gluecose

The following dataset is obtained from here. The patients.csv contains:

• id: Integer, Primary Key
• age: Integer, patient’s current age
• name: String, patient’s name
• diabetes_type: String, either “type_1” or “type_2”, describes cause of diabetes
• join_date: Date when user started using the app. Format is YYYY-MM-DD

The by_reading.csv contains:

• id: Integer, Primary Key
• bg: Float, describes the patient’s Blood Glucose at time of measurement
• patient_id: Foreign key to the Patient Information table
• date: String, describes date on which measurement was taken. Format is YYYY-MM-DD

library(ggplot2)
library(dplyr)

## 
## Attaching package: 'dplyr'
## 
## The following object is masked from 'package:stats':
## 
##     filter
## 
## The following objects are masked from 'package:base':
## 
##     intersect, setdiff, setequal, union

reading = read.csv('bg_reading.csv', header=TRUE)
patient = read.csv('patients.csv', header=TRUE)

reading$age = c(1:dim(reading)[1])
reading$type = c(1:dim(reading)[1])

for (i in 1:dim(reading)[1]){
  reading$age[i] = patient[patient$id==reading$patient_id[i],c('age')]
  reading$type[i] = patient[patient$id==reading$patient_id[i],c('diabetes_type')]
}

head(reading)

##         date       bg    id patient_id age type
## 1 2006-08-28 85.84510 40817         28  26    1
## 2 2006-03-19 94.33209 14718         10  66    2
## 3 2005-08-28 90.36989 12761          9  51    2
## 4 2006-11-29 96.14667 57766         39  26    2
## 5 2007-09-28 75.95284 31844         22  49    2
## 6 2006-10-02 82.43022 10054          7  70    1

Here is a scatterplot of age vs mean blood gluecose reading:

age_grouped = reading  %>%  group_by(age) %>%
  summarise(mean_bg = mean(bg),
            n = n())

ggplot(age_grouped, aes(age, mean_bg)) +
  geom_point()

Here is a line plot of the same data:

ggplot(age_grouped, aes(age, mean_bg)) +
  geom_line()

If we look at the number of patients that fall into each age group, we can see that only a few groups have more than one patient:

table(patient$age)

## 
## 21 23 25 26 27 28 31 32 34 35 37 39 41 42 43 44 46 48 49 51 55 58 61 62 63 
##  1  1  1  2  1  1  1  2  2  2  3  3  2  2  1  2  1  2  2  2  1  1  1  1  3 
## 64 65 66 70 
##  3  2  1  2

Thus, the spikes shown at age 28, 41 and 55 are not representitive for that age in the population.

age_type_grouped = reading  %>%  group_by(age, type) %>%
  summarise(mean_bg = mean(bg),
            n = n())
age_type_grouped$type = as.factor(age_type_grouped$type)
ggplot(age_type_grouped, aes(age, mean_bg)) +
  geom_point(aes(color=type)) + 
  geom_line(aes(color=type))

It seems that for type 1 patients, they controlled their gluecose readings very well. But for type 2 patients, the gluecose level fluctuates.