One night in 1999, a rash of frightening sensations hit Paul Titus all at once.

His left arm went numb. His left eye began twitching. He couldn't speak without slurring.

Unaware what the symptoms meant, he was slow to call for help. When his ischemic stroke was finally over, he was paralyzed on his left side and for 14 years he needed a leg brace and cane just to stay upright.

One morning last week, Titus smiled as he loped along on a treadmill in a makeshift gym. A high-tech, brace-like device wrapped his left ankle, monitoring his gait 200 times per second and supplying energy boosts as needed.

"I think I'm [finally] getting close to getting rid of my cane," he said, huffing as he went.

Titus, 47, of Middle River, is a participant in a two-year Veterans Health Administration study of the ankle machine — an $80,000 rehabilitative robot known as the Anklebot — and of how much it helps stroke survivors improve their gait even years after a disabling attack.

"It's too early to draw scientific conclusions, but, anecdotally, many participants are saying [the Anklebot] is a big help," said Larry Forrester, a rehabilitation scientist with the VA Maryland Health Care System and associate professor at the University of Maryland School of Medicine.

Forrester and Anindo Roy, a robotics engineer with the Baltimore Veterans Affairs Medical Center, are co-directors of the study, which will compare how participants fare while wearing the Anklebot on a treadmill with others who wear it in a seated position as they operate a foot-controlled video game.

A $199,000 pilot grant from the Veterans Health Administration is funding the project, which is scheduled to last through next summer. Twenty volunteers have taken part so far.

According to the National Stroke Association, a stroke occurs when a clot or rupture cuts off blood flow to the brain. This causes the death of cells in the brain. If the region happens to control a particular body function, that function can be permanently damaged.

Until about 20 years ago, according to Forrester, conventional wisdom in medical science held that stroke survivors who have persistent disabilities have about six months in which to improve. After that, the brain's neural pathways were thought to be more or less permanently re-established, and the best stroke survivors could hope for was to use exercise to maintain good overall health.

"Rehabilitation didn't fix you," Forrester said.

Starting in the early 1990s, though, researchers across several disciplines began to realize that exercise and other external changes could restore damaged connections and even generate new ones.

That included research teams at the Baltimore VA, where Forrester and a mentor, University of Maryland neurologist Richard Macko, proved during the early 2000s that stroke survivors who worked out on treadmills saw increased neural as well as motor improvement, each side apparently reinforcing the other.

This realization — that the brain is "plastic," not static — opened "huge new vistas," Roy said, for those in the rehabilitation sciences.

Researchers at the Massachusetts Institute of Technology, for example, decided to try using robot technology to feed information into the rehabilitation process. During the early 2000s, they developed a device, "MIT-MANUS," that could ask a stroke survivor to perform certain tasks with his or her arm. The robot supplied assistance only as needed. Patients using the robot were found to have improved much more quickly than those who didn't.

Forrester, Macko and others at the Baltimore VA sought to apply a similar technology toward helping stroke survivors walk. They decided to focus on the ankle, the joint that generates the most biomechanical power during walking, and collaborated with the MIT team to create Anklebot.

Roy, then a postdoctoral student at MIT, took the engineering lead, programming the device to sense changes in four types of movement — upward, downward and inward and outward — and supply support in small increments as needed.

This capacity to absorb information from the user and adjust is the key feature of both MIT-MANUS and Anklebot, said Roy, now an assistant professor at the University of Maryland in addition to his VA appointment. Each, he said, creates an exercise regimen tailor-made to users and their progress.

As Titus lumbered along on the treadmill at the Baltimore VA, the Anklebot worked smoothly. Electronic switch plates in his shoes monitored pressure, checking how well he lifted the front of his foot. The switch plates fed that data into a computer, which then drove a piston-shaped pump on either side of his left ankle, supplying just enough of a boost that he could complete each step.

"When I first started, I let the robot do all the work. Now I'm doing most of it," Titus said.

Across the room, another participant used her feet to work the controls of a video game, maneuvering a flashing disk through a series of moving gates. The movements worked the muscles and neuromotor connections involved in walking.

Tentative results suggest that both the walking and seated participants are seeing improvement in balance, propulsion and other elements of gait, findings that Roy shared at a recent conference in Germany.

Titus certainly seemed a happy customer. Since starting the study, he said, he has found himself walking 25 feet without a brace and taking stairs two at a time, feats he couldn't perform two months ago.

He also no longer fears falling, which has helped him conceive of returning to a normal life.

"I can get on the bus to get here and not even worry about it," he said. "Do you know how good that feels? For me, this robot is like a miracle."

jonathan.pitts@baltsun.com