Only one quarter of senior cit­i­zens meet the national guide­lines for reg­ular phys­ical activity, according to a report from the Amer­ican Col­lege of Sports Med­i­cine and the Amer­ican Heart Association. This is despite clear evi­dence that exer­cise is asso­ci­ated with lower risk of car­dio­vas­cular dis­ease, Type 2 dia­betes, osteo­porosis, stroke, breast cancer, colon cancer, and disability—all prob­lems of increased con­cern for older adults.

In an effort to pro­mote phys­ical activity across ages, smart­phone and tablet appli­ca­tions are rapidly emerging and stand to improve a variety of health out­comes. But older adults tend to lack the com­puter skills to uti­lize these tech­nolo­gies, according to asso­ciate pro­fessor Tim­othy Bick­more in the Col­lege of Com­puter and Infor­ma­tion Sci­ence, adding that older adults from ethnic minority groups who could ben­efit the most also tend to have lower health lit­eracy. Few, if any, of the cur­rently avail­able health-​​promotion apps are tar­geted specif­i­cally toward the needs and pref­er­ences of these communities.

Backed by a new National Insti­tutes of Health grant, Bick­more and co-​​principal inves­ti­gator Abby King of Stan­ford University’s School of Med­i­cine are seeking to bridge this gap.

In pre­vious research, Bick­more has devel­oped sys­tems that look, sound, and act more like humans than they do com­puters. These agents are actu­ally char­ac­ters with a per­sonal his­tory and the ability to sim­u­late empathy. They have names, like Carmen or Karen; they can speak a variety of lan­guages; and they con­tain mas­sive data­bases of health infor­ma­tion. Bick­more believes that Carmen and Karen could do as good a job as human exer­cise coaches, but with the added ben­e­fits of scal­a­bility, reli­a­bility, and lower cost.

Bick­more and his col­leagues put this hypoth­esis to the test in two recently com­pleted studies—one at Boston Med­ical Center out­pa­tient clinics, and another he ran with King at a Latino com­mu­nity center in Cal­i­fornia. In the BMC study, the team showed that its char­ac­ters helped older adults increase the number of steps they walked daily when they had access to them on take-​​home tablet computers.

We wanted to see if people in an urban low lit­eracy pop­u­la­tion would be able to use a tech­nology that was designed for them that would improve their health,” said Bick­more, “and we seemed to show that it did.”

The Cal­i­fornia study also demon­strated promising results when the sys­tems were deployed in pri­vate rooms at the com­mu­nity center: The par­tic­i­pants walked eight times as much as those in the con­trol group.

Par­tic­i­pants inter­acted with the rela­tional agents for two or four months depending on the study, set­ting short– and long-​​term exer­cise goals, get­ting pos­i­tive rein­force­ment from the rela­tional agents when they met those goals, and dis­cussing bar­riers to suc­cess when they fell short.

With the $3 mil­lion NIH grant, Bick­more and King now plan to repeat the Cal­i­fornia study on a broader scale. This time around, they will deploy the sys­tems at 16 com­mu­nity cen­ters around the state and com­pare the vir­tual coach to a human coun­selor, whom they hope to show is at least on par with the human inter­ven­tion. “That’s good enough,” Bick­more said, “because this is scal­able, less expen­sive, and it’s higher fidelity so it’s the same thing every time—it doesn’t have bad days.”