UTS Hospital Data Analysis Report

This report analyzes data from UTS Hospital, focusing on the relationship between ICU hours, length of stay, and marital status, as well as differences in length of stay between genders. The analysis uses descriptive statistics, frequency distributions, and correlation analysis. The results indicate a moderate positive linear relationship between ICU hours and length of stay, with no significant difference in median ICU hours across marital statuses or median length of stay between genders. The report recommends further analysis using regression models and additional demographic variables.

Table of Contents
Introduction and Objective..........................................................................................................................2
Literature Review........................................................................................................................................2
Methods......................................................................................................................................................2
Results.........................................................................................................................................................3
Discussion....................................................................................................................................................9
Conclusion and recommendation................................................................................................................9
References.................................................................................................................................................10

Introduction and Objective
The details concerning the variables gender, marital status, length of stay, and ICU hours
for UTS Hospital is given in the data.
The objective of this report is to know the relationship of ICU hours with the length of
stay and marital status. Another area of concern is to know if there is any significant difference
in the average length of stay between male and female.
Literature Review
An article "Predict hourly patient discharge probability in intensive care units using data
mining" written by Portela Filipe, Veloso Rui, Santos Manuel Filipe, Machado José Manuel,
Abelha António, Silva Álvaro, Rua Fernando, Oliveira Sérgio Manuel Costa was issued in the
year 2014. According to this article prediction regarding the probability of a patient to be
discharged in the next hour was made by constructing and model on the basis of length of stay
and intensive care units (ICU). On the basis of this Model, the bed Occupancy rate in the ICU
was also predicted for the next hour.
Methods
The method of descriptive statistics is used to understand the data better. Pie chart and
frequency distributions are used to know the demographics and descriptive. The bar graph is
used as a graphical representation of the objective.

Excel software is used to obtain the output. Functions like AVERAGE, STDEV,
MEDIAN, CORREL, MODE, Pivot tables, and charts are used to obtain the results.
Results
The frequency distribution and the corresponding pie chart for gender are given below.
Row Labels
Count of
Gender
Female 20200
Male 14424
Grand Total 34624
58%
42%
Pie chart
Female
Male
The percentage of females in the data is Greater as compared to the percentage of males.
There is 58 percent of females as compared to 42% of males.
Frequency distribution of marital status is given below.
Row Labels
Count of
MaritalStatus
Divorced 1003
Married 15400
Seperated 503
Single 13678
Unknown 1151
Widowed 2889

Grand Total 34624
3%
44%
1%
40%
3% 8%
Pie chart
Divorced
Married
Seperated
Single
Unknown
Widowed
There are 45% of married people, followed by 40% singles. The percentage of divorced
and unknown is only 3% in the survey. It is also observed that there are 1% separated people in
the data.
The frequency distribution for the level of stay and the corresponding histogram is given below.
Row Labels
Count of
LOS
1-20 33473
21-40 826
41-60 198
61-80 78
81-100 27
101-120 11
121-140 6
141-160 1
161-180 2
181-200 1
221-240 1
Grand Total 34624

1-20 21-40 41-60 61-80 81-100 101-120 121-140 141-160 161-180 181-200 221-240
0
5000
10000
15000
20000
25000
30000
35000
40000
histogram
length of stay
Frequency
The histogram indicates that the distribution of length of the day is positively skewed.
There are very few observations with a high length of stay.
The frequency distribution and the corresponding histogram for ICU hours are given below.
Row Labels Count of ICUHours
0-299 34354
300-599 149
600-899 60
900-1199 25
1200-1499 16
1500-1799 5
1800-2099 9
2100-2399 4
2400-2700 2
Grand Total 34624

0-299 300-599 600-899 900-1199 1200-
1499 1500-
1799 1800-
2099 2100-
2399 2400-
2700
0
5000
10000
15000
20000
25000
30000
35000
40000
Histogram
ICU Hours
frequency
The distribution of ICU hours is positively skewed. This indicates that there are very few
patients with high ICU hours in the data.
Scatterplot depicting the relationship between the length of stay and ICU hours is given below.
0 50 100 150 200 250
0
500
1000
1500
2000
2500
3000
f(x) = 4.17683680168092 x − 9.56723634108446
R² = 0.184106544041978
Scatterplot
Length of Stay
ICU hours

An upward trend is observed in the gutter plot and all the points are plotted moderately
close to each other. This indicates that there is a moderate positive linear relationship between
ICU hours and the length of stay. This indicates that as the value of the length of stay increases,
the value of ICU hours also increases. The value of correlation Coefficient being equal to 0.42
indicates that there is a moderate positive linear relationship between ICU hours and length of
stay.
The descriptive statistics for ICU hours for various categories of marital status is given below.
ICUHours_Div
orced
ICUHours_m
arried
ICUHours_sep
arated
ICUHours_s
ingle
ICUHours_un
known
ICUHours_wi
dowed
mea
n 3 3 4 16 3 4
medi
an 0 0 0 0 0 0
mod
e 0 0 0 0 0 0
sd 23.1784 35.0549 37.01495 115.4175 36.71462 32.59835
var 537.2384 1228.846 1370.107 13321.2 1347.963 1062.652
IQR 0 0 0 0 0 0
skew 13.09666 28.04309 16.5356 11.7632 20.49593 13.42688
From the value of skewness, it is observed that the distribution of ICU hours for all the
categories of marital status is positively skewed. The best measure of Central tendency is
median. There is not a much significant difference in the median ICU hours between Divorced,
married, separated, single, unknown, and widowed.
The average length of stay for each category of gender is summarised in the table and the
bar graph below.
LOS_female LOS_male
mean 4.157277 4.477121
median 2 2

mode 1 1
sd 7.39796 8.707255
var 54.7298 75.8163
IQR 3 3
skew 6.569049 7.315787
LOS_female LOS_male
0
0.5
1
1.5
2
2.5
Bar Graph
median
The average length of stay for males is slightly greater as compared to the average length
of stay for females. But the value of standard deviation for the length of stay for each gender is
high indicating that then mean is not reliable. The distribution of length of stay for male as well
as female is observed to be positively skewed. This indicates that the best measure of Central
tendency and dispersion is median and interquartile range respectively. There is no significant
difference in the median length of stay between male and female. The low value of the
interquartile range indicates that the median length of stay is a reliable measure of Central
Tendency.
Discussion
The method of regression analysis should be used to build a linear regression model. The
model is expected to be in the form of, ICU hours = Bo + b1* length of stay. The concept of

one way ANOVA can be used to test the significant difference in the average ICU hours between
the various types of marital status. T-test for independent samples can also be used to test the
significant difference in the average length of stay between male and female.
Conclusion and recommendation
The percentage of females in the data is Greater as compared to the percentage of males.
There are 45% of married people, followed by 40% singles. There are very few observations
with a high length of stay. There are very few patients with high ICU hours in the data.
There is a moderate positive linear relationship between ICU hours and length of stay.
This indicates that as the value of the length of stay increases, the value of ICU hours also
increases.
There is not a much significant difference in the median ICU hours between Divorced,
married, separated, single, unknown, and widowed.
There is no significant difference in the median length of stay between male and female.
The low value of the interquartile range indicates that the median length of stay is a reliable
measure of Central Tendency.
It is recommend that other demographic variables like income, location, occupation, age
should also be considered. Multiple linear regression model should be used to build the model
includes all these recommended variables. Stepwise regression model, adjusted R^2 should be
used to select the best model among these.

References
Portela, F., Veloso, R., Santos, M. F., Machado, J. M., Abelha, A., Silva, Á.,... & Oliveira, S. M.
C. (2014). Predict hourly patient discharge probability in Intensive Care Units using Data
Mining. ScienceAsia.
Wildemuth, B. M. (Ed.). (2016). Applications of social research methods to questions in
information and library science. ABC-CLIO.
Chatfield, C. (2018). Statistics for technology: a course in applied statistics. Routledge.
Chatterjee, S., & Hadi, A. S. (2015). Regression analysis by example. John Wiley & Sons.