Corrected for multiple instances

R/fresh_install.R

@@ -77,7 +77,9 @@ fresh_install <- function(admin, pass, domain_id, domain_name)
 
 pop_chru <- function()
 {
-  readr::read_csv("~/2016/pims16.csv", col_types = readr::cols(.default = readr::col_character())) %>%
+  readr::read_csv("~/2016/pims16.csv", col_types = readr::cols(.default = readr::col_character())) -> patients
+
+  patients %>%
   import_patients_visits("CHRU")
 
   readr::read_csv("~/2016/diags16.csv", col_types = readr::cols(.default = readr::col_character())) %>%
@@ -86,6 +88,30 @@ pop_chru <- function()
   readr::read_csv("~/2016/actes16.csv", col_types = readr::cols(.default = readr::col_character())) %>%
   import_actes("CHRU")
 
+  readr::read_csv("~/2016/mensurations16.csv", col_types = readr::cols(.default = readr::col_character())) %>%
+  import_mensurations(patients, "CHRU")
+
+  readr::read_csv("~/2016/bio16_1.csv", col_types = readr::cols(.default = readr::col_character())) %>%
+  import_bios(patients, "CHRU")
+
+  readr::read_csv("~/2016/bio16_2.csv", col_types = readr::cols(.default = readr::col_character())) %>%
+  import_bios(patients, "CHRU")
+
+  readr::read_csv("~/2016/pims17.csv", col_types = readr::cols(.default = readr::col_character())) -> patients
+
+  patients %>%
+  import_patients_visits("CHRU")
+
+  readr::read_csv("~/2016/diags17.csv", col_types = readr::cols(.default = readr::col_character())) %>%
+  import_diagnostics("CHRU")
+
+  readr::read_csv("~/2016/actes17.csv", col_types = readr::cols(.default = readr::col_character())) %>%
+  import_actes("CHRU")
+
+  readr::read_csv("~/2016/mensurations17.csv", col_types = readr::cols(.default = readr::col_character())) %>%
+  import_mensurations(patients, "CHRU")
 
+  readr::read_csv("~/2016/bio17.csv", col_types = readr::cols(.default = readr::col_character())) %>%
+  import_bios(patients, "CHRU")
 }
 

R/import.R

@@ -79,19 +79,96 @@ import_patients_visits <- function(patients, project)
 
   # Observations : Age à l'hospitalisation
   patients %>%
-    dplyr::select(patient_ide, encounter_ide, start_date, birth_date, provider_id) %>%
+    dplyr::select(patient_ide, encounter_ide, start_date = rum_start, birth_date, provider_id) %>%
     dplyr::distinct() %>%
     dplyr::mutate(concept_cd = "HOS:age_hospit",
            provider_id = stringr::str_c("STRUCT:", provider_id),
            modifier_cd = "@",
-           instance_num = 1,
            valtype_cd = "N",
            tval_char = "E",
            nval_num = as.numeric(start_date - birth_date)/365.25) %>%
     dplyr::select(-birth_date) %>%
+    dplyr::group_by(patient_ide, encounter_ide, start_date, provider_id, concept_cd, modifier_cd) %>%
+    dplyr::mutate(instance_num = 1:n()) %>%
+    dplyr::ungroup() %>%
   add_observations(project)
 }
 
+import_mensurations <- function(mensurations, patients, project)
+{
+  mensurations %>%
+    stats::setNames(c("patient_ide", "encounter_ide", "poids", "taille", "IMC")) %>%
+    dplyr::filter(!is.na(patient_ide)) %>%
+    dplyr::mutate(encounter_ide = ifelse(encounter_ide %>% stringr::str_detect("\\."),
+                                         stringr::str_c(encounter_ide, lubridate::day(start_date) %>% stringr::str_pad(2, "left", "0")),
+                                         encounter_ide)) %>%
+    dplyr::mutate(patient_ide = ifelse(patient_ide %>% as.numeric > 2^32,
+                                patient_ide %>% stringr::str_sub(2),
+                                patient_ide)) ->
+  mensurations
+
+  patients %>%
+    stats::setNames(c("patient_ide", "encounter_ide", "start_date", "end_date", "sex_cd", "birth_date", "death_date", "rum_start", "rum_end", "provider_id", "project")) %>%
+    dplyr::mutate(start_date = start_date %>% as.Date(format = "%Y/%m/%d %H:%M:%S"),
+           end_date = end_date %>% as.Date(format = "%Y/%m/%d %H:%M:%S"),
+           birth_date = birth_date %>% as.Date(format = "%Y/%m/%d %H:%M:%S"),
+           death_date = death_date %>% as.Date(format = "%Y/%m/%d %H:%M:%S"),
+           sex_cd = ifelse(sex_cd == "1", "M", "F")) %>%
+    dplyr::filter(!is.na(patient_ide)) %>%
+    dplyr::mutate(encounter_ide = ifelse(encounter_ide %>% stringr::str_detect("\\."),
+                                         stringr::str_c(encounter_ide, lubridate::day(start_date) %>% stringr::str_pad(2, "left", "0")),
+                                         encounter_ide)) %>%
+    dplyr::mutate(patient_ide = ifelse(patient_ide %>% as.numeric > 2^32,
+                                patient_ide %>% stringr::str_sub(2),
+                                patient_ide)) ->
+  patients
+
+  patients %>%
+    dplyr::left_join(mensurations) -> mensurations
+
+  mensurations %>%
+    dplyr::select(patient_ide, encounter_ide, provider_id, start_date = rum_start, end_date = rum_end, poids) %>%
+    dplyr::mutate(concept_cd = "HOS:poids",
+                  provider_id = stringr::str_c("STRUCT:", provider_id),
+                  modifier_cd = "@",
+                  valtype_cd = "N",
+                  tval_char = "E",
+                  nval_num = as.numeric(poids)) %>%
+    dplyr::select(-poids) %>%
+    dplyr::group_by(patient_ide, encounter_ide, start_date, provider_id, concept_cd, modifier_cd) %>%
+    dplyr::mutate(instance_num = 1:n()) %>%
+    dplyr::ungroup() %>%
+    add_observations(project)
+
+  mensurations %>%
+    dplyr::select(patient_ide, encounter_ide, provider_id, start_date = rum_start, end_date = rum_end, taille) %>%
+    dplyr::mutate(concept_cd = "HOS:taille",
+                  provider_id = stringr::str_c("STRUCT:", provider_id),
+                  modifier_cd = "@",
+                  valtype_cd = "N",
+                  tval_char = "E",
+                  nval_num = as.numeric(taille)) %>%
+    dplyr::select(-taille) %>%
+    dplyr::group_by(patient_ide, encounter_ide, start_date, provider_id, concept_cd, modifier_cd) %>%
+    dplyr::mutate(instance_num = 1:n()) %>%
+    dplyr::ungroup() %>%
+    add_observations(project)
+
+  mensurations %>%
+    dplyr::select(patient_ide, encounter_ide, provider_id, start_date = rum_start, end_date = rum_end, IMC) %>%
+    dplyr::mutate(concept_cd = "HOS:IMC",
+                  provider_id = stringr::str_c("STRUCT:", provider_id),
+                  modifier_cd = "@",
+                  valtype_cd = "N",
+                  tval_char = "E",
+                  nval_num = as.numeric(IMC)) %>%
+    dplyr::select(-IMC) %>%
+    dplyr::group_by(patient_ide, encounter_ide, start_date, provider_id, concept_cd, modifier_cd) %>%
+    dplyr::mutate(instance_num = 1:n()) %>%
+    dplyr::ungroup() %>%
+    add_observations(project)
+}
+
 import_diagnostics <- function(diags, project)
 {
   # Observations : Diagnostics
@@ -120,7 +197,8 @@ import_actes <- function(actes, project)
   actes %>%
     stats::setNames(c("patient_ide", "encounter_ide", "provider_id", "concept_cd", "start_date")) %>%
     dplyr::distinct() %>%
-    dplyr::filter(!is.na(concept_cd)) %>%
+    dplyr::filter(!is.na(concept_cd),
+                  !is.na(start_date)) %>%
     dplyr::mutate(concept_cd = stringr::str_c("CCAM:", concept_cd),
            provider_id = stringr::str_c("STRUCT:", provider_id),
            modifier_cd = "@") %>%
@@ -136,3 +214,48 @@ import_actes <- function(actes, project)
   add_observations(project)
 }
 
+import_bios <- function(bios, patients, project)
+{
+  readr::read_csv("../inst/bio.map") -> mapping
+
+  bios %>%
+    stats::setNames(c("patient_ide", "encounter_ide", "start_date", "concept_cd", "nval_num")) %>%
+    dplyr::filter(!is.na(concept_cd),
+                  !is.na(nval_num),
+                  !is.na(start_date)) %>%
+    dplyr::filter(!concept_cd %in% c("MB_SGT_AER_CB", "MB_SGT_ANA_CB", "MB_LP_TC", "MB_SGT_PED_CB", "MB_CS_NUM_DON_RC", "MB_ANTIBIO_RC")) %>%
+    dplyr::left_join(mapping, by = c("concept_cd" = "from")) %>%
+    dplyr::mutate(concept_cd = ifelse(!is.na(to), to, concept_cd)) %>%
+    dplyr::mutate(concept_cd = stringr::str_c("BIO:", concept_cd)) %>%
+    dplyr::mutate(start_date = start_date %>% as.Date(format = "%Y/%m/%d %H:%M:%S")) %>%
+    dplyr::select(-to) -> bios
+
+  patients %>%
+    stats::setNames(c("patient_ide", "encounter_ide", "start_date", "end_date", "sex_cd", "birth_date", "death_date", "rum_start", "rum_end", "provider_id", "project")) %>%
+    dplyr::mutate(rum_start = rum_start %>% as.Date(format = "%Y/%m/%d %H:%M:%S"),
+                  rum_end = rum_end %>% as.Date(format = "%Y/%m/%d %H:%M:%S")) %>%
+    dplyr::filter(!is.na(patient_ide)) %>%
+    dplyr::mutate(encounter_ide = ifelse(encounter_ide %>% stringr::str_detect("\\."),
+                                         stringr::str_c(encounter_ide, lubridate::day(start_date) %>% stringr::str_pad(2, "left", "0")),
+                                         encounter_ide)) %>%
+    dplyr::select(patient_ide, encounter_ide, rum_start, rum_end, provider_id) %>%
+    dplyr::mutate(patient_ide = ifelse(patient_ide %>% as.numeric > 2^32,
+                                       patient_ide %>% stringr::str_sub(2),
+                                       patient_ide)) ->
+  patients
+
+  bios %>%
+    dplyr::left_join(patients, by = c("patient_ide", "encounter_ide")) %>%
+    dplyr::filter(start_date >= rum_start & start_date <= rum_end) %>%
+    dplyr::select(-rum_start, -rum_end) %>%
+    dplyr::mutate(provider_id = stringr::str_c("STRUCT:", provider_id),
+                  modifier_cd = "@",
+                  valtype_cd = "N",
+                  tval_char = "E") %>%
+    dplyr::mutate(nval_num = nval_num %>% stringr::str_replace(",", ".")) %>%
+    dplyr::group_by(patient_ide, encounter_ide, start_date, provider_id, concept_cd, modifier_cd) %>%
+    dplyr::mutate(instance_num = 1:n()) %>%
+    dplyr::ungroup() %>%
+  add_observations(project)
+}
+