Present: Marianne Burke, Juvena Hitt, Kairn Kelley, Mike LaMantia, Ben
Littenberg, Gail Rose, Liliane Savard, Connie van Eeghen
Start Up: (Not sure… wasn’t there in
time!)
1.
Book Club:
Weapons of Mass Destruction, Cathy O’Neil
a. Well-supported
criticisms; not so well-supported solutions
b. Prescient, with
respect to the use of big data models and the experience of the past election
season
c. How does this
relate to our work?
i.
Identifying a group of people (e.g. high volume consumers of
alcohol) can be used by health care providers to aid those in need… and by
insurers to restrict coverage
d. Mediating
alternatives
i.
Require individual oath-taking/personal responsibility
ii.
Develop methods of identifying bias
iii.
Establish consequences to “crossing the line,” such as civil court
or class-action remedy
iv.
Re-engineer algorithms so that they are:
1. Transparent:
the features of the algorithm are packaged in a message that is delivered to
each case (person) whenever the algorithm is run, including how that case came
to be included in the algorithm
2. Feedback of
success/failure: study false positives, build in feedback either through the
algorithm or the system in which it is used
3. Does not
increase harm or decrease benefit, thereby intensifying the categorization that
again increases harm/decreases benefit
4. Not scaled
until tested
2.
Spring Semester
schedule:
No change; keep to Fridays, 2p – 3:15p.
Let Kairn, Marianne, or Connie know if this presents a problem for you.
3.
Next Workshop Meeting(s): Fridays, 2:00 p.m.
– 3:15 p.m., at Given
Courtyard South Level 4 until end of Dec.
a.
Dec 9:
Marianne’s data
b.
Dec 16:
Nancy Gell’s manuscript
c.
Jan 6: Kairn
(no Connie, Marianne)
d.
Jan 13: CvE
manuscript review
e.
Future
topics:
a.
Juvena:
protocol development
b.
LaMantia:
predictors of successful R01 applications: http://journals.plos.org/plosone/article?id=10.1371/journal.pone.0155060
Posts
No comments:
Post a Comment
Note: Only a member of this blog may post a comment.