Health care is both the largest segment of the U.S. economy -- and one of the most troubled. Despite a reputation as the best in the world, the American health-care system suffers from spiraling costs, fragmentation, and general mediocrity. That last claim was made in a study published last week in the New England Journal of Medicine, which found that -- regardless of income, race, or location -- most Americans receive about half the care they should.

The study, conducted by Rand Health, cited a lack of technology, notably electronic medical records and decision support software, as a key reason why patients in the United States don't get better care.

"There is great consensus that the health-care system in the U.S. is not functioning well," says David Lansky, senior director of the Health Program at the Markle Foundation in New York City. Lansky says the system has gone through 20 years of reform, with little positive result.

He sees a ray of hope, though, from an unlikely source: chip-maker Intel. Historically, Intel has made nothing directly for the medical field. Yet in January 2006 it began its first clinical trial, tracking the progression of Parkinson's disease, a debilitating neurological disorder, by measuring gross motor movements and changes in speech patterns to the microsecond. To run the trial, the company's engineers spent a year developing a specialized box to make and analyze the measurements. Separately, in January, the company received a grant from the National Institutes of Health to start a 300-person clinical trial on Alzheimer's monitoring.

Both ventures come out of the Health Research and Innovation Group, part of the nascent Digital Health Group at Intel. Eric Dishman, a sociologist and ethnographer who previously spent time in research laboratories, first at Interval Research, and then Intel, was named general manager and global director of the Health Research and Innovation Group in July 2005.

Dishman's PhD thesis was in the field of communications, where he combined techniques from sociology and anthropology to study how doctors and patients interact with each other. The research involved taping conversations and analyzing them microsecond by microsecond. "It was useful but painful work," Dishman admits. "Ironically, it helped us to invent technology that may end up helping to detect Alzheimer's 10 years earlier."

Dishman's earlier work at Interval Research used crude sensor networks and data-fusion algorithms to study interactions among family members. His team found that shifts in conversation as short in length as a few tenths of a second could provide insight into whether someone was likely to develop Alzheimer's.

While Alzheimer's already afflicts four million people in the United States, Dishman notes that 100 million Americans have memory problems not related to the disease that could benefit from Intel's work.

Beyond Intel, Dishman also founded and chairs the Center for Aging Services Technologies (CAST), a coalition of 400 corporations, universities, and service agencies, created to develop technologies for helping the growing population of older people. Back at Intel, his group is hiring as many as 50 researchers to explore other ways in which technology can improve health care, especially in the home.