Function reference
-
.disease_markov_model_for_stitch_model() - build an Markov probability model from multi-age prediction models
-
.mldpEHR.cv_train_outcome() - train an xgboost cross validation classification model with k-fold cross-validation
-
.mldpEHR.disease_empirical_prob_for_disease() - this method computes the probability for future disease according to the availablitiy of data
by basically computing a stitched markov model
as score availability is not unbiased (patients with less measurements tend to be healthier),
the probaility will be computed seperately for patients with / without score
we will define the following modes:
0 has required conditions at young age and known score / outcome at older age
1 has required conditions at young age and no restrictions at older age
2 no restrictions at young/older age
-
.mldpEHR.disease_expected() - calculate expected number of disease patients
-
.mortality_markov_model_for_stitch_model() - build an Markov probability model from multi-age prediction models
-
mldpEHR.disease_markov() - To implement this, all patients at a given age will be binned according to their model score (using quantiles). Each bin is assigned a state, and we are computing the probability for traversing from each state to the next model state. Patients with missing score are also included for this model to reflect actual population numbers
-
mldpEHR.disease_multi_age_predictors() - build an xgboost cross validation classification model with k-fold cross-validation for each featureset provided, assumed that the classification is defined by the previous model
-
mldpEHR.mortality_markov() - build a Markov probability model from multi-age prediction models To implement this, all patients at a given age will be binned according to their model score (using quantiles). Each bin is assigned a state, and we are computing the probability for traversing from each state to the next model state Patients with missing score are also included for this model to reflect actual population numbers
-
mldpEHR.mortality_multi_age_predictors() - build an xgboost cross validation classification model with k-fold cross-validation for each featureset provided, assumed that the classification is defined by the previous model
-
mldpEHR.prediction_model_features() - analyze feature significance for a k-fold cross validation model using shaply values