search

ucl-pond / pySuStaIn

Subtype and Stage Inference (SuStaIn) algorithm with an example using simulated data.
MIT License
135 stars 64 forks source link

Mixed Data Types #55

Open abadgerw opened 4 months ago

abadgerw commented 4 months ago

Thank you for a fantastic resource to model disease trajectories. I wanted to inquire about whether it was possible to model both continuous and ordinal data within the same model?

abadgerw commented 2 months ago

I just wanted to circle back to see if you had a chance to review this query?

ayoung11 commented 2 months ago

It is possible to model continuous and ordinal data that can be described using either the event-based or scored events (ordinal) model in the same model.

The scored events model is a generalised version of the event-based model that allows for multiple events per biomarker rather than just a single transition from a normal to an abnormal level. Both the event-based and scored events model take as input probabilities a biomarker has reached a certain score (an abnormal score for the event-based model, and a set of scores for the scored events model) and so can be used with continuous and ordinal data by using appropriate probability distributions.

The z-score model requires z-scored data as input and it is not possible to combine this with the event-based or scored events model using the current version of pySuStaIn. We're planning to incorporate this functionality into later versions.

abadgerw commented 2 months ago

Thanks, @ayoung11! Is there a vignette or any resources on how to do the approach you've outlined?