A Collaborative Modeling Framework for Multi-Study Coordinated Analyses ======================================================== - Andriy Koval, Andrea Piccinin, Scott Hofer (University of Victoria) - Graciela Muniz (University College London) - Will Beasley (University of Oklahoma) [IALSA network](http://ialsa.org) July-22-2015, *Wellcome Trust Conference* Portland Workshop ====== - examine associations between changes in **physical** functioning (i.e., grip strength, pulmonary function, chair stands, walking speed) - and **cognitive** functioning (i.e., measures of speed, memory, reasoning, executive functioning) - in **multiple-study** comparative framework using - **bivariate growth models** Publication Model ======  ## Association between changes in - [**pulmonary**](./projects/pulmonary/README.md), [**grip**](./projects/grip/README.md), [**gait**](./projects/gait/README.md) and cognitive functioning - [**cognitive**](./projects/cognitive/README.md) variables, [**physical**](./projects/physical/README.md) variables Open Science & Reproducible Research ==== - [GitHub repository](https://github.com/IALSA/IALSA-2015-Portland) - no data is shared, only models - all scripts and reports available online - 1649 models submitted so far  Dashboard ===  - built with ```shinydashboard``` R package - interactive html interface on top of R script - what exactly do those number count? About: Model ID ===  - each file with model output provided a unique model ID - ID is parsed into variables we can use in meta-analysis over studies About: Domain Mapping ==== - constructs/domains are flexible - here is our current organization - first six characters, counts  Model Collective ==== - counts models in each combination - each color adds to 100% - pivot table - study_name: EAS - physical_measure: grip - model_type: aeh - list unique values - pc_CORR_11 - what are the key info about the model that we want to collect? (ISR) Specification === - random intercepts, slopes, residuals (ISR) - Bivariate ISR (BISR) graph BISR === - view for one study and one physical measure - variance, p-value, 95CIs Compare - (EAS, grip, aeh) vs. (EAS, grip, a) - (EAS, grip, aeh) vs. (OCTO, grip, aeh) Forest === Compare - view for all studies with this physical measure - block : (eas,males) vs. (octo, females) - mmse : (octo, males) vs. (oct, females) Future === - [IalsaSynthesis](https://github.com/IALSA/IalsaSynthesis/blob/master/README.md) - R package to service collaborative workshops - add data : new **models** - add data : new **studies** - add data : new **outcomes** - add models: new **specifications** - add questions: new **analyses** Thanks ==== - IALSA team - Andrea Piccinin - Scott Hofer - Graciela Muniz - Will Beasley