Learn AI in 5 minutes a day
What’s the secret to staying ahead of the curve in the world of AI? Information. Luckily, you can join 1,000,000+ early adopters reading The Rundown AI — the free newsletter that makes you smarter on AI with just a 5-minute read per day.
Hi!
Welcome again to AIMedily.
This week, Microsoft presented a preprint on MAI-DxO, which simulated a panel of physicians handling complex clinical cases (304 from NEJM).
Paired with LLMs, MAI-DxO reached 80% accuracy—4× better than generalist doctors (20%).
It cut diagnostic costs by up to 70%.
This is an example of how AI tools can help clinicians improve accuracy and reduce cost.
Now, let’s dive into today’s issue.
.🤖AIBytes: Social Robots to improve physical activity. Robotic glove + Mirror therapy in Stroke. Do exoskeletons reduce therapy time?
🦾TechTool: Wearables to improve Balance.
🧬AI Medily Snaps: AI events (virtual and in person) coming. An article on LLMs from Apple. AI for diagnostics and surgery.
🧩TriviaRX: Find the new question (and the answer from last week).
🤖 AIBytes
🔬 Methods:
This systematic review analyzed 19 papers published through February 2024.
They evaluated AI-enabled social robots to promote physical activity in older adults.
Robots were used in:
Nursing homes
Rehabilitation clinics
Community centers
Private homes
Data extracted: robot type, intervention design, and outcomes related to physical activity engagement.
🤖 Robots Used:
Humanoid social robots (check these social robots Pepper, NAO): Delivered exercise prompts, conversation, and social presence
Therapeutic/companion robots (e.g., a companion robot Paro): Encouraged movement through emotional bonding and sensory feedback.
Robotic exercise coaches: Used sensors and AI (e.g., computer vision, neural networks) to monitor form and give real-time feedback.
Robots varied in form and function but shared key features: Personalization and behavioral reinforcement
📊 Results:
All studies reported improvements in exercise engagement and adherence.
Robots support routine physical activity, deliver personalized prompts, and foster social motivation.
High user satisfaction. Interactions were engaging and enjoyable.
🔑 Key Takeaways:
AI-powered social robots increase physical activity in older adults in clinical and home settings.
Personalized, emotionally engaging interactions promote long-term effectiveness.
Different robot types serve different needs
🤖Humanoids for coaching
🤖Companions for comfort
🤖Robotic coaches for precision feedback
Robots may offer health benefits in sleep, medication routines, emotional well-being, and independence.
🔗 Shen J, Yu J, Zhang H, Lindsey MA, An R. Artificial intelligence-powered social robots for promoting physical activity in older adults: A systematic review. J Sport Health Sci. 2025. doi:10.1016/j.jshs.2025.101045
🔬 Methods:
Study design: Single-blind, randomized controlled pilot trial.
Participants: 66 patients - subacute stroke (less 6 months) randomly assigned to:
Mirror Therapy (MT)
Robot-assisted Therapy (RT)
Combined Robot-assisted Mirror Therapy (RMT)
All patients also received standard rehabilitation for 4 weeks (30 min/day, 5 days/week).
Robot: Yisheng SY-HR06 pneumatic glove.
Mirror therapy: Sagittal mirror for visual feedback.
Robot-assisted Mirror Therapy integrated simultaneously.
Assessments:
Fugl-Meyer Assessment for Upper Extremity (FMA-UE)
Functional Independence Measure (FIM)
Brunnstrom Scale for Upper Limb and Hand
📊 Results:
All groups showed significant improvement across all measures after 4 weeks of intervention.
Robotic + Mirror Therapy group (RMT) achieved the greatest functional gains:
FMA-UE scores increased significantly more in the RMT group than in the Mirror Therapy group (+9.4 points, p = 0.006).
Brunnstrom scores (upper limb and hand) improved more in the RMT group than in the MT group (p = 0.003).
🔑 Key Takeaways:
Robotic glove therapy + mirror therapy enhances upper limb motor recovery in subacute stroke patients.
This combination approach outperformed the other groups on motor performance and daily activity independence (FIM).
🔗Qian J, Liang C, Liu R, et al. Combination of robot-assisted glove and mirror therapy improves upper limb motor function in subacute stroke patients: a randomized controlled pilot study. Front Neurol. 2025;16:1602896. doi:10.3389/fneur.2025.1602896
🔬 Methods
Study Design: Single-blinded randomized controlled trial
Participants: 30 neurological patients with upper limb motor deficits.
Randomized into two groups:
AGREE exoskeleton group
Conventional therapy group
Each patient received 15 sessions (45 minutes, 3×/week).
The AGREE exoskeleton is a motorized upper limb exo with 4 degrees of freedom at the shoulder and elbow.
Control modes: passive, assist-as-needed, resistance, transparent and visual feedback via LEDs.
Clinical assessments:
Fugl–Meyer Assessment (FMA)
Box and Block Test
Motricity Index
Usability: System Usability Scale (SUS)
Dosage: Total and active therapy time
📊 Results:
Clinical improvements were comparable in both groups.
The exoskeleton group achieved the same results, with less actual treatment time.
Therapists operated the AGREE system independently.
🔑 Key Takeaways:
Robot-assisted therapy needs less treatment time to achieve similar motor gains compared to conventional rehab in patients with upper limb deficits.
🔗Gandolla M, Luciani B, Longatelli V, et al. AGREE: an upper limb motorized exoskeleton for restoring arm functions: a single‑blinded randomized controlled trial. J Neuroeng Rehabil. 2025 Jun 11;22:134. doi: 10.1186/s12984‑025‑01651‑7
🦾TechTool
🔬 Methods
Design: Scoping review (2014–2024) of Smart Wearable Balance Systems + SWOT (strengths, weaknesses, opportunities, and threats) market comparison.
Patients: Older adults (≥50 years) at risk of falls.
Tools: Identified 17 systems (10 investigational, 7 commercially available).
Commercial Devices:
📊 Results
17 smart wearable systems were identified (10 investigational, 7 commercial).
Motion tracking: 8/10 investigational systems used sensors/IMU.
Visual interfaces: 9/10 investigational devices incorporated Virtual Reality/Augmented Reality.
Feedback: 3 systems provided visual or auditory feedback.
Gamification: In 8/10 systems.
Balance exercises: 7 systems targeted specific balance tasks.
Clinician supervision: 5 studies.
Clinical Report: 4 studies.
AI: 1 system employed AI personalization.
🔑 Key Takeaways
TeleRehab DSS standout is a platform that combines AI, Augmented Reality, remote monitoring, and clinician integration.
Offers a more personalized experience.
TeleRehab DSS needs to improve usability for older adults with limited tech skills or mild cognitive impairments.
Future research should focus on:
Long-term clinical validation
Simplified, user-friendly interfaces
Scalable, accessible deployment models
🔗 Nairn B, Tsakanikas V, Gordon B, Karapintzou E, Kaski D, Fotiadis DI, Bamiou DE. Smart Wearable Technologies for Balance Rehabilitation in Older Adults at Risk of Falls: Scoping Review and Comparative Analysis. JMIR Rehabil Assist Technol. 2025;12:e69589. doi:10.2196/69589
🧬AIMedily Snaps
The Illusion of Thinking: Understanding the Strengths and Limitations of Reasoning Models by Apple.
🧩TriviaRX
What’s the earliest known brain-computer interface (BCI) experiment on humans?
A) Electrode cap for epilepsy patients
B) EEG-based cursor control
C) Visual prosthesis in blind patients
D) Implanted BCI in stroke patients
(✅The answer will be in the next issue)
👉Now, the answer from last Trivia:
c) Pyrotherapy (induced fever)
Julius Wagner- Jauregg won the Nobel Prize in 1927 for his malarial treatment of neurosyphilis. He inoculated patients with malaria to induce fever.
That’s all for today! Thank you for taking the time to read.
Please send me feedback (hit Reply to this email 📧).
And before I go, I have one big favor to ask: please help me share AIMedily.
You can forward this email to your colleagues or share this link on your phone.
Thank you!
See you next week,
Itzel Fer
MD PM&R
Forwarded this email? Sign up here
How did you like today's newsletter?
Stop Asking AI Questions, and Start Building Personal AI Software.
Transform your AI skills in just 5 days through this free email course. Whatever your starting point, by Day 5 you'll be building working software without writing code.
Each day delivers actionable techniques and real-world examples straight to your inbox. No technical skills required, just knowledge you can apply immediately.
