TUM's soft glove let a man with ALS hold a fork, controlled by one thumb muscle

At a glance

Status: research prototype; the paper lists safety certification and regulatory steps as still ahead.

Skip it if: you want something you can buy or a controlled trial in people with near-complete paralysis.

Wearing the final version of a soft glove from the Technical University of Munich (TUM), a 65-year-old man with ALS held a fork for the first time in four years, TUM says. His right hand couldn't move any finger apart from small thumb-tip movements; with the glove, a Nature Machine Intelligence paper published in June 2026 reports, he moved up to five blocks in 60 seconds on the Box and Blocks Test, with minimal assistance, and fed himself. The first prototype scored 0.[1] [2]

Researcher John Nassour sewed the glove himself, TUM says, from ordinary clothing fabric; it weighs 352 g and moves fingers, thumb and wrist with 13 textile air actuators. Plain finger bending wasn't enough for this patient, so the team added thumb opposition, thumb abduction and wrist extension, letting the hand position itself before it closes. Pressure stays below 2 bar, and the compressor and valves sit in a 2 kg control box that can mount on a wheelchair.[1] [3] [2]

Knowing when to grip is the harder job. One sensor sits over the flexor pollicis longus, the only muscle in that hand he could still control voluntarily, and its signal was noisier and weaker than in 15 healthy participants. With an adaptive threshold backed by a learned error filter, the system still detected 97.1% of his cued grasp attempts in a game, and motion sensors keep the grip closed while an object travels. He preferred muscle control to a button; study leader Gordon Cheng told Tech Xplore he insisted on instructing the hand himself.[1] [3] [2]

Six people in early stroke rehabilitation tried the glove, grouped by unassisted Box and Blocks score. The three who couldn't lift a single block gained most: 10 points on average on grasp tasks and 4.67 on grip tasks of a modified Action Research Arm Test, and the abstract totals their gain at 17. Of the three moderately impaired participants, two gained one point each by lifting a heavy cube, while pinch scores fell, an average drop of 9 in the abstract. The authors call that a mismatch between device design and impairment level.[1] [3]

Two earlier studies from other groups used similar gloves with simpler commands. In Cappello and colleagues' 2018 Harvard study of nine people with C4-C7 spinal cord injuries, scores rose by about a third of the maximum while a researcher ran the controls. In Lim and colleagues' 2023 case series of eight chronic stroke patients, who pressed buttons with their unaffected hand, open-hand task scores climbed from a mean of 8.13 to 18.38 out of 21, and the participants with the lowest unassisted scores gained most.[4] [5]

The paper is free to read, with supplementary videos of the fork task. The glove is a research prototype tailored to one co-designer and checked on six stroke patients with their arms manually supported; the paper lists safety certification and regulatory steps as still ahead. TUM calls the fabric cheap, and the authors' to-do list is a lighter air unit, automated calibration and larger controlled studies. The result to wait for is a controlled trial in people with near-complete paralysis.[1] [2]