Patient population

[Sietze-McR]

Which algorithm or pipeline should be used for which patient population?

[AKuederle]

Hi @Sietze-McR,

From our testing within Mobilise-D we came up with two slightly different pipelines depending on the patient cohorts. See Fig. 7 here: https://www.nature.com/articles/s41598-024-51766-5
All the algorithms listed there are implemented and can be assembled according to the outlined pipeline.
Note the correction recently published for the algorithm comparison paper: https://link.springer.com/content/pdf/10.1186/s12984-024-01361-6.pdf
The used citation and description within the paper for the GSD_B algorithm was wrong and hence the citation in Fig.7 is also incorrect. A correction will be published soon as well.

In terms of algorithms this means the "normal" GsdIonescu instead of the GsdAdaptiveIonescu should be used for the P2 pipeline.

These different pipeline structures will also become more clear once #145 is merged. It will implement the pipelines as explicit classes. You can already see the algorithm mapping here:

mobgap/mobgap/pipeline/_mobilised_pipeline.py

Line 53 in a8d2c4b

normal_walking: Final = MappingProxyType(

normal_walking -> P1
imparied_walking -> P2