ehr2pheno: Phenotyping from EHR data

In borishejblum/phenotypr: Elastic-net logistic regression for phenotyping prediction using EHR data

Phenotyping from EHR data

Description

Using both diagnosis codes and CUIs extracted from NLP processed notes and selecting predicting features

Usage

ehr2pheno(codes, nlp, covariates = NULL, ...)

Arguments

codes	a n x q matrix containg the occurences of the q diagnosis codes related to the phenotype of interest for all n patients
nlp	a n x p matrix containg the occurences of the q diagnosis codes related to the phenotype of interest for all n patients
covariates	an optional matrix of dimension n x r containing additional potential predictors
...	further arguments to be passed to get_thresholds or pheno_feat_sel subfunctions

Value

an object of type ehr2pheno, which is a list of the following elements:

• feat_sel: itself a list with the three glmnet models, "codes", "nlp" and "both"

• surrogates: itself a list with the three constructed surrogates, "codes", "nlp" and "both"

borishejblum/phenotypr documentation built on May 2, 2022, 11:04 p.m.