All Episodes
The Health AI Brief — 159 episodes
Managing ‘Needle in a Haystack’ Context - Why AI Struggles with the Middle of Your Notes
Can the WHO’s AI Fix Medical Misinformation?
AI Just Beat Harvard Doctors?
Google DeepMind AI Co-Clinician Tries to Examine Patients
XML Tags for Data - How Tech Giants Structure Medical Charts for AI
The Negative Prompt Strategy for LLMs
Politeness vs Performance – Why Saying Please may Killing Your AI’s Accuracy
What Blindness is Warning Us About AI
Pre-, mid-, post-training - The Complete LLM Training Guide
Model Context Protocol (MCP) - the 'universal adaptor' for artificial intelligence
Small Language Models (SLMs) - The Lean Machine
How Bixonimania Fooled the World's Leading AI Models
Test-Time Inference - The High Cost of Thinking
Multimodal AI - Seeing the X-Ray
People Use AI Instead of Doctors (Here’s Why)
Temperature & Top-P- The Creativity Dial for Controlling the Chaos
Meta Muse Spark - New Standard for Healthcare AI?
Vector Databases - The AI's Filing Cabinet
System Prompts - The Secret Instructions How System Prompts Define AI Personality
AI Scribes in 2026: What Every Leader Needs to Know
AI Scribes Worth It? - New JAMA Study Analysis
Heidi Remote - Dedicated Hardware for Ambient Clinical AI Scribe
RAG - Ending Hallucinations and Confabulation with Retrieval Augmented Generation
Doctronic AI is Legally Prescribing Drugs (And Doctors Agree 99% of the Time)
How to Spot AI Writing
Chain-of-Thought (CoT) - Making The AI 'Think' Out Loud
Why This AI Failed to Catch 17 Fatal "Never Events" - Validating NG tube Position AI
Benchmarks - How We Grade AI
Foundation Models & Patient Privacy: Is There a Risk?
Zero-Shot vs Few-Shot - The Secret of Few-Shot Prompting
Is Perplexity Health the Future of Medical AI? The Surprises Behind the Launch
Fitbit AI Health Coach - Medical Records & Google Gemini Integration
Data Privacy & HIPAA - Is Patient Data Leaking
How ChatGPT and AlphaFold Helped Shrink a Terminal Tumour by 75%
Microsoft Copilot Health AI and The Systemic Failures Driving Us Towards Similar Medical AI
The Age of the Medical Generalist: Foundation Models in Healthcare
Google AI vs Human Doctor - AMIE AI Clinical Trial - Real-World Primary Care Results
Amazon Health AI Explained - Workflow & Medical Records
What Medicine Can Learn From Consumer AI Trends
SleepFM: The AI Foundation Model for Disease Prediction
Context Windows - The ‘Short-Term Memory’
Am I an AI? And the bot that tried to destroy a human’s career
Quantization - Shrinking the LLM 'Brain'
Scaling Neuro-Rehab: How Nyra Health’s €20M Series A Might Redefine AI Therapy
064 Parameters - Does LLM Size Actually Matter
Amazon Connect Health Agentic AI Aiming to Eliminate Administrative Burden
Why Medical AI Fails: Lessons About Model Collapse From Purely AI Synthetic Data
World Models vs LLMs for Healthcare - Master the Next Frontier According to Yann LeCun
063 Reinforcement Learning from Human Feedback (RLHF) - Human-in-the-Loop
Elon Musk's Grok Medical "Second Opinion" Suggestion
ChatGPT Health Deemed Unsafe: Nature Medicine Urgent Alert
062 Fine-Tuning - Specialist vs Generalist - Sending AI to Medical School
Health AI in 2026: The Definitive Guide to Big Tech Strategies
Heidi Acquires Automedica and Rolls Out New Products
061 Pre-Training - What LLMs Learn Before They Become 'Medical'
Deep Dive into Ant Afu Health AI App - Affiliate of Jack Ma’s Alibaba
060 Positional Encoding - How AI Knows the Order of Your Symptoms
059 BERT vs. GPT - Understanding the Difference Between an Encoder and Decoder
Agentic AI Era in Healthcare: Lessons from OpenClaw
058 Regular Expressions (Regex) - The ‘Old School’ Trick That (Sometimes) Still Beats Modern AI
AI scribe transcripts - Digital Exhaust or Digital Gold
Are LLMs Reliable for Medical Advice? Nature Medicine Study
Nature medicine principles to guide AI readiness
International AI Safety Report 2026
Large Language Models for Complex Cardiology? Google AMIE/Stanford Study Review
057 Named Entity Recognition (NER) - How AI Spots Drugs and Dosages Instantly
Lotus Health AI - AI Aimed at Primary Care
AI Stethoscope TRICORDER Trial Results - Why Good AI Doesn't Guarantee Clinical Impact
056 TF-IDF - How to Find a Needle in a Digital Haystack
AI For Mammography Screening - Results of Lancet RCT
055 N-grams - Why Blood Pressure is One Word, Not Two
OpenEvidence: The $12B AI Changing Clinical Decisions
AlphaGenome Explained - Decoding the Non-Coding Genome
054 Stop Words - The ‘Invisible’ Words That Are Troubling Your AI Models
Sued for using AI scribe
053 Stemming & Lemmatization - How AI Groups Healing and Healed Automatically
Google MedGemma 1.5 and MedASR - 3D Scans & Medical Speech
052 Tokenization - How AI Slices Symptoms into Tokens
Amazon's New 'Agentic' Health AI
Gates and OpenAI Deploying AI in Rwanda
How to Use Ambient AI Scribes - The Ultimate Clinical Guide to Stop Editing AI Notes
051 Natural Language Processing (NLP) - teaching a computer to read a chart
NVIDIA’s SurgWorld - Solving the Surgical Robotics Data Gap with AI Synthesis
050 Diffusion Models - Sculpting from Noise
049 The Transformer - The Architecture Behind the Revolution
Google AI Overviews vs Clinical Safety - Why Reputable Sources Aren't Always Enough for AI
Claude for Healthcare - The AI Arms Race Hits the Clinic
048 Attention - How AI Focuses on What Matters
ChatGPT Health - Consolidating Health Data - Diagnosis or Data-Mine
047 Embeddings - The Medical Dictionary for AI
AI vs Pancreatic Cancer - Can PANDA Solve the Early Detection Problem
The Wealth of Health - Avoiding an Inequality Spiral in the AI Age
046 Generative Adversarial Networks - The Forger and the Detective
045 LSTMs - Fixing the AI's Short-Term Memory
Hugging a PowerPoint - Why ChatGPT makes a Terrible (but Addictive) Therapist
The 'Duck Test' for AI And Why Your Chatbot Might Be an Illegal Medical Device
044 RNNs - AI With a Memory for Sequences
043 CNNs - How AI See Medical Images
042 Activation Functions - To Fire or Not To Fire
041 Neurons & Weights - The Synapse of AI
040 K-Means - The Automated Triage Nurse
Beyond the Benchmark - How do we test a 'superhuman' doctor
039 SVMs - Drawing the Clearest Line Between Sick and Healthy
The Verdict on AI Scribes - Analysing their First Major RCTs
038 Random Forests - The Wisdom of the MDT
037 Decision Trees - The Algorithms You Already Use
036 Regression - The Statistical Bedrock of AI
035 The Architecture Shift - From Mechanics to Models
The WHO AI Report and Why Governance is Stalling Health AI
034 Regularisation - the antidote to overfitting
033 The Bias-Variance Tradeoff - The Specialist vs The Generalist
032 Underfitting - the overly simplistic intern
031 Overfitting - the brittle genius
What DeepMind's AI That Mastered Video Games Taught Us for Building Better Health AI
030 Actor-Critic Partnership - The Best of Both Worlds
029 Policy-Based Methods - Learning How To Act Directly
When the Chatbot is More Humane Than the Doctor
028 Value-Based Methods and Their Limits - The World of Q-learning
Epic's New AI - A Crystal Ball for Medicine, or a Look in the Rear-View Mirror
027 Curriculum learning - teaching an agent step-by-step
026 Exploration-Exploitation Dilemma
AI That Speaks the Language of the Cell and Contributs to Biomedical Discoveries
025 Reinforcement learning - rewards and punishments
Charting a Course for Safe AI in Medicine to Prevent Us Flying Blind - Report from the JAMA Summit on Artificial Intelligence
024 Efficient Learning and Smart Shortcuts for Medical AI - Transfer learning and Active learning
023 Self- and Semi-Supervised Learning - Learning from Imperfect Data
An AI Doctor Takes on the NEJM: Dr CaBot, the CPC-Bench AI and the Dawn of the Diagnostic Co-Pilot
022 Unsupervised learning - finding the patterns we might not have seen
021 Supervised learning - like learning with flashcards
AI Caught 'Cheating' Its Medical Exams - New Research Paper from Microsoft
020 Hyperparameters - the AI's recipe
019 Learning rates and Gradient descent - finding the bottom of the valley
The UK's New Health AI MHRA Commission - Rewriting the Rulebook or More Red Tape?
018 The AI's Scorecard - What is a Loss Function
017 How models actually learn
Forecasting Health with AI - A Deep Dive into the Delphi-2M AI Transformer Model for Health Records
AI in the NHS: A Reality Check on a National Rollout
016 DICOM, HL7, FHIR: The Languages of Medical Data Exchange.
015 - Training data vs validation data vs test data
014 Data annotation and labelling
AI stethoscopes research trial - press release vs peer review
013 Data cleaning and preprocessing
012 Data Quality - Junk in, junk out
011 Structured vs. Unstructured Data
AI robot surgeon that corrects its own mistakes
010 The AI Tipping Point in Medicine - Why Now?
Endoscopist deskilling risk after exposure to artificial intelligence in colonoscopy - Does 'helpful' tech actually make us worse?
009 AI Agents: Automating Healthcare or a New Clinical Safety Risk?
008 Narrow vs General AI, AGI and ASI
OpenAI's New Release: Why 'Open-Weights' Isn't 'Open-Source' - And Why They're Both Relevant For Clinicians
007 Parameters
006 Deep learning
Flok Health - an AI Physio for Back Pain - A cure for health service waiting lists? But is it safe?
005 Neural networks
004 Models
DD - OpenAI and Penda Health's AI Consult Real-World Study
003 AI vs machine learning
002 What is AI?
001 Introduction