In partnership with
Welcome to AIMedily—a newsletter where I curate AI research in rehabilitation medicine to help you stay updated.
Hello!
How is your week going?
Before we jump into today’s articles. Let’s talk about how AI is impacting the health ecosystem.
That’s a huge difference! This number will keep going up.
AI improves through continuous learning from data (like you and I).
As AI evolves, how do you think AI will integrate into your practice in 5 or 10 years?
Now, let's dive into today's issue:
🤖AIBytes: Exoskeletons on sit‑to‑stand transitions, robotic rehabilitation on arm function and Surface EMG + Exoskeletons.
🦾TechTool: Do people with disabilities want robots to help them at home?
🧬AIMedily snaps: Interesting links
🧩TriviaRX: Find the new question (and the answer from last week).
🤖 AIBytes
🔬 Methods:
This pilot clinical study recruited 8 stroke patients with hemiparesis and 3 healthy controls.
Participants performed sit-to-stand transitions:
a) With a powered knee exoskeleton that assists the affected knee joint using electromiographic control (paretic vastus medialis).
b) Without assistance
They analyze: Motion capture, Force plates, Inverse dynamics and Electromiography.
📊 Results:
Stand-up time reduced by 8.8% (p<0.05)
Weight-bearing symmetry improved by +13.7% (p<0.05)
The exoskeleton supplemented the lack of strength in the affected knee. Torque increased by 59% (p<0.05).
Less effort on the affected side. Quadriceps EMG peak activation dropped ~32% (p<0.05)
Margin of stability (balance) improved significantly (p<0.05)
🔑 Key Takeaways:
Powered exoskeleton improved speed, symmetry, and strength during sit-to-stand transitions.
Fatigue reduction: EMG-timed assistance lowers muscle demand.
EMG-powered exoskeletons show potential to improve independence, balance, and mobility in stroke rehabilitation.
🔗 Gunnell AJ, Sarkisian SV, Hayes HA, Foreman KB, Gabert L, Lenzi T. Powered knee exoskeleton improves sit-to-stand transitions in stroke patients using electromyographic control. Commun Eng. 2025;4:Article 104. Published June 7, 2025. doi:10.1038/s44172-025-00440-3
🔬 Methods:
This systematic review and network meta-analysis included:
31 randomized controlled trials (1,494 stroke patients).
Participants received either:
Robot-Assisted Therapy + Conventional Rehabilitation Therapy
Conventional Rehabilitation Therapy
They evaluated:
FMA-UE (Fugl-Meyer Assessment–Upper Extremity) – measures arm motor impairment
MBI (Modified Barthel Index) – measures functional independence
MAS (Modified Ashworth Scale) – measures spasticity
WMFT (Wolf Motor Function Test) – measures timed upper‑limb tasks
FIM (Functional Independence Measure) – measures independence
Elbow extension angle
📊 Results:
Patients in the robot-assisted therapy group:
Improved their arm movement significantly more.
On average, they scored nearly 6 points higher on the Fugl-Meyer Upper Extremity Test
(Better if therapy started within the first 6 weeks).
Improved daily activities like eating, dressing, and bathing.
Scored about 8 points higher on the Modified Barthel Index.
These gains were strongest in longer programs (6+ weeks).
Had less spasticity.
Their scores on the Modified Ashworth Scale dropped by about 0.6 points.
The results were consistent across studies.
🔑Key Takeaways:
Best robot for movement recovery:
End-effector robots were most effective at improving arm and hand function.
(End-effector robots are connected to patients at one distal point; their joints do not match human joints.)
Best robot for daily independence:
Exoskeletons
🔗 Wang H, Wu X, Li Y, Yu S. Efficacy of robot-assisted training on upper limb motor function after stroke: a systematic review and network meta-analysis. Arch Rehabil Res Clin Transl. 2025;7:100387. DOI: 10.1016/j.arrct.2024.100387
🔬 Methods:
How surface electromyography (sEMG) helps robotic exoskeletons understand a patient’s movement intentions.
The authors analyzed 321 studies covering:
sEMG signals recollection.
How machine learning and deep learning interpret signals.
How do these methods work alone or alongside other tools?
Like EEG.
Control strategies for making real-time, adaptive movement possible.
📊 Results:
sEMG can detect the intention to move before the limb moves. Making the exoskeleton response faster and more intuitive.
Traditional machine learning methods were accurate.
Deep learning models were better at handling real-world noise and variability.
Combining sEMG with other signals (like EEG) improved safety and reliability.
These systems can predict motion paths and adapt to the user.
🔑 Key Takeaways:
Combining muscle (EMG) and brain (EEG) signals using deep learning helps exoskeletons understand movement intentions.
This approach improves control, especially for complex or subtle movements.
Personalized AI models will enable exoskeletons to work naturally with the body. Offering smoother and more intuitive assistance in real-world rehab.
🔗 Zhang X, Qu Y, Zhang G, Wang Z, Chen C, Xin. Review of sEMG for exoskeleton robots: motion intention recognition techniques and applications. Sensors (Basel). 2025;25(8):2448. doi:10.3390/s25082448
🦾TechTool
🔬 Methods:
This qualitative study involved 9 adults with physical disabilities (ages 27–78).
They interacted with a humanoid robot named EVE in a simulated home setting.
Participants used voice commands and evaluated both autonomous and remote-controlled functions.
📊 Results:
Most participants were eager to explore robotic assistance.
Users preferred helpfulness over humanlike features.
Participants found the robot easy to use, even when it made occasional errors.
Privacy concerns were low, similar to using smartphones or smart homes.
The robot was useful for daily tasks, not intimate care.
Has the potential to reduce caregiver burden.
🔑 Key Takeaways:
Independence matters: Users want robots to enhance autonomy.
Design should be prioritized over function
Voice control and personalization boost usability and engagement.
Robots could adapt to support each user’s home and routine.
🔗 Sørensen L, Johannesen DTS, Melkas H, Johnsen HM. JMIR Rehabil Assist Technol. 2025;12:e63641. doi: 10.2196/63641
🧬 AIMedily Snaps
🧩TriviaRX
A Roman physician documented the 1st recorded use of electrical stimulation therapy in 46 AD for treating gout and headaches.
What was his method?
A) Rubbing amber against wool to create static electricity
B) Using copper and iron rods in saltwater baths
C) Placing live torpedo fish on the affected body part
D) Creating sparks with flint and steel on skin
The answer will be in the next issue (send me your answer, I dare you not to check online)
Answer from last issue: A) Cerebellum predicts self-touch
The cerebellum dials down the tickling sensation in the somatosensory cortex when it detects that the touch is self-generated.
That's all!
I would love to know what you are working on. If you have suggestions hit reply.
If you enjoy reading AIMedily, maybe your friends will like it too! You can forward this email 📥
Until next week,
Itzel Fer, MD PM&R
Forwarded this email? Sign up here
How did you like today's newsletter?
Find out why 1M+ professionals read Superhuman AI daily.
AI won't take over the world. People who know how to use AI will.
Here's how to stay ahead with AI:
Sign up for Superhuman AI. The AI newsletter read by 1M+ pros.
Master AI tools, tutorials, and news in just 3 minutes a day.
Become 10X more productive using AI.
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.
