Algorithm-based trunk support system shows potential for recovery of walking ability


A multidirectional gravity-assist device that delivers precise trunk support to stroke and spinal cord injury (SCI) patients via an artificial intelligence algorithm has demonstrated significantly improved locomotor performance beyond treadmill-based systems in a new study.

The harness device used in the study adjusts patients’ balance while they stand still or walk by employing a unique, adaptive multidirectional gravity-assist (MGA) algorithm tailored to the specific needs of the patient, according to Jean-Baptiste Mignardot, PhD , of the Center for Neuroprosthetics and Brain Mind Institute, Swiss Federal Institute of Technology, Lausanne, Switzerland, and fellow investigators.

“The MGA establishes a safe and natural rehabilitation environment wherein individuals with neurological deficits can perform basic and skilled locomotor activities that would not be possible without robotic assistance,” according to the investigators. “The immediate and short-term ameliorations of gait performance during locomotion with MGA illustrate the potential of this environment to augment motor recovery.”

Current gait rehabilitation methods in stroke or SCI patients most commonly involves counterweight mechanisms or force-controlled equipment that apply upward support while walking on a treadmill. However Dr. Mignardot and his colleagues believe these methods are flawed.

“Treadmill-restricted environments markedly differ from the rich repertoire of natural locomotor activities underlying daily living,” the investigators wrote. “Vertically restricted trunk support creates undesired forces that impede gait execution.”

To counteract these negative effects, the MGA adjusts both upward and forward forces on the patient’s body, re-creating a more naturally occurring gait posture, which investigators have likened to an inverted pendulum with a natural forward tilt. In order to create the algorithm, investigators ran through a series of procedures, starting with calibrations based on the gait of healthy subjects and adjusting for necessary upward and forward assistance for stroke and SCI patients.

The artificial neural network within the algorithm analyzes patients’ support needs, a job that therapists currently have to do based on visual observations. This opens a window to faster and more accurate estimations, according to the investigators ( Sci Transl Med. 2017 Jul 19. doi: 10.1126/scitranslmed.aah3621 ).

Investigators tested the algorithm on 15 SCI patients and 12 stroke patients. The stroke patients had an average age of 51 years, with length of time after stroke varying from 8 to 235 months. The SCI patients had an average age of 47 years, with a length of time since injury ranging from 12 to 264 months. Most patients in both groups were male.

When tested, the algorithm showed varying success depending on the severity of the injury, according to the researchers.

“For example, the MGA enabled subjects who could not stand independently to walk overground with or without assistive device.” Subjects who were able to move around only with crutches or a walker progressed without the use of assistive devices and exhibited improved spatiotemporal gait features, according to Dr. Mignardot and fellow investigators. “Individuals with stroke exhibited similar or even superior amelioration of locomotor performance and showed that individuals who could only walk with crutches exhibited improved intralimb coordination.”

After initial efficacy tests, the researchers tested the MGA’s effectiveness in five SCI patients immediately after 1 hour of training with the device and found that their gait speed increased during the training. However, the improvements were not evident in a similar test 1 week later. Similar tests using treadmill-restricted step training without the MGA device did not show any improvement during either week of testing.

Although the study’s small sample size limited the conclusions that could be reached, the investigators were encouraged by the overall effects of the algorithm. They noted that further tests are required to test the potential sensitivity and accuracy of the software.

The study was supported by the European Commission’s Seventh Framework Programme, various foundations, and the Swiss National Science Foundation.

Investigators reported holding patents on the step-by-step procedure and use of the MGA algorithm in this study.

On Twitter @eaztweets