of LLMs
  • Global GenAI Awards — Best Healthcare AI Application
  • AI Breakthrough Award — Best Use of AI for Healthcare (Generative AI Lab)
  • Oracle GenAI Innovation Award

2024

  • AI Breakthrough Award — Best MLOps Company
  • Cloud Awards — Best SaaS Product; Most Innovative Use of Data; Best Open Source Solution
  • Corporate Vision — Best Generative AI in Healthcare Solution Provider
  • Global 100 — Best Medical Application of LLMs

2023

  • AI Breakthrough Award — Best Natural Language Recognition Solution
  • AI TechAward — Best NLP Technology
  • Juniper Research Platinum Award (2×)
  • Global 100 Award; Gamechangers Award (2×)