John Snow Labs
HCC coding #ai requires more than extracting diagnosis mentions from clinical text. A valid coding recommendation must be supported by current documentation, MEAT evidence, ICD/HCC mapping logic, p...
Manual chart abstraction is the bottleneck for real-world evidence. Specialty-specific point tools fragment the work across cardiology, oncology, neurology, and rare disease teams, each with its ow...
Jahnavi Kachhia presents “Agentic AI for Intelligent Patient Call Triage” — a multi-modal, multi-agent healthcare framework designed to reduce clinician overload, improve patient prioritisation, an...
In healthcare, an AI hallucination is not a UX problem. It is a clinical liability.
Standard RAG architectures fail when semantic similarity is mistaken for medical accuracy.
Dippu Kumar Singh (Le...
Medical education is shifting from multiple-choice recall toward reasoning, communication, and real-world clinical judgement.
Janet Mee (Measurement Scientist at NBME) demonstrates how John Snow L...
Deploying one healthcare model is difficult. Deploying 2,000+ without operational fragmentation is an infrastructure problem, not a modelling problem.
Eric Hixson (VP of Data Science & Methodolog
A de-identification model can achieve 97% F1 and still expose thousands of patients to re-identification risk.
HIPAA does not evaluate token accuracy. It evaluates whether a real person can still b...
9 million biomedical samples already exist. Most researchers still cannot meaningfully use them.
AI is shifting public research data from static archives into searchable discovery engines.
Arjun K...
AI governance is becoming fragmented while AI systems are becoming global.
The risk is no longer model capability alone, but inconsistent rules, accountability gaps, and ungoverned deployment.
Dur...
Financial fraud detection and healthcare AI share the same core challenge: rare events in noisy, multimodal data where missing a positive case is costly.
Timestamps:
00:00 The “accuracy illusion” ...
Faster models don’t save patients. Faster decisions do.
In healthcare, communication latency is often the real failure point, not model accuracy.
Akhilesh Bollam (Software Development Engineer at ...
Clinical AI fails where context matters most: across encounters, modalities, and time.
Vector search retrieves fragments. Clinical reasoning requires relationships.
Join Krishnendu Dasgupta (Found...
Clinical trials optimise for control. Real patients do not behave that way.
The real signal of treatment effectiveness is when patients actually switch.
Timestamps
00:00 Why clinical trials fail t...
Shipping AI without validation is still the default in most organisations. Especially in regulated industries, that approach does not scale.
Anil Pantangi (Sr Product & Tech Leader, AI & Anal
What if complex diseases could be screened before symptoms become obvious, without new tests, labs, or patient questionnaires?
This talk presents a zero-burden AI screening approach that uses exis...
Employer-sponsored insurance covers 160 million Americans, yet healthcare costs keep rising and outcomes remain uneven.
AI is one of the few levers that can improve quality, reduce waste, and clos...
You can build a great healthcare AI model — and still not be allowed to use it.
The real problem is not accuracy, but proving safety, privacy, and control in production systems.
⏱️ Key moments:
00...
Most clinical AI risks don’t come from the model — they come from how it’s used.
This talk shows how governance-first design makes AI safe, auditable, and usable in practice.
⏱️ Key moments:
00:00...
Healthcare AI doesn’t fail at the model — it fails at the last mile.
This session breaks down why even high-performing LLMs don’t translate into real clinical impact — and what actually determines ...
Clinical documentation AI is one of the biggest bottlenecks in healthcare — not because models are плохие, but because data pipelines and workflows break in production.
This session shows what act...