John Snow Labs

John Snow Labs Helping healthcare and life science organizations put AI to work faster with state-of-the-art LLM & NLP.

John Snow Labs, an AI and NLP for healthcare company, provides state-of-the-art software, models, and data to help healthcare and life science organizations build, deploy, and operate AI projects. John Snow Labs, the AI for healthcare company, provides state-of-the-art software, language models, and data to help healthcare and life science organizations build, deploy, and operate AI, LLM, and NLP projects faster.

The most common question we hear from health system data teams is not about model accuracy. It is: how do we connect the...
05/03/2026

The most common question we hear from health system data teams is not about model accuracy. It is: how do we connect the data we already have?

Clinical truth is distributed. It lives in Epic flowsheets, scanned faxes from referring physicians, pathology PDFs, and social history free-text blocks that no structured field captures.

Patient Journey Intelligence was built to address this specific problem, ingesting, enriching, and harmonizing multimodal clinical data into a unified OMOP-based longitudinal record that teams can actually use for analytics, research, and agentic workflows.

If your organization is still treating fragmentation as a pre-processing problem, it is worth reconsidering the architecture.

Explore the platform: https://hubs.ly/Q04fbz_x0

Most de-identification tools are built for structured fields. They fail when PHI is buried inside a social history note,...
05/02/2026

Most de-identification tools are built for structured fields. They fail when PHI is buried inside a social history note, a discharge narrative, or a handwritten addendum.

Healthcare NLP handles de-identification across all document types, clinical notes, scanned PDFs, DICOM metadata, and tabular data, using context-aware models that distinguish a patient name from a physician name, and a date of birth from a procedure date.

For organizations building secondary-use data programs, the depth of de-identification determines the safety of the pipeline.

Review the methodology: https://hubs.ly/Q04fbNMd0

Medical document processing fails for the same reason every time: the tool reads characters but loses structure. A lab ...
05/02/2026

Medical document processing fails for the same reason every time: the tool reads characters but loses structure.

A lab result table in a pathology report contains information in its geometry, which biomarker belongs to which result, which value is flagged as abnormal. Standard OCR flattens that structure into a string and discards the relationships.

Visual NLP preserves document geometry. It parses nested tables, multi-column layouts, and mixed text-image formats while maintaining the clinical meaning embedded in the original layout.

For teams processing discharge summaries, pharmacy charts, and radiology reports at scale, the starting point is how the AI sees the document.

See it in action: https://hubs.ly/Q04fbyJc0

Clinical trial matching is one of the most resource-intensive tasks in oncology research. Identifying eligible patients ...
05/02/2026

Clinical trial matching is one of the most resource-intensive tasks in oncology research. Identifying eligible patients requires reading thousands of charts, cross-referencing complex inclusion and exclusion criteria, and doing it at a speed that the trial timeline will not wait for.

Patient Journey Intelligence automates this process by building a structured longitudinal patient view, linking diagnoses, biomarkers, medications, and staging data from clinical notes and EHR records, then matching against protocol criteria with agentic AI workflows.

Organizations are reducing screening time from weeks to hours.

Learn how it works: https://hubs.ly/Q04fbFSh0

ICD-10 has over 70,000 codes. SNOMED CT contains more than 350,000 active concepts. RxNorm maps tens of thousands of dru...
05/02/2026

ICD-10 has over 70,000 codes. SNOMED CT contains more than 350,000 active concepts. RxNorm maps tens of thousands of drug synonyms.

No human coder can maintain consistent precision across all three standards at scale.

Healthcare NLP resolves clinical entities to standardized terminologies automatically, linking a free-text mention of "metformin 500mg" to its RxNorm concept, or "type 2 diabetes" to its ICD-10 and SNOMED equivalents in a single pipeline.

Explore medical terminology mapping: https://hubs.li/Q04fbG8P0

If a clinician cannot trace an AI-generated fact back to the source sentence in the patient record, that AI is a liabili...
04/30/2026

If a clinician cannot trace an AI-generated fact back to the source sentence in the patient record, that AI is a liability, not an asset. This is why "Black Box" models fail in high-stakes healthcare. The Generative AI Lab is built on the principle of provenance, providing full transparency into how data points are extracted. By giving clinicians, a no-code environment to verify and tune models, we ensure the person responsible for the patient is the one validating the technology.

Verify the output: https://hubs.li/Q04d-Kd_0

The paradox of modern medicine: we have AI-guided surgery, yet the clinical pipeline still depends on faxed documents an...
04/30/2026

The paradox of modern medicine: we have AI-guided surgery, yet the clinical pipeline still depends on faxed documents and scanned PDFs. Generic OCR tools fail because they lose the document’s geometry, the relationship between a column header and a lab result. Visual NLP preserves this structure, ensuring that when you extract data from a complex pathology report, the clinical meaning stays intact. Accuracy in a production environment starts with how the AI "sees" the document, not just how it reads the characters.

See the document clearly: https://hubs.li/Q04d-LqQ0

A single medical note is a snapshot; a patient’s life is a trajectory. The primary challenge in clinical research today ...
04/30/2026

A single medical note is a snapshot; a patient’s life is a trajectory. The primary challenge in clinical research today is fragmentation; the truth is scattered across unstructured narrative notes, scanned faxes, and siloed EHR fields. Patient Journey Intelligence solves this by connecting these disparate signals into a unified longitudinal view. This isn't about better search; it’s about understanding the clinical history of a patient to improve future care outcomes. If you aren't connecting the dots across time, you aren't seeing the full patient.

Connect the trajectory: https://hubs.li/Q04d-Jsl0

It’s today! Join our next webinar to see how AI agents become practical tools for: ✔ Cohort identification and populatio...
04/29/2026

It’s today!

Join our next webinar to see how AI agents become practical tools for:

✔ Cohort identification and population analysis
✔ Patient-level investigation across care settings
✔ Clinical trial matching
✔ Secondary use of clinical data with governance built in

We’ll demonstrate how a unified, OMOP-based foundation enables analytics, research, and agentic healthcare workflows while maintaining data quality and regulatory readiness.

Presented by Veysel Kocaman, PhD, CTO at John Snow Labs.
📅 April 29, 2026
🕑 2:00 PM ET

👉 Register now to explore Patient Journey Intelligence in action: https://hubs.li/Q04dWgTq0

In healthcare, "trust but verify" is the rule. Our models aren't just high-performing on internal tests; they are valida...
04/27/2026

In healthcare, "trust but verify" is the rule. Our models aren't just high-performing on internal tests; they are validated in peer-reviewed journals like JAMIA. When we claim 99% accuracy in de-identification or clinical extraction, it's backed by rigorous, independent evaluation.

See the papers: https://hubs.li/Q04dsrC-0

Most healthcare AI pilots fail because they focus on the model instead of the workflow integration. To reach production,...
04/27/2026

Most healthcare AI pilots fail because they focus on the model instead of the workflow integration. To reach production, you must solve for the "last mile", how the AI interacts with the existing EHR and clinical staff. We focus on industrial-grade tools that fit into the high-stakes hospital environment, not just the lab.

Implementation Strategy: https://hubs.li/Q04dsnL10

The people who understand clinical "truth" are often the ones who don't code. Our Generative AI Lab provides a no-code e...
04/26/2026

The people who understand clinical "truth" are often the ones who don't code. Our Generative AI Lab provides a no-code environment where clinicians and subject matter experts can tune NLP pipelines and verify outputs themselves. We are removing the technical barrier so that medical experts can stay in the loop.

No-Code Clinical AI: https://hubs.li/Q04dsb530

Address

16192 Coastal Highway Lewes, DE , USA
Delaware City, DE
19958

Telephone

+13027865227

Website

Alerts

Be the first to know and let us send you an email when John Snow Labs posts news and promotions. Your email address will not be used for any other purpose, and you can unsubscribe at any time.

Contact The Practice

Send a message to John Snow Labs:

Shortcuts

Share