Open gdevenyi opened 2 years ago
Hi @gdevenyi — This is certainly conceptually possible, but not necessarily straightforward to implement in pySuStaIn.
We have some work (in progress) along these lines, but nothing that's ready yet.
Happy to discuss/collaborate — perhaps you could kickoff the discussion by giving us (POND group) an overview of the kinds of data you have.
If you fancy a(nother!) virtual meeting, we'd be happy to have you give a talk.
perhaps you could kickoff the discussion by giving us (POND group) an overview of the kinds of data you have.
We (CobraLab, PI Mallar Chakravarty) have a number of different datasets either in-progress or complete (numbers are fuzzy for me right now, but they're <200 for humans, and ~50 for mice)
Healthy aging with, T1, T2(0.67mm iso), T1maps, T2star Alzhimer's Biomarkers, T1, T2(0.67mm iso), T1maps, T2star High Risk for Psychosis, T1, T2(0.67mm iso), T1maps, 4 longitudinal timepoints over 18 months
Human measures of subcortical, hippocampal volumes, cortical thickness, longitudinal DBM, and quantitative measures therein
Triple-Transgenic Alzheimer Mice, longitudinal evaluation, with WT controls Triple-Transgenic Alzheimer Mice, longitudinal with diet and exercise interventions, WT controls Mice with maternal immune activation, longitudinal, WT controls Mice with THC maternal exposure, longitudinal, WT controls Mice model of parkinsons, longitudinal, alpha-syn injection, human alpha-syn gene knock in, WT controls
We've also curated a lot of public human data both longitudinal and cross sectional. We're always thinking about how we might better understand longitudinal data in particular.
Hi @gdevenyi thanks for the interest. We're developing a time-dependent version of the Event-Based Model, called the Temporal Event-Based Model (TEBM), that is designed to use longitudinal data. We have a journal paper about the TEBM in review and will be releasing the accompanying code shortly. If this is of interest in its own right then let us know - we could look at just running a TEBM analysis on your data first. But if you're only interested in subtyping with longitudinal data you'll have a to wait a bit longer for a "Temporal SuStaIn" - we're working on this now. As @noxtoby says, perhaps worth setting up a meeting with our team to discuss in more detail?
Lets sort out a meeting to discuss potential data/collaboration, mchakrav@gmail.com & gdevenyi@gmail.com please!
Apologies that this seems to have gotten away from us. @gdevenyi — if you fancy meeting, send me and @pawij an email.
Hi @gdevenyi thanks for the interest. We're developing a time-dependent version of the Event-Based Model, called the Temporal Event-Based Model (TEBM), that is designed to use longitudinal data. We have a journal paper about the TEBM in review and will be releasing the accompanying code shortly. If this is of interest in its own right then let us know - we could look at just running a TEBM analysis on your data first. But if you're only interested in subtyping with longitudinal data you'll have a to wait a bit longer for a "Temporal SuStaIn" - we're working on this now. As @noxtoby says, perhaps worth setting up a meeting with our team to discuss in more detail?
Hi, my data is longitudinal. I have read the conference paper of Subtype and stage inference with timescales, and I wonder whether Temporal SuStaIn is public avaliable now?
SusStaIn is a cool idea for extracting models from cross-sectional data, but one idea I have is, if my data is longitudinal, I could constrain the possible progression models that are possible?
Can such a feature go into SuStaIn? Does it conceptually make sense?
We have lots of animal longitudinal developmental and interventional data which could be used to test this.