Matchmaking For Health Facilitating mentoring in peer health communities Finding the right mentor in an online health community is challenging Motivation Wanda Pratt, PhD David McDonald, PhD Andrea Hartzler, PhD Albert Park, MS Jina Huh, PhD Troy Griffiths Megan Taylor Approach Supported by NSF SHB #111718 Recommend peers as potential mentors o Social matching algorithms o Personalized mentor recommendations o Informative mentoring profiles o Different mentoring contexts One-on-one buddy system One-shot anonymity Group campaign Process posts to extract characteristics in individual mentoring profiles o Extract health interests using MetaMap with select UMLS semantic types o Extract personal qualities using Linguistic inquiry & word count (LIWC) for sentiment analysis o Extract interpersonal connections using social network analysis Mentor matching connects patients with helpful, experienced peers o Social matching tools provide personalized mentor recommendations o Illustrate factors that make a good match between peers o Facilitate good matchmaking and connect peers for mentorship o Help ‘newcomers learn what to expect’ and help ‘veterans to give back’ Diagnosed with kidney cancer Starting chemotherapy Avid knittera Progress Who in this community can answer my questions? Who has experiences that I can learn from? Interactive Prototyping: Developed visualizations of profiles using word cloud, timeline, and text based approaches Studies: • Evaluating text extraction process • Validating health interest profiles • Understanding needs and preferences of community peers • Assessing peer mentor recommendations with patients • Understanding characteristics of posts that encourage community participation Social Matching Algorithms: Developed algorithms to recommend candidate peer mentors based upon health interests, demographics, and personal style How will chemotherapy affect my joints? 1. 2. 3. Evaluate mentor matching with participants o What characteristics are important for a good match? o What types of mentoring contexts do people prefer? o What types of automatically extracted information do people want to see and what are they willing to share? Kidney cancer survivor Underwent chemotherapy Learning to knit Text Extraction Techniques: Developed text processing pipeline to determine health interest personal style profiles based on online community posts
Text of Matchmaking For Health - depts.washington.edu
Matchmaking For Health Facilitating mentoring in peer health
communities!
Finding the right mentor in an online health community is
challenging"
Motivation"
Albert Park, MS!
Megan Taylor! !
o Social matching algorithms! o Personalized mentor
recommendations! o Informative mentoring profiles! o Different
mentoring contexts!
u One-on-one buddy system! u One-shot anonymity! u Group
campaign!
Process posts to extract characteristics in individual mentoring
profiles" o Extract health interests using MetaMap with
select
UMLS semantic types!
o Extract personal qualities using Linguistic inquiry & word
count (LIWC) for sentiment analysis !
o Extract interpersonal connections using social network analysis
!
Mentor matching connects patients with helpful, experienced peers "
o Social matching tools provide personalized mentor
recommendations!
o Illustrate factors that make a good match between peers!
o Facilitate good matchmaking and connect peers for
mentorship!
o Help ‘newcomers learn what to expect’ and help ‘veterans to give
back’! ü Diagnosed with kidney cancer !
ü Starting chemotherapy! ü Avid knittera!
Progress"
Interactive Prototyping: Developed visualizations of profiles using
word cloud, timeline, and text based approaches!
Studies:" • Evaluating text extraction process! • Validating health
interest profiles! • Understanding needs and preferences of
community peers! • Assessing peer mentor recommendations with
patients! • Understanding characteristics of posts that
encourage community participation!
How will chemotherapy affect my joints?!
1."
2."
3." Evaluate mentor matching with participants" o What
characteristics are important for a
good match?!
o What types of mentoring contexts do people prefer?!
o What types of automatically extracted information do people want
to see and what are they willing to share?!
ü Kidney cancer survivor! ü Underwent chemotherapy! ü Learning to
knit!
Text Extraction Techniques: Developed text processing pipeline to
determine health interest personal style profiles based on online
community posts !
