.jpg "Cursor Launches a New AI Agent Experience to Take On Claude Code and Codex")
.jpg "Anthropic Says That Claude Contains Its Own Kind of Emotions")
Stroke is one of the leading causes of long-term disability, with roughly two-thirds of survivors experiencing significant impairments in their hands and arms. While some people eventually regain that function, many live with persistent paralysis or weakness. Epia Neuro, a newly launched startup out of San Francisco, wants to help more stroke patients regain hand function with a brain implant and motorized glove.
It’s among a growing number of companies developing brain-computer interfaces, devices that read neural signals from the brain and translate them into specific actions. The space has seen a huge influx of investment in recent years, with Elon Musk’s Neuralink raising $500 million last year and Sam Altman’s Merge Labs emerging from stealth in January with $252 million in funding.
Neuralink and others are building devices that give people with severe motor disabilities the ability to control a computer or speak with a digital voice. Epia’s technology aims to help people move their own hands again.
“These patients have very weak grip. It’s a very common problem,” says Michel Maharbiz, Epia’s CEO and a professor of electrical engineering and computer science at UC Berkeley. “If you could just give them the grip back reliably, an enormous number of things would open up in their daily life.”
Improvements in hand function can mean the difference between being able to dress or eat independently and relying on constant care.
Epia’s disk-shaped implant is inserted in the skull and detects brain signals associated with a person’s intent to move their hand. The implant will be used alongside a grip-assist motorized glove that patients will wear during rehab or at home. Neural signals are translated by AI algorithms and combined with data from external sensors on the glove to predict and drive gripping motion. The system learns to associate certain brain signals and contextual data with a person’s desire to open and close their hand.
The device relies on the idea of neuroplasticity, the brain’s ability to change and make new connections. During a stroke, blood flow to part of the brain is interrupted, starving cells of oxygen and damaging precious tissue. Damage to the motor area of the brain can cause paralysis and muscle weakness. When a person with paralysis tries to move, their brain still generates signals related to movement, but the injury means these signals are blocked from getting to the muscles. Epia’s implant collects neural signals from an uninjured part of the brain, determines an intent to move, and converts this intent to hand movement through the glove.
“We can train the system to learn the user’s intent with regards to the function they’re trying to compensate for,” Maharbiz says.
Repeated use of the system could strengthen the neural pathways associated with movement, reducing a person’s reliance on the glove.
“A lot of brain-computer interfaces allow a person to type on a computer screen or to move a robotic arm to achieve a task,” says David Lin, a critical care neurologist and director of the Neuro-recovery Clinic at Massachusetts General Hospital who is advising the company. “That’s different from a rehabilitative solution, where using that device in and of itself leads to plasticity of the brain, or changing of the brain and the connections to the spinal cord, so that once you take the glove away, that the native function of the arm and hand gets better.”
One of the hurdles for brain-computer interfaces is scalability. These devices will need to be relatively easy to implant with few risks for people to want to get them. Neuralink is trying to get around this problem by developing a robot to insert its BCI. Another company, Synchron, has a stent-like implant that’s inserted into the blood vessel instead of requiring brain surgery.
