This section is under development, because we don´t have a example data to show the results. Anyway, you can get idea of how functions are working.
This example shows deeper detail of the healthpopR
functions after your datasets are in right shape. Check section Data Handling first to get your
data in right shape.
Load Data
Assuming that you have created datasets which are ready to use and analyzed. We can load datasets to use.
loc = "data/"
population <- arrow::read_parquet(file = paste0(loc, "population.parquet"))
diagnoses <- arrow::read_parquet(file = paste0(loc, "diagnoses.parquet"))
data_dates <- arrow::read_parquet(file = paste0(loc, "data_dates.parquet"))
population_variables <- arrow::read_parquet(file = paste0(loc, "population_variables.parquet"))Let’s define some sample case of interest.
selected_exposure_icd10 = "^E11"
selected_exposure_icd9 = "^250A"
selected_exposure_icd8 = "^250"
selected_exposure_regsrc = c("avohilmo", "erko", "hilmo", "local", "ksyy", "soshilmo", "syopa")
selected_response_icd10 = "^I2[0-5]"
selected_response_icd9 = "^41[0-4]"
selected_response_icd8 = "^41[0-4]"
selected_response_regsrc = c("avohilmo", "erko", "hilmo", "local", "ksyy", "soshilmo", "syopa")Diagnoses data
Creating exposure diagnosis data
exposure_diagnoses <- search_diagnoses(
regex_icd10 = selected_exposure_icd10,
regex_icd9 = selected_exposure_icd9,
regex_icd8 = selected_exposure_icd8,
registry_source = selected_exposure_regsrc,
data_diagnoses = diagnoses
)Creating response diagnosis data
response_diagnoses <- search_diagnoses(
regex_icd10 = selected_response_icd10,
regex_icd9 = selected_response_icd9,
regex_icd8 = selected_response_icd8,
registry_source = selected_response_regsrc,
data_diagnoses = diagnoses
)Checking what registry sources are found of the diagnoses
plot_diagnoses_src(exposure_diagnoses)
plot_diagnoses_src(response_diagnoses, per_source = TRUE)Population Data
Just exposure population without all the exposure codes
dpop_exposure <- classify_population(
exposure_icd10 = selected_exposure_icd10,
exposure_src = selected_exposure_regsrc)Both, exposure and response population in one
dpop <- classify_population(exposure_icd10 = selected_exposure_icd10,
exposure_icd9 = selected_exposure_icd9,
exposure_icd8 = selected_exposure_icd8,
exposure_src = selected_exposure_regsrc,
response_icd10 = selected_response_icd10,
response_icd9 = selected_response_icd9,
response_icd8 = selected_response_icd8,
response_src = selected_response_regsrc)Population Analytics
Exposure age distribution
plot_age_distribution(dpop, group = "exposure")
plot_age_distribution(dpop, group = "exposure", subgroups = TRUE)Response age distribution
plot_age_distribution(dpop, group = "response")
plot_age_distribution(dpop, group = "response", subgroups = TRUE)Age Distribution Tables
table_age_distribution(dpop, group = "exposure")
table_age_distribution(dpop, group = "response")
table_age_distribution(dpop, group = "response", subgroups = TRUE)
table_age_distribution(dpop, group = "exposure", subgroups = TRUE)Summaries of Exposure Response Diagnoses
summary_exp_resp_order(dpop)
summary_exp_resp_crosstabulation(dpop, output = "viewer")Health Analytics
Creating A ICD-10 profile of the group
health_profiles <- classify_icd10_profile(dpop, diagnoses = diagnoses, exposure_icd10 = selected_exposure_icd10, exposure_src = selected_exposure_regsrc)
plot_health_icd10_profile(health_profiles)Finding top ICD-10 diagnoses on exposure group (compared to no-exposure group)
tbl_health <- tbl_icd10_diff_by_exposure(dpop, diagnoses = diagnoses, exposure_icd10 = selected_exposure_icd10, exposure_src = selected_exposure_regsrc)
plot_icd10_diff_by_exposure(tbl_health)Survival Analysis
Creating a survival object and plotting Kaplan Meier and Competing Risk
dsurv <- create_dsurv(dpop,
censoring_date = as.Date("2023-12-21"),
filter_early_responses = FALSE)
plot_survival_km(dsurv)
plot_survival_cr(dsurv)Cox Analysis
This is under development and is currently relying on two
extra datasets data_dates and
data_socioeconomic, which are still under development of
scaling to more universal format.
Creating a cox dataset
cox_base <- cox_create_data(dpop,
data_dates = ostpre_vastpaiv,
data_socioeconomic = edumiage,
reference_values = list("bmi_cat1" = "Healthy Weight",
"bmi_cat2" = "Healthy Weight",
"edu" = "3 - High"),
censoring_date = as.Date("2023-12-31"))Plotting a survival element
cox_plot_overall(cox_base, type = "event")Creating a model for Cox analysis
Checking spline variable
cox_plot_spline(model1, spline_var = "age_bs")Other packages has some good functions to create plots
Poisson Regression Model (SIR)
These functions are still bit use case defined, and may not work properly, if data differs of how we have data in our dataset.
In case we have selected interested exposure and response diagnoses, we can calculate and plot SIR values. Analyzes is given in list format back.
dpirr <- pirr_data(d_exposure = exposure_diagnoses,
d_response = response_diagnoses,
d_population = dpop,
var_age_start = 50 )
results1 <- pirr_results(adat = dpirr) ## DGAnother example is that we can have a dataset which included fractures, and check same outcomes with fractures data. Analyzes is given in list format back.
dpirr2 <- pirr_data(d_exposure = exposure_diagnoses,
d_response = diagnoses %>% filter(DGREG == "FRACTURES" ),
d_population = dpop,
dg_list = list(any_fracture = "ankle+forearm+hip+humerus+vertebral",
ostheoporotic = "forearm+hip+humerus+vertebral"
))
results2 <- pirr_results(adat = dpirr2)