Evaluation_OneDef.Rmd

---
output: html_document
---

```{r setup, include=FALSE}

knitr::opts_chunk$set(echo = TRUE)

files.source <- list.files("SupportCode")
sapply(paste0("SupportCode/",files.source),source)

setup_information <- read_excel("DefinitionInformationTable.xlsx", sheet = "Setup", range = "A2:F3")

query_start <- as.Date(setup_information$StartDate)
query_end <- as.Date(setup_information$EndDate)
jurisdiction <- setup_information$Jurisdiction

select_fields_table <- read_excel("DefinitionInformationTable.xlsx", sheet = "AppliedFields", range = "A2:A22") %>%
  filter(!is.na(Field)) 
select_fields <- select_fields_table$Field

information_table <- read_excel("DefinitionInformationTable.xlsx", sheet = "DefinitionInformation", range = "A2:G3")

```

```{r def1_setup, echo=FALSE, message=FALSE, warning=FALSE}

def1_table <- information_table[information_table$defX=="def1",]

def1_name <- def1_table$Syndrome
def1_short <- def1_table$Abbreviation
def1_structure <- clean_query_essence(def1_table$Structure)
def1_structure_print <- str_replace_all(def1_table$Structure,"\\^","\\\\^")
def1_url <- def1_table$API

```

```{r def1_extract, echo=FALSE, message=FALSE, warning=FALSE}

def1_full <- return_longterm_query(url = def1_url, loop_start = query_start, loop_end = query_end)

def1 <- def1_full %>%
  dplyr::select(Date, EssenceID, select_fields)  %>%
  mutate(def1 = 1) %>%
  mutate(Date = as.Date(Date, format='%m/%d/%Y')) %>%
  filter(Date >= query_start) %>%
  filter(Date <= query_end)

def1_total <- nrow(def1)
def1_total_pretty <- format(def1_total, big.mark = ",", scientific = FALSE)

```

```{r def1_elements, echo=FALSE, message=FALSE, warning=FALSE}

def1_elements <- (def1_structure$`Syndrome Element`)

def1_elements_detected <- def1 %>%
  dplyr::select(EssenceID, ChiefComplaintParsed, DischargeDiagnosis) %>%
  clean_ChiefComplaintParsed() %>%
  clean_DischargeDiagnosis() %>%
  mutate(CCDDclean = paste(ChiefComplaintParsed, DischargeDiagnosis)) %>%
  select(EssenceID, CCDDclean) %>%
  detect_elements(data = ., def1_elements, text_field = "CCDDclean") %>%
  janitor::adorn_totals(where = "row")

def1_elements_detected_table <- def1_elements_detected %>%
  select(-TruePositive, -CCDDclean) %>%
  filter(EssenceID=="Total") %>%
  pivot_longer(cols = starts_with("element"), names_to = "Syndrome Element", values_to = "VisitsMatched") %>%
  mutate(`Syndrome Element` = str_remove(`Syndrome Element`,"element_")) %>%
  mutate(`Syndrome Element` = str_replace_all(`Syndrome Element`, "\\."," ")) %>%
  full_join(def1_structure) %>%
  mutate(`Syndrome Element` = ifelse(`Element Type`=="Diagnosis Code", str_to_sentence(`Syndrome Element`), `Syndrome Element`)) %>%
  mutate(`Syndrome Element` = ifelse(`Element Type`=="CCDD Category (see ESSENCE)", is.na(`Syndrome Element`), `Syndrome Element`)) %>%
  select(`Syndrome Element`, `Element Type`, `CCDD Matches` = VisitsMatched) %>%
  arrange(desc(`CCDD Matches`))

```

---
title: "Definition Evaluation: `r def1_name`"
author: `r query_start` to `r query_end`
---  
  
**Jurisdiction**: `r jurisdiction`  
**Report Created**: `r Sys.Date()`    
**Point of Contact**: `r setup_information$PointOfContact` (`r setup_information$POCEmail`)    
**Organization**: `r setup_information$Organization`



***  
### Individual Definition Information  
  
Inclusion terms and codes are displayed in the table below (exclusion terms are not shown in the table). Visits may match on multiple syndrome elements. Refer to the csv outputs for visit-level information about which visits matched on which elements.  
  
#### `r def1_name` (`r def1_short`)   
  
**Total Visits Identified:**  
  
`r def1_total_pretty`  
  
**Query description:**    
  
`r def1_table$Description`  
  
**Full query:**  
`r def1_structure_print`  
  
```{r def1_table, echo=FALSE, message=FALSE, warning=FALSE}

DT::datatable(def1_elements_detected_table, rownames = FALSE, options = list(pageLength = nrow(def1_structure), dom = "t"), filter = "top", escape = FALSE)

```
  
  
#### Visits over time     
  
Use the slider at the bottom of the figures to adjust the start and end dates visible in the graph.   
  
```{r time_series, echo=FALSE, message=FALSE, warning=FALSE}

fill_dates <- data.frame(Date = seq.Date(from = as.Date(query_start), to = as.Date(query_end), by = "day"))

time_series_def1 <- def1 %>%
  dplyr::count(Date, name = "Visits") %>%
  filter(Date<=query_end) %>%
  full_join(fill_dates) %>%
  mutate(Syndrome = def1_short) %>%
  arrange(Date) %>%
  replace_na(list(Visits = 0))

plot_ly() %>%
  add_lines(data = time_series_def1,
            x = ~Date, y = ~Visits, name = ~Syndrome,
            line = list(color = 'rgb(226,78,66)')) %>%
  layout(
    hovermode = "compare",
    xaxis = list(
      range = c(query_start,query_end),
      rangeslider = list(type = "date", thickness = 0.1))
  )


```

  
***   
### Top terms (besides "patient")
#### Top 5 occurring terms for each definition and each field of interest:

```{r word_prep, echo=FALSE, message = FALSE, warning = FALSE, results='hide'}

my_file <- def1 %>%
  mutate(Date = mdy(Date))

my_file <- clean_Admit_Reason_Combo()
my_file <- clean_ChiefComplaintOriginal()
my_file <- clean_ChiefComplaintUpdates()
my_file <- clean_ChiefComplaintParsed()
my_file <- clean_CCDD()
my_file <- clean_DischargeDiagnosis()
my_file <- clean_TriageNotesOrig()


```


```{r top_terms, echo=FALSE, message = FALSE, warning = FALSE, fig.width=10, fig.height=8}

def_sets <- list(my_file[my_file$def1==1,])

def_sets_names <- c(def1_short)

top_words <- list()
word_count <- list()

for (i in 1:length(def_sets)) {
  
  for(j in 1:length(select_fields)) {
    
    data <- as.data.frame(def_sets[i])
    
    field <- select_fields[j]
    
    word_count[[j]] <- data %>%
      dplyr::select(field) %>%
      unnest_tokens(word, !!field) %>%
      anti_join(stop_words, by = "word") %>%
      dplyr::count(word, name = "Count", sort=TRUE) %>%
      filter(word!="patient") %>%
      top_n(5) %>%
      mutate(def = def_sets_names[i],
             field = field)

    names(word_count)[[j]] <- select_fields[j]  
    
    }
  
  top_words[[i]] <- plyr::ldply(word_count, data.frame)

  names(top_words)[[i]] <- def_sets_names[i]
  
}

top_words_combined <- ldply(top_words, data.frame) %>%
  dplyr::arrange(Count) %>%
  group_by(def,field) %>%
  dplyr::mutate(order = row_number())

ggplot(data = top_words_combined) +
  geom_bar(aes(x = Count, y = as.factor(order)), stat="identity", fill = "#a4dba4") +
  geom_text(aes(x = 0, y = as.factor(order), label = paste0(word,'  (',Count,")"), hjust = "left"), size = 3) + 
  facet_grid(field ~ def, switch = "y", scales = "free") +
  theme_minimal() +
  theme(
    axis.text = element_blank(),
    panel.grid = element_blank(),
    strip.text.y.left = element_text(angle = 0),
    panel.background = element_rect(fill = NA, color = "grey40"),
    axis.title = element_blank()
  )

```

```{r output, echo=FALSE, message=FALSE, warning=FALSE}

# All Visits
write.csv(file = paste0("Output_OneDef/",def1_short," All Visits.csv"),
          x = def1,
          row.names = FALSE)

# Def1 matched elements
write.csv(file = paste0("Output_OneDef/",def1_short," Matched Elements.csv"),
          x = def1_elements_detected,
          row.names = FALSE)


```


___
*The template for this report was created by Sara Chronister (Maricopa County Department of Public Health). For questions, ideas for improvement/collaboration, or attribution, please reach out to <sara.chronister@maricopa.gov>.*