Video games that aim to improve children's health are still in their infancy, but already a few are showing demonstrable results.

What began as, and remains, a niche concept in the $7 billion-a-year video game industry is now getting some science behind it. And the new health-themed games are tested not only for whether kids will play them but also for whether playing them changes their behavior in a healthful way.

For years, researchers have thought that a medium that sometimes turns kids into video zombies could be mobilized to help young people fighting cancer, diabetes, obesity and other health problems. But the commercial prospects of such games -- and their ability to draw more financial backing -- depended on scientifically demonstrating their value.

"When we start developing games with measurable health impacts in mind, that's when we can really start advancing public health through games," said Erin Edgerton, a program analyst with the National Center for Health Marketing, part of the Centers for Disease Control and Prevention.

One game that has already begun to produce measurable results is Re-Mission by HopeLab, a Palo Alto, Calif., nonprofit that aims to help young cancer patients.

The game features a microscopic "nanobot" named Roxxi, a shapely brunette who, at the player's direction, travels through the body blasting away at cancer cells and bacteria with a sidearm loaded with chemotherapy drugs and antibiotics. The game's 20 levels simulate seven types of cancer.

The game takes aim not at the disease but at the obstinacy of adolescence: Studies show that teenagers are more likely than young children or adults to stray from their treatment regimens. In playing the game, young people learn, for instance, that failing to eradicate every cancer cell can lead to a recurrence -- bad for their score and their health.

In a scientific trial of 375 cancer patients age 13 to 29 last year, those who played Re-Mission adhered more closely to antibiotic treatments and maintained higher levels of chemotherapy drugs in their blood. They also understood cancer better and were more confident of their ability to fight it.

"It's stealth learning," said Steve W. Cole, vice president for research at HopeLab. "The things that happen inside the game don't stay in the game; they get in your head, and they change the way you approach the world. . . . Cancer is not death knocking on your door, but basically an opponent whose butt you are going to kick."

Rashida Wilkins, 16, who took part in the study, began playing the game last year while undergoing treatment for a brain tumor at Children's Hospital of the King's Daughters in Norfolk. Now in remission, she still plays every day.

"It showed me how the chemo goes through my body and kills the cancer cells . . . and it was fun to play," she said. "I even let my little brother play it with me. He liked it. He said he learned about what I was going through."

HopeLab has distributed 40,000 copies of the game since April. It is free for cancer patients; HopeLab asks others for a $20 donation.

Not quite as far along is Immune Attack, being developed by the nonprofit Federation of American Scientists with a $1.3 million grant from the National Science Foundation. Like Re-Mission, it features a journey inside the body of a patient without a functioning immune system. Players must learn how the system works and how to activate parts of it, such as infection-fighting white blood cells, to ascend to higher levels.

The federation's goal is to get kids excited about immunology and biology, but early tests did not go smoothly.

Only 28 percent of the 250 students in the trial this spring at five high schools filled out surveys properly both before and after playing the game, said Eitan Glinert, a research assistant at the federation. That left researchers without a statistically meaningful sample. The federation plans to test the game in 200 schools next spring.

"The evaluation is useful because we're a very small organization and we're very new at making video games," Glinert said. "If we just go and make a game and we don't do any evaluation, there is a very good chance that we're going to put something in there that we find very interesting but that the average high school student wouldn't."

Efficacy alone will not guarantee that a game survives.

In the early 1990s, media researcher Debra Lieberman helped design a Nintendo-compatible game called Packy & Marlon. It featured two diabetic elephants who attend a summer camp for diabetics. Rodents scatter the camp's food and diabetes supplies, and players must help the cartoon elephants overcome those problems in order to eat well and maintain good blood sugar levels over four days.

A three-month clinical trial of 59 children with diabetes, funded partly by the National Institutes of Health, found that the game's users had a 77 percent decline in diabetes-related emergency and urgent care visits compared with a control group, Lieberman said.

"When you apply a skill in a game, there is a lot of learning and rehearsal that goes on," said Lieberman, now a lecturer at the University of California at Santa Barbara. "And in many cases, you are transferring what you've learned into your real life. So it's not just fun and games."

For a while the game was used by several clinics and HMOs, she said, but the company that owns the rights to it is now "in hibernation." Still, it is often cited by developers as a health game that was constructed the right way.

In Escape From Diab, a game being developed by Archimage Inc., a Houston-based design studio, a kid is thrust into a world where an evil king denies his subjects healthful food and exercise. The main character, Deejay, helps other kids get healthy and escape to a "golden city."

The game began as part of a nationwide study funded by the National Institute of Diabetes and Digestive and Kidney Diseases to see whether school-age children could be motivated to eat better and exercise more, reducing their risk of diabetes and obesity.

Spun off into a study of its own, Escape From Diab and a related game soon will undergo a series of four-month trials involving 200 kids to see whether players consume more fruits and vegetables and become more active, said Tom Baranowski, a psychologist and professor of pediatrics at Baylor College of Medicine in Houston. He will run the study. The project has $9 million in federal funding.

"This is a video game for health, and its prospect in the marketplace is very much dependent on whether or not it works," said Richard Buday, president of Archimage, which hopes to begin selling the games in 2008. "It's one thing to have a game that's fun and compelling, . . . but it's open field now for determining how well video games work for changing behavior."