Print

Print
Hi Anderson,
                  This is with reference with my previous question interaction about sexes with age as covariates.The data is taken from 3 sites.
I have two groups concussed and contact-control

Now there are three sites

Site1: TrimTrio scanner with 12 Ch head coil

Site2: TrimTrio scanner with 32CH head Coil , Prisma Scanner 32 CH head Coil

Site3: TrimTrio scanner with 32CH head Coil , Prisma Scanner 32 CH head Coil

Now my design.mat file according to out discussion will be


EV1: Males, concussed
EV2: Males, controls
EV3: Females, concussed
EV4: Females, controls
EV5: Age
EV6: +1 for scanner 1,  0 otherwise  ....SIte 1
EV7: +1 for scanner TrimTrio (32CH), -1 for Prisma(32 CH), 0 otherwise...Site 2
EV8: +1 for scanner TrimTrio(32CH), -1 for Prisma(32CH), 0 otherwise.....Site 3

Now I work very closely with Dr Jaroslaw Harezlak of Indiana University and we were trying to find a consistency of TBSS GLM and GLM written in R .

Now in the above according to Dr Jaroslaw we have age, Site1_Scanner1, Site2_Scanner1, Site2_Scanner2, Site3_Scanner1 and Site3_Scanner2.
6 variables so we should put 6-1 =5 covariets in R coding.

However in TBSS we have 4 covariates in the measure ? Do you know how we can compare the TBSS coding with R scripting of this setup?

Kind regards
Sourajit




On Tue, Nov 21, 2017 at 8:56 AM, Anderson M. Winkler <[log in to unmask]> wrote:
Hi Sourajit,

Yes, that's right.

All the best,

Anderson


On 19 November 2017 at 18:17, sourajit mitra <[log in to unmask]> wrote:
Dear Anderson,
                      Thank you for the comments.

Also in the contrast in addition to interaction if we consider male>female, male < female in concussed as well as control
groups then it should be additional

C3: [1 0 -1 0 0 0 0 0]
C4: [-1 0 1 0 0 0 0 0]
C5: [0 1 0 -1 0 0 0 0]
C6: [0 -1 0 1 0 0 0 0]

in contrast file.

Please give me a feedback on this additional feature.

Kind regards
Sourajit




On Sat, Nov 18, 2017 at 7:14 PM, Anderson M. Winkler <[log in to unmask]> wrote:
Hi Sourajit,

Before suggesting a model, here are two general comments:

- Site is a qualitative variable, not quantitative, such that coding as 1, 2, 3 isn't adequate. Instead, need to code the three sites using two EVs (or three, depending on how it's coded).

- Sites with same brand/model of scanner, and even same coils, can still differ due to calibration, software version, etc. Thus, instead of considering two types of scanners, consider each individual machine.

From your example it seems site 1 has two scanners, whereas sites 2 and 3 have one scanner each. Thus, it seems you're working with four individual machines.

Consider then the following design:

EV1: Males, concussed
EV2: Males, controls
EV3: Females, concussed
EV4: Females, controls
EV5: Age
EV6: +1 for scanner 1, -1 for scanner 2, 0 otherwise
EV7: +1 for scanner 2, -1 for scanner 3, 0 otherwise
EV8: +1 for scanner 3, -1 for scanner 4, 0 otherwise

The contrasts of interest for the interaction are then:
C1: [1 -1 -1 1 0 0 0 0]
C2: [-1 1 1 -1 0 0 0 0]

All the best,

Anderson


On 16 November 2017 at 15:00, Sourajit Mitra Mustafi <[log in to unmask]> wrote:
Dear FSL users,
                        I have two group of subjects one is Contact-Control and another is Concussed. In each group there are both male and female.
Now I want to see interaction of sexes in groups. The study is multi-site, 3 sites, 2 type of scanners.

So I thought of 2 possible ways

Method 1
design.mat

Group1 Group2 Sex(Group1) Sex (Group2) CoV1(age) Cov1(Site) CoV3(Scanner)
1              0             2                0                    18              1               1
1              0             1                0                    19              1               2
1              0             2                0                    20              2               1
1              0             1                0                    19              3               2
0              1             0                1                    21              2               1
0              1             0                2                    22              3               2
0              1             0                2                    19              2               1
0              1             0                1                    20              2               1
......
......

2= female 1-male

and the design.con

Group1 Group2 Sex(Group1) Sex (Group2) CoV1(age) Cov1(Site) CoV3(Scanner)
0             0            1                 -1                     0               0               0                     (interaction 1)
0             0           -1                  1                     0               0               0                     (interaction 2)
0             0             0                  1                    0                0               0                     (positive correlation)
0             0             0                 -1                    0                0               0                     (negative correlation)
0             0             1                  0                    0                0               0                      (positive correlation)
0             0            -1                  0                    0                0               0                      (negative correlation)

Another alternate approach I thought of is


Method 2
design1.mat

MALE FEMALE  MALE (GROUP) FEMALE(Group) CoV1(age) Cov1(Site) CoV3(Scanner)
1              0             2                0                    18              1               1
1              0             1                0                    19              1               2
1              0             2                0                    20              2               1
1              0             1                0                    19              3               2
0              1             0                1                    21              2               1
0              1             0                2                    22              3               2
0              1             0                2                    19              2               1
0              1             0                1                    20              2               1
......
......
where 2=Contact-Control and 1=concussed.

and design1.con looks something like

MALE FEMALE  MALE (GROUP) FEMALE(Group) CoV1(age) Cov1(Site) CoV3(Scanner)
0             0            1                 -1                     0               0               0                     (interaction 1)
0             0           -1                  1                     0               0               0                     (interaction 2)
0             0             0                  1                    0                0               0                     (positive correlation)
0             0             0                 -1                    0                0               0                     (negative correlation)
0             0             1                  0                    0                0               0                      (positive correlation)
0             0            -1                  0                    0                0               0                      (negative correlation)

Could you please specify which approach is appropriate in this regard?
Which will give me interaction of sexes in groups?

Thanking you
Sourajit