Healthcare Analysis
Oppy
2023-11-17
Background
This report provides a comprehensive overview of key metrics that are vital for hospital management and policy-making. It helps in understanding the demographic profile of patients, the financial aspects of healthcare delivery, and the efficiency of hospital services, all of which are critical for informed decision-making by the hospital board.
Introduction
Healthcare is an important part of human existence and would usually come at a cost directly or indirectly. Directly in terms of paying out of pocket or indirectly through a founded program. Also, health challenges vary from person to person due to varying factors. Some health challenges are generally attributed to certain age groups while same may arise due to other socioeconomic factors or lifestyle.
Objective
1. To explore trends in the given healthcare dataset
2. Use visualization techniques to understand and communicate patterns in the dataset
Data Source
The data used in this analysis is a public dataset from Kaggle kaggle.. A copy of the dataset can be obtained here: Healthcare Dataset
Scope
* Explore the data
* Descriptive statistics
* Use visualization tools to present findings
* Document each step in a communicable manner.
Process Flow
A. Install r
B. Install Rstudio
C. Set work directory
D. Install and load tidyverse - to input dataset into r & data manipulation.
E. Telescope your dataset - to gain an initial overview of the data
F. Identify data types and convert as necessary
__install package - tidyverse - used for data manipulation_
options(repos = c(CRAN = "https://cran.rstudio.com"))
install.packages("tidyverse")## Installing package into 'C:/Users/fadar/AppData/Local/R/win-library/4.3'
## (as 'lib' is unspecified)## package 'tidyverse' successfully unpacked and MD5 sums checked
##
## The downloaded binary packages are in
## C:\Users\fadar\AppData\Local\Temp\Rtmp40pc2Z\downloaded_packagesload tidayverse into the R session
library(tidyverse)## ── Attaching core tidyverse packages ──────────────────────── tidyverse 2.0.0 ──
## ✔ dplyr 1.1.3 ✔ readr 2.1.4
## ✔ forcats 1.0.0 ✔ stringr 1.5.0
## ✔ ggplot2 3.4.4 ✔ tibble 3.2.1
## ✔ lubridate 1.9.3 ✔ tidyr 1.3.0
## ✔ purrr 1.0.2
## ── Conflicts ────────────────────────────────────────── tidyverse_conflicts() ──
## ✖ dplyr::filter() masks stats::filter()
## ✖ dplyr::lag() masks stats::lag()
## ℹ Use the conflicted package (<http://conflicted.r-lib.org/>) to force all conflicts to become errorsimporting your dataset
healthcare <- read.csv("healthcare_dataset.csv")
View(healthcare)to have a quick view of the dataset
summary(healthcare)## Name Age Gender Blood.Type
## Length:10000 Min. :18.00 Length:10000 Length:10000
## Class :character 1st Qu.:35.00 Class :character Class :character
## Mode :character Median :52.00 Mode :character Mode :character
## Mean :51.45
## 3rd Qu.:68.00
## Max. :85.00
## Medical.Condition Date.of.Admission Doctor Hospital
## Length:10000 Length:10000 Length:10000 Length:10000
## Class :character Class :character Class :character Class :character
## Mode :character Mode :character Mode :character Mode :character
##
##
##
## Insurance.Provider Billing.Amount Room.Number Admission.Type
## Length:10000 Min. : 1000 Min. :101.0 Length:10000
## Class :character 1st Qu.:13507 1st Qu.:199.0 Class :character
## Mode :character Median :25258 Median :299.0 Mode :character
## Mean :25517 Mean :300.1
## 3rd Qu.:37734 3rd Qu.:400.0
## Max. :49996 Max. :500.0
## Discharge.Date Medication Test.Results
## Length:10000 Length:10000 Length:10000
## Class :character Class :character Class :character
## Mode :character Mode :character Mode :character
##
##
## dim(healthcare) # return number of rows vs column## [1] 10000 15str(healthcare)## 'data.frame': 10000 obs. of 15 variables:
## $ Name : chr "Tiffany Ramirez" "Ruben Burns" "Chad Byrd" "Antonio Frederick" ...
## $ Age : int 81 35 61 49 51 41 82 55 33 39 ...
## $ Gender : chr "Female" "Male" "Male" "Male" ...
## $ Blood.Type : chr "O-" "O+" "B-" "B-" ...
## $ Medical.Condition : chr "Diabetes" "Asthma" "Obesity" "Asthma" ...
## $ Date.of.Admission : chr "17/11/2022" "1/06/2023" "9/01/2019" "2/05/2020" ...
## $ Doctor : chr "Patrick Parker" "Diane Jackson" "Paul Baker" "Brian Chandler" ...
## $ Hospital : chr "Wallace-Hamilton" "Burke, Griffin and Cooper" "Walton LLC" "Garcia Ltd" ...
## $ Insurance.Provider: chr "Medicare" "UnitedHealthcare" "Medicare" "Medicare" ...
## $ Billing.Amount : num 37491 47304 36875 23303 18086 ...
## $ Room.Number : int 146 404 292 480 477 180 161 384 215 310 ...
## $ Admission.Type : chr "Elective" "Emergency" "Emergency" "Urgent" ...
## $ Discharge.Date : chr "1/12/2022" "15/06/2023" "8/02/2019" "3/05/2020" ...
## $ Medication : chr "Aspirin" "Lipitor" "Lipitor" "Penicillin" ...
## $ Test.Results : chr "Inconclusive" "Normal" "Normal" "Abnormal" ...head(healthcare)## Name Age Gender Blood.Type Medical.Condition Date.of.Admission
## 1 Tiffany Ramirez 81 Female O- Diabetes 17/11/2022
## 2 Ruben Burns 35 Male O+ Asthma 1/06/2023
## 3 Chad Byrd 61 Male B- Obesity 9/01/2019
## 4 Antonio Frederick 49 Male B- Asthma 2/05/2020
## 5 Mrs. Brandy Flowers 51 Male O- Arthritis 9/07/2021
## 6 Patrick Parker 41 Male AB+ Arthritis 20/08/2020
## Doctor Hospital Insurance.Provider Billing.Amount
## 1 Patrick Parker Wallace-Hamilton Medicare 37490.98
## 2 Diane Jackson Burke, Griffin and Cooper UnitedHealthcare 47304.06
## 3 Paul Baker Walton LLC Medicare 36874.90
## 4 Brian Chandler Garcia Ltd Medicare 23303.32
## 5 Dustin Griffin Jones, Brown and Murray UnitedHealthcare 18086.34
## 6 Robin Green Boyd PLC Aetna 22522.36
## Room.Number Admission.Type Discharge.Date Medication Test.Results
## 1 146 Elective 1/12/2022 Aspirin Inconclusive
## 2 404 Emergency 15/06/2023 Lipitor Normal
## 3 292 Emergency 8/02/2019 Lipitor Normal
## 4 480 Urgent 3/05/2020 Penicillin Abnormal
## 5 477 Urgent 2/08/2021 Paracetamol Normal
## 6 180 Urgent 23/08/2020 Aspirin Abnormaltail(healthcare)## Name Age Gender Blood.Type Medical.Condition
## 9995 Jorge Obrien 69 Male A+ Diabetes
## 9996 James Hood 83 Male A+ Obesity
## 9997 Stephanie Evans 47 Female AB+ Arthritis
## 9998 Christopher Martinez 54 Male B- Arthritis
## 9999 Amanda Duke 84 Male A+ Arthritis
## 10000 Eric King 20 Male B- Arthritis
## Date.of.Admission Doctor Hospital
## 9995 25/12/2021 Frank Miller Scott LLC
## 9996 29/07/2022 Samuel Moody Wood, Martin and Simmons
## 9997 6/01/2022 Christopher Yates Nash-Krueger
## 9998 1/07/2022 Robert Nicholson Larson and Sons
## 9999 6/02/2020 Jamie Lewis Wilson-Lyons
## 10000 22/03/2023 Tasha Avila Torres, Young and Stewart
## Insurance.Provider Billing.Amount Room.Number Admission.Type
## 9995 UnitedHealthcare 16793.598 341 Elective
## 9996 UnitedHealthcare 39606.840 110 Elective
## 9997 Blue Cross 5995.717 244 Emergency
## 9998 Blue Cross 49559.203 312 Elective
## 9999 UnitedHealthcare 25236.345 420 Urgent
## 10000 Aetna 37223.966 290 Emergency
## Discharge.Date Medication Test.Results
## 9995 6/01/2022 Penicillin Inconclusive
## 9996 2/08/2022 Ibuprofen Abnormal
## 9997 29/01/2022 Ibuprofen Normal
## 9998 15/07/2022 Ibuprofen Normal
## 9999 26/02/2020 Penicillin Normal
## 10000 15/04/2023 Penicillin Abnormalclass(healthcare)## [1] "data.frame"glimpse(healthcare)## Rows: 10,000
## Columns: 15
## $ Name <chr> "Tiffany Ramirez", "Ruben Burns", "Chad Byrd", "Ant…
## $ Age <int> 81, 35, 61, 49, 51, 41, 82, 55, 33, 39, 45, 23, 85,…
## $ Gender <chr> "Female", "Male", "Male", "Male", "Male", "Male", "…
## $ Blood.Type <chr> "O-", "O+", "B-", "B-", "O-", "AB+", "AB+", "O-", "…
## $ Medical.Condition <chr> "Diabetes", "Asthma", "Obesity", "Asthma", "Arthrit…
## $ Date.of.Admission <chr> "17/11/2022", "1/06/2023", "9/01/2019", "2/05/2020"…
## $ Doctor <chr> "Patrick Parker", "Diane Jackson", "Paul Baker", "B…
## $ Hospital <chr> "Wallace-Hamilton", "Burke, Griffin and Cooper", "W…
## $ Insurance.Provider <chr> "Medicare", "UnitedHealthcare", "Medicare", "Medica…
## $ Billing.Amount <dbl> 37490.983, 47304.065, 36874.897, 23303.322, 18086.3…
## $ Room.Number <int> 146, 404, 292, 480, 477, 180, 161, 384, 215, 310, 3…
## $ Admission.Type <chr> "Elective", "Emergency", "Emergency", "Urgent", "Ur…
## $ Discharge.Date <chr> "1/12/2022", "15/06/2023", "8/02/2019", "3/05/2020"…
## $ Medication <chr> "Aspirin", "Lipitor", "Lipitor", "Penicillin", "Par…
## $ Test.Results <chr> "Inconclusive", "Normal", "Normal", "Abnormal", "No…names(healthcare)## [1] "Name" "Age" "Gender"
## [4] "Blood.Type" "Medical.Condition" "Date.of.Admission"
## [7] "Doctor" "Hospital" "Insurance.Provider"
## [10] "Billing.Amount" "Room.Number" "Admission.Type"
## [13] "Discharge.Date" "Medication" "Test.Results"the dataset can be grouped into three types of data
i. categorical
ii. date
iii. numeric
Categorical data
an initial review of the data suggest the following are categorical data to be converted to factors Gender, Blood.Type, Admission.Type, Test results
healthcare$Gender <- as.factor(healthcare$Gender)
healthcare$Blood.Type <- as.factor(healthcare$Gender)
healthcare$Admission.Type <- as.factor(healthcare$Admission.Type)
healthcare$Test.Results <- as.factor(healthcare$Test.Results)date date
convert columns with date to date format
healthcare$Date.of.Admission <- as.Date(healthcare$Date.of.Admission, format="%d/%m/%Y")
healthcare$Discharge.Date <- as.Date(healthcare$Discharge.Date, format="%d/%m/%Y")numeric data
summary(healthcare$Age)## Min. 1st Qu. Median Mean 3rd Qu. Max.
## 18.00 35.00 52.00 51.45 68.00 85.00Findings
Basic descriptive statistics for the age variable. The minimum age is 18 years. 1st Qu. (First Quartile): 25% of the patients are aged 35 or younger.Median: the 50th percentile. Half of the patients are younger than 52 years, and half are older. Mean: The mean age in your dataset is approximately 51.45 years.3rd Qu. 75% of the patients are aged 68 or younger. Max The maximum age in your dataset is 85 years. These statistics give a general idea of the age distribution of the patients. The median and mean being close to each other suggests that the age distribution is fairly symmetrical. The range from the minimum to the maximum age (18 to 85 years) indicates the breadth of ages covered.
to check the skewness
install the e1071 package
install.packages("e1071")## Installing package into 'C:/Users/fadar/AppData/Local/R/win-library/4.3'
## (as 'lib' is unspecified)## package 'e1071' successfully unpacked and MD5 sums checked
##
## The downloaded binary packages are in
## C:\Users\fadar\AppData\Local\Temp\Rtmp40pc2Z\downloaded_packagesto load the e1071 package
library(e1071)to check the skewness of age, Billing.Amount
age.skewness <- skewness(healthcare$Age)
print(age.skewness)## [1] -0.01214421Findings
The skewness value for the Age variable is -0.01214421, which is very close to 0. This small negative value indicates a very slight skew to the left. However, given how close the value is to zero, it suggests that the age distribution of the patients is almost symmetrical. In practical terms, this means that the ages of the patients are fairly evenly distributed around the median, with no significant skew towards younger or older ages
BillingAmount.skewness <- skewness(healthcare$Billing.Amount)
print(BillingAmount.skewness)## [1] 0.01271741Findings
BillingAmount, data does not have a pronounced long tail on either the right or left side and is approximately symmetric. This symmetry means that there aren’t extreme values (outliers) that are significantly distorting the distribution in one direction.
check if there is a relationship between the age and admission period
check if there is a relationship between billing amount and admission period
Unique Counts
Doctors <- unique(healthcare$Doctor)
View(Doctors)Medications <- unique(healthcare$Medication)
View(Medications)Test.Results <- unique(healthcare$Test.Results)
View(Test.Results)Frequency Analysis
table(healthcare$Test.Results)##
## Abnormal Inconclusive Normal
## 3456 3277 3267Findings_
The table indicates a relatively even distribution among the three categories of test results. This imply a diverse patient population with varying health conditions.
visualization
to load the ggplot2
library(ggplot2)Histograms
1. Age
2. Billing.Amount
ggplot(healthcare, aes(x = Age)) +
geom_histogram(bins = 30, fill = "grey", color = "black") +
ggtitle("Age Distribution") +
xlab("Age") +
ylab("Frequency")
ggplot(healthcare, aes(x = Billing.Amount)) +
geom_histogram(bins = 30, fill = "grey", color = "black") +
ggtitle("Billing Amount Distribution") +
xlab("Billing Amount") +
ylab("Frequency")
Bar chart
1. Gender
2. Blood.Type
3. Medical.Condition
ggplot(healthcare, aes(x = Gender, fill = Gender)) +
geom_bar() +
scale_fill_manual(values = c("black", "yellow")) +
ggtitle("Gender Distribution") +
xlab("Gender") +
ylab("Frequency")
Blood.Type
ggplot(healthcare, aes(x = Blood.Type)) +
geom_bar(fill = "green") +
ggtitle("Blood Type Distribution") +
xlab("Blood Type") +
ylab("Frequency")
Medical.Condition
ggplot(healthcare, aes(x = Medical.Condition)) +
geom_bar(fill = "grey") +
ggtitle("Medical Condition Distribution") +
xlab("Medical Condition") +
ylab("Frequency")
Calculate the length of stay for each patient
healthcare$Length.of.Stay <- as.numeric(difftime(healthcare$Discharge.Date, healthcare$Date.of.Admission, units="days"))
View(healthcare$Length.of.Stay)
head(healthcare$Length.of.Stay)## [1] 14 14 30 1 24 3summary(healthcare$Length.of.Stay)## Min. 1st Qu. Median Mean 3rd Qu. Max.
## 1.00 8.00 16.00 15.56 23.00 30.00Histogram for ‘Length of Stay’
ggplot(healthcare, aes(x = Length.of.Stay)) +
geom_histogram(bins = 30, fill = "yellow", color = "black") +
ggtitle("Length of Stay Distribution") +
xlab("Length of Stay (days)") +
ylab("Frequency")
Correlation Analysis
- Age
- Billing Amount
- Length of Stay
cor(healthcare[c("Age", "Billing.Amount", "Length.of.Stay")], use = "complete.obs")## Age Billing.Amount Length.of.Stay
## Age 1.000000000 -0.009483329 0.009111433
## Billing.Amount -0.009483329 1.000000000 -0.013506706
## Length.of.Stay 0.009111433 -0.013506706 1.000000000Findings
Age and Billing.Amount: The correlation coefficient is approximately -0.0095, indicating a very weak negative correlation between age and billing amount. This suggests that there’s almost no linear relationship between these two variables. Age and Length.of.Stay: The correlation coefficient is approximately 0.0091, suggesting a very weak positive correlation between age and length of stay. Again, this implies that there’s almost no linear relationship. Billing.Amount and Length.of.Stay: The correlation coefficient is approximately -0.0135, indicating a very weak negative correlation between billing amount and length of stay. This suggests there’s almost no linear relationship between these variables.
In summary, the output suggests that there are no strong linear relationships between age, billing amount, and length of stay among the patients. All these correlations are very close to zero, indicating that changes in one of these variables do not reliably predict changes in the others.
Grouped Analyses
Mean Age by Gender
aggregate(Age ~ Gender, data = healthcare, mean)## Gender Age
## 1 Female 51.60847
## 2 Male 51.29117Findings
The average age of both male and female patients are the same. All patient’s regardless of their gender fall within the same age group.
Mean Billing Amount by Medical Condition
aggregate(Billing.Amount ~ Medical.Condition, data = healthcare, mean)## Medical.Condition Billing.Amount
## 1 Arthritis 25187.63
## 2 Asthma 25416.87
## 3 Cancer 25539.10
## 4 Diabetes 26060.12
## 5 Hypertension 25198.03
## 6 Obesity 25720.84Finding
For instance, the data suggests that on average, diabetes has the highest associated billing amount, while arthritis and hypertension have relatively lower average billing amounts.
Average Length of Stay by Admission Type
aggregate(Length.of.Stay ~ Admission.Type, data = healthcare, mean)## Admission.Type Length.of.Stay
## 1 Elective 15.60117
## 2 Emergency 15.60974
## 3 Urgent 15.47656Finding
Elective: The average length of stay is approximately 15.60 days. Emergency: The average length of stay is approximately 15.61 days. Urgent: The average length of stay is approximately 15.48 days. These results suggest that the length of hospital stay is quite similar across the three types of admission, with very slight variations. In particular, there’s only a marginal difference in the average length of stay between elective and emergency admissions.
Summary
Overview of Patient Demographics and Admission Patterns
The analysis of the data set reveals insightful trends and characteristics about patient population and hospital admissions. Patient age range spans from 18 to 85 years, with a median and mean age close to 52 years, indicating a well-distributed age profile across young, middle-aged, and elderly patients. This diversity in age groups underscores the need for a wide range of medical services and specialized care to cater to the varied health needs of different age groups. Additionally, admissions data show a near-even split between genders, suggesting that healthcare services are accessed equally by both male and female patients. In terms of medical conditions, the average billing amounts vary slightly across different diagnoses such as Arthritis, Asthma, Cancer, Diabetes, Hypertension, and Obesity, but not significantly, pointing towards a consistent billing policy.
Analysis of Billing and Length of Stay
The financial aspect of healthcare services was examined through the billing amounts and length of stay associated with various medical conditions and admission types. The skewness of the billing amount distribution is remarkably low (0.0127), indicating a symmetrical distribution with no significant outliers, which suggests a consistent and equitable billing practice. This is crucial for maintaining trust and transparency with patients and their families. Furthermore, the average length of stay for patients, regardless of the admission type (Elective, Emergency, or Urgent), hovers around 15.5 days, suggesting effective and efficient patient care and hospital resource utilization. This consistency in length of stay, irrespective of the urgency of admission, demonstrates hospital’s commitment to providing timely and quality care to all patients.