Breast­feeding for the first six months of an infant’s life is linked to a lower risk of sev­eral health prob­lems for both the mother and child. For instance, chil­dren who are breastfed are less likely to suffer from asthma, obe­sity, child­hood leukemia and both type 1 and type 2 dia­betes. Mothers who breast­feed their chil­dren have lower risk of type 2 dia­betes, breast cancer, ovarian cancer and post­partum depression.

Nonethe­less, only 15 per­cent of Amer­ican babies are breastfed exclu­sively for the first six months of their lives, a time­frame rec­om­mended by more than 40 health­care orga­ni­za­tions. The per­centage of breastfed babies is much higher in other parts of the world.

Roger Edwards of the Bouvé Col­lege of Health Sci­ences and Tim Bick­more, of the Col­lege of Com­puter and Infor­ma­tion Sci­ence, are trying to reverse those trends through devel­oping a com­put­er­ized lactation-​​education con­sul­tant that helps sup­port and inform expecting mothers. Bick­more has devel­oped sim­ilar sys­tems for use in walking-​​promotion studies, oncology clin­ical trials and patient adher­ence programs.

Backed by funding from a North­eastern Seed Grant, the duo car­ried out a 15-​​person pilot study to test the system’s poten­tial impact. The researchers found that the inter­ven­tion strategy sig­nif­i­cantly increased both breast­feeding knowl­edge and the intent to breastfeed.

Bick­more and Edwards spent four months devel­oping the system, working closely with a lac­ta­tion con­sul­tant at Melrose-​​Wakefield Hos­pital, where the pilot test was conducted.

The researchers video­taped the lac­ta­tion con­sul­tant giving breast­feeding tuto­rials. In all of the avatars Bick­more designs, the goal is the same: “We’re always going back to source videos and dia­logues to cap­ture the essence of the expert,” he explained. “We look a lot at non­verbal behav­iors, like hand ges­tures, gaze cues and head nods.”

Armed with data, the team is now working to secure funding for a broad-​​based lon­gi­tu­dinal study, to test whether mothers who interact with the system would breast­feed for a longer period of time than new moms who don’t.

The team also hopes to upgrade the system by designing a 3-​​D ver­sion of the com­put­er­ized con­sul­tant and building a training system in which mothers-​​to-​​be would receive feed­back on holding a baby doll that con­tains embedded sensors.

In a 500-​​patient study of a sim­ilar system not related to breast­feeding, Bick­more found that 74 per­cent of users pre­ferred to hear infor­ma­tion from a com­puter rather than a live nurse. But Edwards said that the tech­nology is designed to aug­ment human exper­tise, not sub­sti­tute for it.

Tech­nology is not going to change a sit­u­a­tion where the hos­pital norms are not sup­porting breast­feeding,” he said. “We have to dili­gently work on changing these mater­nity practices.”