Predict, Prevent, Personalise
From Cancer Clues to Fall Forecasts, AI is stepping up, just don’t call it a nurse!
Handoff #5 | Reading time: 5 minutes
Good morning. Last week marked five years since the WHO declared COVID-19 a pandemic, a moment that reshaped healthcare, accelerated AI adoption, and made us all experts in sourdough baking. But while we can’t predict the next global crisis, AI is proving it can predict falls, flu, and even how long a patient might stay in hospital.
In today's handoff:
Putting Nighttime Scratching to Bed
Detecting Colorectal Cancer with Precision
AI Can’t Play Nurse: Oregon Draws the Line
AI in the NHS: From Baby Steps to Big Leaps
Stopping Falls and Flu in Their Tracks
🩺 Quick Assessment
The one story every healthcare pro needs to know this week.
🛌🏻 Putting Nighttime Scratching to Bed
Imagine you’re fast asleep, until the relentless itch of atopic dermatitis drags you into a scratching frenzy. By morning, your skin is raw, and your sleep? A total disaster.
How AI Helped?
A smart AI-powered wearable is changing the game. Using haptic feedback, it gently vibrates when it detects scratching, reducing scratch time by 28% and scratch events by 50%, all while you stay asleep.
The Science
Researchers at JAMA Dermatology trained the device to track nocturnal scratching patterns, using AI to detect when it’s time to intervene. Think of it as a tiny, wearable "stop that" coach for your skin.
The Outcome
Patients with mild atopic dermatitis experienced less skin damage, fewer nighttime wakeups, and improved sleep quality, without extra medication.
Why It Matters?
Chronic itching wrecks sleep and worsens skin conditions. This AI-driven wearable could offer a drug-free way to break the scratch cycle, helping patients heal while they rest.
🚨 Critical Updates
Fresh, impactful news on AI’s real-world applications in healthcare.
🔬 Detecting Colorectal Cancer with Precision
Researchers at the University of Jyväskylä have developed an AI-powered tool that identifies colorectal cancer from tissue samples with a staggering 96.74% accuracy. Instead of relying solely on pathologists hunched over microscopes, this AI scans digital microscopy slides to pinpoint cancerous tissue faster and more reliably than traditional methods.
So What? Faster and more accurate cancer detection means earlier treatment, better patient outcomes, and a reduced workload for histopathologists, allowing them to focus on complex cases.
🚫 AI Can’t Play Nurse: Oregon Draws the Line
Oregon’s House Bill 2748 is making waves by banning AI from using the title "nurse". Lead by State Rep. Travis Nelson, a nurse himself, the bill is a direct response to a tech company promoting an AI-powered “nurse” for $9 an hour. While AI can assist in clinical tasks, lawmakers argue it lacks the empathy, critical thinking, and human intuition that make nursing irreplaceable.
So What? This isn’t just a debate over job titles, it’s about maintaining trust, safety, and the human connection in healthcare. AI can support nurses, but replacing them? That’s a hard no from Oregon.
🏥 AI in the NHS: From Baby Steps to Big Leaps
The NHS is shifting from small-scale AI pilots to full-blown adoption, using AI to predict patient deterioration, reduce emergency admissions, and streamline admin workflows. The challenge? Seamless integration, ensuring AI actually improves patient care without overwhelming staff or adding unnecessary complexity. Successful use cases include AI-driven diagnostic tools and predictive analytics that are already reducing hospital admissions.
So What? This shift is about making AI a standard tool in NHS operations. If done right, it could mean shorter waiting times, better resource allocation, and more consistent patient care across the UK.
📋 Follow-Up Notes
Demystifying tricky AI concepts with simple, relatable explanations.
💡 Multimodal AI
The Breakdown
Most AI models process one type of data at a time, text, images, or numbers. Multimodal AI? It’s the overachiever of the AI world, capable of analysing multiple types of data simultaneously, text, images, speech, and even sensor readings.
The Analogy
Imagine trying to understand a movie by only reading the script, no visuals, no sound, no expressions. You’d miss the full picture. Multimodal AI watches the entire movie, processing the dialogue, facial expressions, background music, and even the setting, giving a richer, more complete understanding.
Why It Matters
Healthcare isn’t one-dimensional, and neither are its data sources. Multimodal AI can integrate medical imaging, patient records, genetic data, and even voice analysis to improve diagnoses, personalise treatments, and flag potential risks before they escalate.
🔍 Incidental Findings
The AI twist you didn’t see coming.
🚷 Stopping Falls and Flu in Their Tracks
Trips, slips, and surprise sniffles. What if AI could warn you before they happen? Well, the NHS might have just cracked it. A new AI tool developed by Cera is rolling out across millions of home care visits, using patient blood pressure, heart rate, temperature to predict falls and flag early signs of winter illnesses like COVID-19, flu, RSV, and norovirus. Oh, and did we mention? It predicts falls with 97% accuracy and could prevent up to 2,000 hospital admissions every day.
Why It’s Wild?
It’s like AI with a sixth sense for health hazards. One moment you’re steady, the next it’s giving you a heads-up before gravity does its thing. And the best part? It’s actually working.
The Takeaway
This is about keeping people out of hospital, freeing up NHS resources, and helping vulnerable patients stay safer at home. Fewer falls, fewer flu cases, and way more peace of mind? Sounds like a win.
📝 Rounds Recap
A quick roundup of key headlines you might’ve missed but should know.
The AI Assistant Your Doctor Always Wanted: Microsoft’s Dragon Copilot is bringing voice AI to clinical workflows, combining speech recognition with ambient listening to handle documentation and automate tasks.
A Lifeline for Learning Disabilities: A new AI model from Loughborough University can predict how long patients with learning disabilities will stay in hospital, using data from 9,600 cases. With a 76% accuracy rate, it’s helping hospitals plan better and personalise care.
Chinese Scientists Make a Breakthrough in Parkinson’s Research: Researchers have identified FAM171A2 as a key risk gene and found a promising new drug that could slow Parkinson’s progression. AI played a huge role in screening 7,000+ compounds to make this possible.
Japan’s Robo-Nurses Answer to Elderly Care: Japan’s AI-powered AIREC robot is tackling its caregiver shortage by helping with patient movement and even changing diapers.
AI Gets a Thumbs-Up from Pharma Giants: Novo Nordisk is now using AI to generate clinical study reports in just 10 minutes, a massive leap from the previous 15-week process. Faster drug approvals? Yes, please.
London’s AI Framework: OneLondon has launched an AI governance framework to ensure safe, ethical, and effective AI integration across health and care.
Heart Hero: Shanghai Zhongshan Hospital’s CardioMind AI helps cardiologists diagnose heart conditions faster and more accurately using ECGs, ultrasounds, and lab tests.
🤝 Final Handoff
Turns out AI is getting scarily good at spotting cancer, predicting hospital stays, and even keeping us from scratching ourselves raw in the middle of the night. But while AI can anticipate a bad fall, it still can’t stop you from walking into your coffee table at 2 AM. Some problems remain deeply human.
Stay curious, stay caffeinated, and see you next week!