Roy Perlis
Holly Sciortino
Patience Gallagher
Isaac Kohane
Susanne Churchill
Vivian Gainer
Griffin Weber
Dan Iosifescu
Shawn Murphy
Sergey Goryachev
Action Items:
Action Item |
Description |
Owner |
Source of Action |
1 |
Ask Lynn if it's possible to query Crimson to see which imaged subjects blood is available for |
Roy |
8.7.09 Meeting |
2 |
Generate list of useful scans |
Dan |
8.7.09 Meeting |
3 |
Results of classification (outcome measures) |
Tianxi |
8.7.09 Meeting |
4 |
Make list about what concepts/covariates are needed to pull out |
Dan and Roy |
8.7.09 Meeting |
5 |
Move classifier to another workstation |
Shawn/Sergey |
8.7.09 Meeting |
6 |
Eventually move classifier to I2B2 server |
Shawn/Sergey |
8.7.09 Meeting |
7 |
Provide instructions for how to get to classifier using remote desktop |
Shawn/Sergey |
8.7.09 Meeting |
8 |
Talk with Dan Mirel – confirm genotyping costs |
Roy |
8.7.09 Meeting |
Minutes:
· IMAGING
i. Non DTI: Healthy volunteers(?): 8 out of 80 usable
1. total of 20 subjects anticipated
2. ~ 1 month to complete phenotype data set
ii. DTI: 170 out of 320 usable
1. total of 150 subjects (excluding children) anticipated
2. ~6 months to complete phenotype data set
iii. 5-10% of these will not be useful (but information about cases not included will be kept to reduce bias)
· Access to DTI data
· Classification
i. Shi and graduate student are working on using notes to determine outcomes
ii. Roy and Dan need to create a list of concepts/covariates to pull out of notes
· A new analyst will be joining the team soon – working under Vivian – but all work still goes through Vivian (She is the point person for DBP and PV)
· Sergey’s classifier
i. According to PHS security guidelines, identified data can only be stored on a PHS server
1. Data can not be open to databases without required approvals
ii. Short term solution = the classifier is currently only on Sergey’s workstation which can be accessed with remote desktop from a Partners’ computer or with VPN access
1. Shawn and/or Sergey will also provide instructions for how to get to the classifier using remote desktop
iii. Long term solution – Shawn plans to integrate the classifier with the i2b2 applications
· PV
i. Not much dose-response, but this is without covariates
ii. Use of multiple antidepressants leads to better signal for most of the poor treatment outcomes
iii. Roy (and others?) need to sort out how to do analysis
1. define clinical terms
· Genetics
The genotyping is expected to take 6 months (start to finish)
October 1, 2009 – Target date to cut the link and move samples over (independent of having imaging data available)