2015 APHA Annual Meeting & Expo

Home Events & Meetings Annual Meeting Schedule & Program

Recorded Annual Meeting Presentations (RAMP) are now available for purchase.
Only $90 for presenters, $140 for registered attendees and $200 for non-attendees.

Online Program

326273
Promise of Artificial Intelligence for Improving Implementation of Network Based HIV Prevention Programs


Tuesday, November 3, 2015 : 8:50 a.m. - 9:10 a.m.

Eric Rice, PhD, Suzanne Dworak-Peck School of Social Work, University of Southern California, Los Angeles, CA
Amulya Yadav
L Marcolino
Milind Tambe
Robin Petering, MSW, School of Social Work, University of Southern California, Los Angeles, CA
Hailey Winetrobe, MPH, CHES, Suzanne Dworak-Peck School of Social Work, University of Southern California, Los Angeles, CA
Harmony Rhoades, PhD, Suzanne Dworak-Peck School of Social Work, University of Southern California, Los Angeles, CA
Audio (mp3) recording Recorded Presentation (mp4)
Issues: Network-based models for HIV prevention have been implemented with varying degrees of success.  One of the enduring problems in mounting these programs is determining which persons are most likely to efficiently disseminate messages in a network.  Several groups have proposed network-based solutions, including determining the most popular persons and selecting those persons who bridge the most social relations in a network (e.g. Schneider et al., 2015; Valente 2012) . Neither solution assumes uncertainty in network structure.  Yet in many public health settings, such as homeless drop-in centers and popular MSM venues, it is almost impossible to know network structure over time.

Description: In collaboration with engineers who develop artificial intelligence computer algorithms to solve complex network strategy problems, we developed PSINET, a decision support system that: handles uncertainties in network structure and evolving network state and addresses these uncertainties by using partially observable Markov decision processes (POMDPs) in influence maximization.

Lessons Learned: Using actual network data collected from homeless youth (n=386), PSINET generates solutions that are more efficient than deterministic solutions such as selecting the most popular persons, or the persons who bridge the most social ties. PSINET allows for the selection of peer leaders from a diversity of network positions, while contending with a large amount of uncertainty in the network structure.

Recommendations: Community-based programs who wish to implement network-based HIV prevention, but who cannot fully map their population’s network, should consider using PSINET to help them best select peer leaders.  We are currently pilot testing this technology.

  

Learning Areas:

Systems thinking models (conceptual and theoretical models), applications related to public health

Learning Objectives:
Demonstrate the applicability of artificial intelligence models to augmenting HIV prevention methods.

Keyword(s): HIV/AIDS, Information Technology

Presenting author's disclosure statement:

Qualified on the content I am responsible for because: I am professor in a School of Social Work, working on issues of HIV prevention since 2002. I have my PhD.
Any relevant financial relationships? No

I agree to comply with the American Public Health Association Conflict of Interest and Commercial Support Guidelines, and to disclose to the participants any off-label or experimental uses of a commercial product or service discussed in my presentation.

Back to: 4044.0: Intersection of Technology and HIV