---
title: "Hospital Patient Analysis Dashboard"
output: 
  flexdashboard::flex_dashboard:
    orientation: rows
    vertical_layout: fill
    theme: flatly
    source_code: embed
---

```{r setup, include=FALSE}
library(flexdashboard)
library(tidyverse)
library(plotly)
library(DT)
library(scales)

# Load clean data
df_clean <- read.csv("data/healthcare_clean.csv") %>%
  mutate(
    date_of_admission = as.Date(date_of_admission),
    discharge_date    = as.Date(discharge_date),
    across(c(gender, blood_type, medical_condition, insurance_provider,
             admission_type, medication, test_results,
             age_group, billing_tier, data_quality_flag), as.factor)
  )

# Billing-safe subset
df_billing <- df_clean %>%
  filter(data_quality_flag == "OK")

# Pre-compute key metrics for value boxes
n_patients   <- nrow(df_clean)
avg_los      <- round(mean(df_clean$length_of_stay), 1)
avg_billing  <- dollar(round(mean(df_billing$billing_amount), 0))
pct_abnormal <- round(mean(df_clean$test_results == "Abnormal") * 100, 1)
```

Row {data-height=150}
-----------------------------------------------------------------------

### Total Patients

```{r}
valueBox(
  value   = format(n_patients, big.mark = ","),
  caption = "Total Patients",
  icon    = "fa-hospital",
  color   = "#2C3E50"
)
```

### Average Length of Stay

```{r}
valueBox(
  value   = paste(avg_los, "days"),
  caption = "Average Length of Stay",
  icon    = "fa-calendar",
  color   = "#2980B9"
)
```

### Average Billing Amount

```{r}
valueBox(
  value   = avg_billing,
  caption = "Average Billing Amount",
  icon    = "fa-dollar-sign",
  color   = "#27AE60"
)
```

### Abnormal Test Results

```{r}
valueBox(
  value   = paste0(pct_abnormal, "%"),
  caption = "Abnormal Test Results",
  icon    = "fa-flask",
  color   = "#E74C3C"
)
```

Row {data-height=850}
-----------------------------------------------------------------------

### Length of Stay by Condition and Admission Type

```{r}
p <- df_clean %>%
  group_by(medical_condition, admission_type) %>%
  summarise(
    avg_los    = round(mean(length_of_stay), 1),
    n_patients = n(),
    .groups    = "drop"
  ) %>%
  ggplot(aes(x    = reorder(medical_condition, avg_los),
             y    = avg_los,
             fill = admission_type,
             text = paste0(
               "Condition: ",  medical_condition,           "\n",
               "Admission: ",  admission_type,              "\n",
               "Avg LOS: ",    avg_los, " days",            "\n",
               "Patients: ",   format(n_patients, big.mark = ",")
             ))) +
  geom_col(position = "dodge", alpha = 0.85) +
  scale_fill_manual(values = c(
    "Elective"  = "#2C3E50",
    "Emergency" = "#E74C3C",
    "Urgent"    = "#F39C12"
  )) +
  coord_flip() +
  labs(
    x    = NULL,
    y    = "Average Length of Stay (Days)",
    fill = "Admission Type"
  ) +
  theme_minimal(base_size = 12) +
  theme(legend.position = "bottom")

ggplotly(p, tooltip = "text") %>%
  layout(legend = list(orientation = "h", y = -0.15))
```

### Patient Data Explorer {data-width=400}

```{r}
# DT renders in the browser so large datasets cause timeouts
# 2,000 rows is enough to demonstrate full interactivity
# We note the sampling in the panel so viewers understand

set.seed(42)
df_clean %>%
  slice_sample(n = 2000) %>%
  select(
    "Condition"    = medical_condition,
    "Age Group"    = age_group,
    "Admission"    = admission_type,
    "LOS (Days)"   = length_of_stay,
    "Billing ($)"  = billing_amount,
    "Insurance"    = insurance_provider,
    "Test Result"  = test_results
  ) %>%
  mutate(`Billing ($)` = round(`Billing ($)`, 0)) %>%
  datatable(
    caption  = "Showing a random sample of 2,000 patients — full dataset contains 55,392 records",
    options  = list(
      pageLength = 12,
      scrollY    = "550px",
      dom        = "ftip"
    ),
    filter   = "top",
    rownames = FALSE,
    class    = "compact"
  )
```