Rex Parsons
@RexParsons8
Followers
229
Following
670
Media
13
Statuses
85
R development and health data things (Senior Data Scientist) at Nous.
Joined December 2019
New 📦: {GLMMcosinor} extends cosinor modelling with GLMM-ness! Lots of flexibility for fitting models appropriate for rhythmic data with non-Gaussian responses and hierachical structures (i.e. repeated measures). pkgsite: preprint:
2
10
41
RT @nicolem_white: Delighted to share our latest paper on clinical prediction models @aidybarnett @GSCollins @elborgo9 @RexParsons8 .We loo….
jclinepi.com
To describe the characteristics and publication outcomes of clinical prediction model studies registered on clinicaltrials.gov since 2000.
0
11
0
RT @biorxivpreprint: GLMMcosinor: Flexible cosinor modeling with a generalized linear mixed modeling framework to characterize rhythmic tim….
0
1
0
@nicolem_white @OliverRawashde1 @OllieJayasinghe All feedback is very welcome - we want this to be as accessible and easy to use as possible.
0
0
2
Thanks to the whole team @nicolem_white, Dr Prasad Chunduri, and @OliverRawashde1, but particularly @OllieJayasinghe who worked on this project as a summer research student at UQ and picked up R development like an absolute pro (way faster than I did)!.
2
0
3
We used glmmTMB as the base for fitting these models so users can specify formulas with the familiar lme4-style. There's a bunch of helpful methods to summarize and visualise the model outputs. (see .
1
0
2
GLMMcosinor solves this problem by allowing the user to specify mixed models and a family as you would when using glm(). This figure from the preprint shows (A) existing analyses where the authors used circacompare, (B) GLMMcosinor (Gamma), (C) GLMMcosinor (Gamma + mixed) model.
1
0
2
The circacompare R package gets a lot of use but sometimes inappropriately: users fitting non-mixed models to hierachical data or to non-Gaussian responses (i.e. measures that can only be positive). The appropriate methods need to be more accessible for these researchers.
1
0
1
In 2020, we published circacompare ( - an R package that uses nonlinear regression to fit models to circadian data. Although it does allow for mixed models, it doesn't extend to generalised non linear models.
1
0
2
RT @AusHSI: When patients fall, their health can decline rapidly. On the #AusHSI blog, read about @RexParsons8's PhD journey exploring digi….
0
3
0
RT @nicolem_white: New preprint with @RexParsons8 @elborgo9 @GSCollins @aidybarnett . We analysed publication outcomes for clinical predict….
osf.io
Objectives: To describe the characteristics and publication outcomes of clinical prediction model studies registered on clinicaltrials.gov since 2000. Design: Retrospective descriptive analysis. Data...
0
7
0
RT @ARDC_AU: 📊 Many models are created for #ClinicalPrediction every year, but not all of them will actually lead to better outcomes for pa….
0
4
0
RT @SusannaCramb: BIG congratulations to superstar #PhD student @RexParsons8 who gave a brilliant final seminar presentation yesterday! In….
0
5
0
RT @GSCollins: In this NEW PAPER in @BMCMedicine (with @nicolem_white @RexParsons8 @aidybarnett) we found an excess of published AUC values….
0
34
0
RT @SusannaCramb: Fabulous talks today by both teams of the 2023 Venables award - congratulations again! @RexParsons8 @aidybarnett @andrewz….
0
9
0
RT @r_medicine: Only 5 days away! @RexParsons8, Ph.D. Candidate at @AusHSI, will be leading a demo at #RMed2023 on “predictNMB: An R Packag….
0
10
0
RT @nicolem_white: Life or death research stats - our preprint on AUC hacking featured in today's campus morning mail @AusHSI. https://t.co….
0
2
0