search

britishredcrosssociety / covid-19-vulnerability

COVID-19 Vulnerability Index for neighbourhoods and Local Authorities in the UK
https://britishredcrosssociety.github.io/covid-19-vulnerability/
Other
33 stars 15 forks source link

Add sex demographics, and review other indicators used in common co-morbidity indices #14

Open henryholms opened 4 years ago

henryholms commented 4 years ago

Hi, it seems the 'Clinical vulnerability' and potentially 'Health and Wellbeing' domains are looking at similar indicators to existing co-morbidity indices, which aim to give a mortality score for a given patient based on various co-morbidity factors, e.g. asthma.

Particular examples are Charlson index, and Elixhauser index. Those papers include lists of indicators and their relative weightings in the indices.

There may be other factors with available data that you want to include, but one that stands out is sex (e.g. male:female ratio), as being male is a significant risk factor for COVID and otherwise. Could that be added as an indicator?

Best, Henry

MikeJohnPage commented 4 years ago

Concering the indicators used for each domain (e.g., 'Clinical vulnerability'), a new branch weighted-indicators has been pushed to the repo. This branch contains some unsupervised learning models (PCA and MLFA) that are to be used to determine the weights for each domain. This process should help correct for the correlation between indicators in each domain. A few more things need ironing out with the code before it can be merged to the master branch.

Thanks for the links to the indices. I will take a look and see if they can be used to improve the current index.

Regarding deomgraphic data, @matthewgthomas is there a reason sex has not been used as an indicator, I imagine the information will be readily avaiable?