Paralyzed Man Walks Again Thanks to AI-Powered System

In a remarkable breakthrough announced this week, scientists have leveraged the power of AI to enable a man paralyzed from the hips down to walk again. A team of Swiss researchers shared their results in the publication Nature: twelve years after a devastating motorcycle accident that left him paralyzed, Gert-Jan Oskam is now able to stand, walk, and even climb ramps, courtesy of an innovative system that translates his thoughts into signals capable of stimulating his lower body.

An AI-powered Brain-Spine Interface

Central to this medical milestone is a Brain-Spine Interface (BSI) enhanced by AI. This system consists of cortical implants that capture the patient’s brain signals and transmits them to a local processing unit. An AI decoder, meticulously calibrated to Oskam’s unique brain signals, powers the translation of signals.

The AI decoder subsequently dispatches commands to a pulse generator installed within Oskam’s spinal cord, facilitating precise control over his leg muscles. This technology bypasses the portion of Oskam’s spine that was injured in his accident, thereby circumventing the paralysis.

In order to interpret and transmit signals, researchers designed an AI decoder composed of multiple models. First, a gating model is designed to estimate the probability of moving a particular joint. Additionally, an independent multi-linear model predicts the amplitude and direction of a specific movement. To ensure the stability and robustness of these predictions, a hidden Markov model oversees the overall operation. Oskam underwent several rounds of calibration in order to improve the accuracy of the model as researchers fine-tuned it to his specific brain signals.

Impressive Results and Unexpected Neurological Recovery

The Brain-Spine Interface demonstrated notable efficiency, achieving a latency as low as 1.1 seconds. Over time, Oskam progressively acquired control over each lower limb joint, with an average accuracy rate of 74%. Most significantly, the BSI helped Oskam perform day-to-day tasks such as transitioning from walking to stopping, climbing steep ramps, and even ascending stairs.

The BSI's impact extended beyond mere physical assistance. It fostered remarkable neurological recovery, which came as a surprise to researchers. Although Oskam had previously participated in a separate study using an implanted interface, his rehabilitation had plateaued. This new AI-powered BSI, however, allowed Oskam to exercise an additional degree of control over his muscles, not directly stimulated by the AI interface. This advancement has significantly improved his ability to walk, carry weight, and transition poses.

Setting a New Benchmark for AI in Mobility Restoration

A number of recent studies have explored the idea of using simulation to restore motor functions, and scientists have shared other breakthroughs that have enabled paralyzed subjects to regain hand movement, ride bicycles and more. Recent studies leveraging generative AI have also demonstrated the power of increasingly powerful AI decoders to read human thoughts with high accuracy.

However, this latest study, with its innovative use of AI to interpret and transmit signals, stands out. The breadth of actions it can support and the high accuracy with which it operates are especially notable, and the technology has functioned as more than a limited science experiment. For the past year, it has allowed Oskam to regain the ability to live a full and varied life.

Oskam, expressing his gratitude, shared with reporters, “For 12 years, I’ve been trying to get back on my feet. Now, I have learned how to walk again in a way that feels normal and natural.”