Patients clustering - is it possible?

[Giuseppe Uras]

Hi all,

I want to apologize in advance for the potentially naive question, I am very new to Stata and my experience is mainly related to laboratory work.

We have now screened about 100 patients for the quantification of 17 biochemical targets. We would like to understand whether some of them cluster together based on this quantification and, if possible, to understand which of these targets is mostly contributing to such distrubution.

Can anyone suggest a way to approach this?

Thank you in advance for all the help.

Regards
Giuseppe

[Erik Ruzek]

I would suggest that you read the Stata pdf documentation on the cluster command. Clustering methods are used to identify groups in the data. In your case, the groups would be defined by the way that the 17 biochemical targets cohere into distinct groups based on their values. A related approach, although perhaps even more opaque in nature, is latent profile analysis (see example 52g of the SEM manual).

[Rich Goldstein]

I find your question a bit unclear but you might be looking for Q-methodology; see Akhtar-Danesh, N. 2018. qfactor: A command for Q-methodology analysis. Stata Journal 18: 432–446.