---
title: "Aberdeen hospital admissions dashboard"
output: 
  flexdashboard::flex_dashboard:
    orientation: columns
    vertical_layout: fill
    source_code: embed

---

```{r setup, include=FALSE}
library(flexdashboard)
library(tidyverse)
library(lubridate) 
library(DataExplorer)

library(kableExtra)
library(scales)
library(ggridges)

library(DT)
library(ggplot2)
library(plotly)
library(shiny)
```

```{r}

## load datasets from Github
hospital_adm_df <- read_csv('https://raw.githubusercontent.com/AbdnCHDS/skills-test/main/hospital_admissions.csv')

pop_dem_df <- read_csv('https://raw.githubusercontent.com/AbdnCHDS/skills-test/main/population_demographics.csv')

## Join in one db by patient 'id' using (left join)
dt <- left_join(hospital_adm_df, pop_dem_df, by = "id")

### After the dataset "summary" check, 3 things require attention. 
# 1) Change the date format; 2) the time mismatch between admit and discharge; 
# 3) Check NA; 4) Age distribution (typos).

dt <- dt %>% 
  mutate(date_admit = dmy(date_admit),
         date_discharge = dmy(date_discharge))

## replace (typos) dates - 2002
dt$date_discharge <- as.Date(sub('2002', '2020', dt$date_discharge))

## calculate length hospital days
dt$hosp_length <- difftime(dt$date_discharge ,dt$date_admit , units = c("days"))
## convert as numeric value fifftime in hospital days
dt$hosp_length <- as.numeric(as.difftime(dt$hosp_length))

## replace age "typos" 
dt$age <- as.numeric(sub('999','99',dt$age))

## Create two variables using the case_when function 
dt$age_group <- case_when(
  dt$age >= 0  & dt$age <= 13 ~ 'Children',
  dt$age >= 14  & dt$age <= 17 ~ 'Young adults',
  dt$age >= 18  & dt$age <= 65 ~ 'Adults',
  dt$age >= 65  & dt$age <= 99 ~ 'Elderly')

dt$age_group <- as.factor(dt$age_group)
```



Part 1
=======================================================================

Column {data-width=400, .tabset}
-----------------------------------------------------------------------

### Age bar chart

```{r}

p1 <- ggplot(data= dt, aes(age, fill = factor(deprivation))) +
  geom_bar() +
  labs(x = "age", y = "") +
  theme_light()

ggplotly(p1, height= 500, width= 700)

```

### Hospital days distribution

```{r}

p2 <- dt %>%
  filter(!is.na(deprivation)) %>%
  ggplot(aes(x = hosp_length, fill = factor(deprivation))) +
  geom_density(alpha=0.4, position = "stack") +
  theme_light()

ggplotly(p2, height= 450, width= 700) 

```


Column {data-width=300}
-----------------------------------------------------------------------

### Dataset

```{r}

datatable(dt, extensions = 'Buttons',
          options = list(pageLength = 15,
                         dom = 'Bfrtip',
                             buttons = c('copy','excel','print')),rownames = FALSE)

```


Part 2
=======================================================================

Column {data-width=450}
-----------------------------------------------------------------------


### boxplot distribution by age groups and days in hosp.

```{r}
p<-dt %>%
  filter(!is.na(deprivation)) %>%
  ggplot( aes(x = age_group, y = hosp_length, fill = age_group)) +
  geom_violin(alpha = 0.5, size = .2) +
  geom_jitter(alpha = .2, width = .2) +
  geom_boxplot(alpha = .2) +
  theme(legend.position = "none") +
  labs(x = "age groups", y = "number of days in hospital")

ggplotly(p, height=500, width=700)
```

Column {data-width=300}
-----------------------------------------------------------------------

### Grouped dataset info

```{r, fig.width=50, fig.height=15}

dt1<-dt %>%
  group_by(age, deprivation) %>%
  summarise(tot_hosp_length = sum(hosp_length)) %>%
  arrange(desc(tot_hosp_length))

datatable(dt1, extensions = 'Buttons',
          options = list(pageLength = 15,
                         dom = 'Bfrtip',
                             buttons = c('copy','excel','print')),rownames = FALSE)

```