Análisis de factores de riesgo para enfermedades cardíacas

Carga de datos y librerías

phd <- read.csv("data/framingham.csv")

library(tidyverse)
library(corrplot)

Limpieza de datos

sapply(phd, function(x) sum(is.na(x)))

##            male             age       education   currentSmoker      cigsPerDay 
##               0               0             105               0              29 
##          BPMeds prevalentStroke    prevalentHyp        diabetes         totChol 
##              53               0               0               0              50 
##           sysBP           diaBP             BMI       heartRate         glucose 
##               0               0              19               1             388 
##      TenYearCHD 
##               0

phd <- na.omit(phd)

sapply(phd, function(x) sum(is.na(x)))

##            male             age       education   currentSmoker      cigsPerDay 
##               0               0               0               0               0 
##          BPMeds prevalentStroke    prevalentHyp        diabetes         totChol 
##               0               0               0               0               0 
##           sysBP           diaBP             BMI       heartRate         glucose 
##               0               0               0               0               0 
##      TenYearCHD 
##               0

write.csv(phd, "data/framingham_sin_NA.csv")

Negativos vs Positivos

phd %>%
  group_by(TenYearCHD) %>%
  summarize(count = n()) %>%
  mutate(perc = 100*count/sum(count)) %>%
  ggplot(aes(x = factor(TenYearCHD), y = perc)) +
  geom_col(aes(fill = TenYearCHD)) +
  theme_classic()+
  scale_x_discrete(labels = c("No TenYearCHD", "TenYearCHD")) +
  theme(axis.title.x = element_blank()) + ylab("% Outcome") + 
  theme(legend.position = "none") 

positivos <- phd %>%
  filter(TenYearCHD == 1) %>%
  summarise(n())

totales <- phd %>%
  summarise(n())

porcentaje_hd <- paste(round((positivos*100)/totales,2),"%")

porcentaje_hd

## [1] "15.24 %"

head(phd)

##   male age education currentSmoker cigsPerDay BPMeds prevalentStroke
## 1    1  39         4             0          0      0               0
## 2    0  46         2             0          0      0               0
## 3    1  48         1             1         20      0               0
## 4    0  61         3             1         30      0               0
## 5    0  46         3             1         23      0               0
## 6    0  43         2             0          0      0               0
##   prevalentHyp diabetes totChol sysBP diaBP   BMI heartRate glucose TenYearCHD
## 1            0        0     195 106.0    70 26.97        80      77          0
## 2            0        0     250 121.0    81 28.73        95      76          0
## 3            0        0     245 127.5    80 25.34        75      70          0
## 4            1        0     225 150.0    95 28.58        65     103          1
## 5            0        0     285 130.0    84 23.10        85      85          0
## 6            1        0     228 180.0   110 30.30        77      99          0

Análisis exploratorio

Matriz de correlación

corrplot(cor(phd), type = "full")



male

phd <- phd %>%
  mutate(male = ifelse(male == 1, "Male", "Female"))

unique(phd$male)

## [1] "Male"   "Female"

  phd %>%
  group_by(male, TenYearCHD) %>%
  summarize(count = n()) %>%
  mutate(perc = 100*count/sum(count)) %>%
  filter(TenYearCHD == 1)

## # A tibble: 2 × 4
## # Groups:   male [2]
##   male   TenYearCHD count  perc
##   <chr>       <int> <int> <dbl>
## 1 Female          1   250  12.3
## 2 Male            1   307  18.9

phd %>%
  group_by(male, TenYearCHD) %>%
  summarize(count = n()) %>%
  mutate(perc = 100*count/sum(count)) %>%
  ggplot(aes(x = factor(TenYearCHD), y = perc)) +
  geom_col(aes(fill = TenYearCHD)) +
  theme_classic()+
  scale_x_discrete(labels = c("No TenYearCHD", "TenYearCHD")) +
  theme(axis.title.x = element_blank()) + ylab("% Outcome by male") + 
  theme(legend.position = "none") +
  facet_wrap(~ male)



age

ggplot(phd, aes(age, TenYearCHD))+
  geom_jitter(alpha = 0.5)+
  geom_smooth(method = "glm", se = FALSE, method.args = list(family = binomial)) +
  theme_bw()

## `geom_smooth()` using formula = 'y ~ x'



  phd %>%
  group_by(age, TenYearCHD) %>%
  summarize(count = n()) %>%
  mutate(perc = 100*count/sum(count)) %>%
  arrange(desc(age)) %>%
  filter(TenYearCHD == 1) %>%
  ggplot(aes(age, perc)) +
    geom_point()+
    geom_smooth()+
    theme_bw()+
    ylab("% TenYearCHD") 



phd <- phd %>%
  mutate(AgeGroup = case_when(age < 40 ~ "18_39", age < 50 ~ "40_49", age < 61 ~ "50_60",  TRUE ~ "61+" ))

  phd %>%
  group_by(age, TenYearCHD) %>%
  summarize(count = n()) %>%
  mutate(perc = 100*count/sum(count)) %>%
  arrange(desc(age)) %>%
  filter(TenYearCHD == 1) %>%
  ggplot(aes(age, perc)) +
    geom_point()+
    geom_smooth()+
    theme_bw()+
    ylab("% TenYearCHD") 



education

unique(phd$education)

## [1] 4 2 1 3

ggplot(phd, aes(education, TenYearCHD)) + 
  geom_col()



phd %>%
  group_by(education, TenYearCHD) %>%
  summarize(count = n()) %>%
  mutate(perc = 100*count/sum(count)) %>%
  ggplot(aes(x = factor(TenYearCHD), y = perc)) +
  geom_col(aes(fill = TenYearCHD)) +
  theme_classic()+
  scale_x_discrete(labels = c("No TenYearCHD", "TenYearCHD")) +
  theme(axis.title.x = element_blank()) + ylab("% Outcome by education") + 
  theme(legend.position = "none") +
  facet_wrap(~ education)



currentSmoker

ggplot(phd, aes(currentSmoker, TenYearCHD)) + 
  geom_col()



phd %>%
  group_by(currentSmoker,TenYearCHD ) %>%
  summarize(count = n()) %>%
  mutate(perc = 100*count/sum(count)) %>%
  ggplot(aes(x = factor(TenYearCHD), y = perc)) +
  geom_col(aes(fill = TenYearCHD)) +
  theme_classic()+
  scale_x_discrete(labels = c("No TenYearCHD", "TenYearCHD")) +
  theme(axis.title.x = element_blank()) + ylab("% Outcome by currentSmoker") + 
  theme(legend.position = "none") +
  facet_wrap(~ currentSmoker)



cigsPerDay

ggplot(phd, aes(cigsPerDay, TenYearCHD)) + 
  geom_jitter(alpha = 0.5)+
  geom_smooth(method = "glm", se = FALSE, method.args = list(family = binomial)) +
  theme_bw()



BPMeds

unique(phd$BPMeds)

## [1] 0 1

phd %>%
  group_by(BPMeds,TenYearCHD ) %>%
  summarize(count = n()) %>%
  mutate(perc = 100*count/sum(count)) %>%
  ggplot(aes(x = factor(TenYearCHD), y = perc)) +
  geom_col(aes(fill = TenYearCHD)) +
  theme_classic()+
  scale_x_discrete(labels = c("No TenYearCHD", "TenYearCHD")) +
  theme(axis.title.x = element_blank()) + ylab("% Outcome by BPMeds") + 
  theme(legend.position = "none") +
  facet_wrap(~ BPMeds)



prevalentStroke

phd %>%
  group_by(prevalentStroke,TenYearCHD ) %>%
  summarize(count = n()) %>%
  mutate(perc = 100*count/sum(count)) %>%
  ggplot(aes(x = factor(TenYearCHD), y = perc)) +
  geom_col(aes(fill = TenYearCHD)) +
  theme_classic()+
  scale_x_discrete(labels = c("No TenYearCHD", "TenYearCHD")) +
  theme(axis.title.x = element_blank()) + ylab("% Outcome by prevalentStroke") + 
  theme(legend.position = "none") +
  facet_wrap(~ prevalentStroke)



prevalentHyp

phd %>%
  group_by(prevalentHyp,TenYearCHD ) %>%
  summarize(count = n()) %>%
  mutate(perc = 100*count/sum(count)) %>%
  ggplot(aes(x = factor(TenYearCHD), y = perc)) +
  geom_col(aes(fill = TenYearCHD)) +
  theme_classic()+
  scale_x_discrete(labels = c("No TenYearCHD", "TenYearCHD")) +
  theme(axis.title.x = element_blank()) + ylab("% Outcome by prevalentHyp") + 
  theme(legend.position = "none") +
  facet_wrap(~ prevalentHyp)



diabetes

phd %>%
  group_by(diabetes,TenYearCHD ) %>%
  summarize(count = n()) %>%
  mutate(perc = 100*count/sum(count)) %>%
  ggplot(aes(x = factor(TenYearCHD), y = perc)) +
  geom_col(aes(fill = TenYearCHD)) +
  theme_classic()+
  scale_x_discrete(labels = c("No TenYearCHD", "TenYearCHD")) +
  theme(axis.title.x = element_blank()) + ylab("% Outcome by diabetes") + 
  theme(legend.position = "none") +
  facet_wrap(~ diabetes)



totChol

  phd %>%
  group_by(totChol, TenYearCHD) %>%
  summarize(count = n()) %>%
  mutate(perc = 100*count/sum(count)) %>%
  arrange(desc(totChol)) %>%
  filter(TenYearCHD == 1) %>%
  ggplot(aes(totChol, perc)) +
    geom_point()+
    geom_smooth()+
    theme_bw()+
    ylab("% TenYearCHD")



sysBP

  phd %>%
  group_by(sysBP, TenYearCHD) %>%
  summarize(count = n()) %>%
  mutate(perc = 100*count/sum(count)) %>%
  arrange(desc(sysBP)) %>%
  filter(TenYearCHD == 1) %>%
  ggplot(aes(sysBP, perc)) +
    geom_point()+
    geom_smooth()+
    theme_bw()+
    ylab("% TenYearCHD")



diaBP

  phd %>%
  group_by(diaBP, TenYearCHD) %>%
  summarize(count = n()) %>%
  mutate(perc = 100*count/sum(count)) %>%
  arrange(desc(diaBP)) %>%
  filter(TenYearCHD == 1) %>%
  ggplot(aes(diaBP, perc)) +
    geom_point()+
    geom_smooth()+
    theme_bw()+
    ylab("% TenYearCHD")



BMI

  phd %>%
  group_by(BMI, TenYearCHD) %>%
  summarize(count = n()) %>%
  mutate(perc = 100*count/sum(count)) %>%
  arrange(desc(BMI)) %>%
  filter(TenYearCHD == 1) %>%
  ggplot(aes(BMI, perc)) +
    geom_point()+
    geom_smooth()+
    theme_bw()+
    ylab("% TenYearCHD")



heartRate

  phd %>%
  group_by(heartRate, TenYearCHD) %>%
  summarize(count = n()) %>%
  mutate(perc = 100*count/sum(count)) %>%
  arrange(desc(heartRate)) %>%
  filter(TenYearCHD == 1) %>%
  ggplot(aes(heartRate, perc)) +
    geom_point()+
    geom_smooth()+
    theme_bw()+
    ylab("% TenYearCHD")



glucose

  phd %>%
  group_by(glucose, TenYearCHD) %>%
  summarize(count = n()) %>%
  mutate(perc = 100*count/sum(count)) %>%
  arrange(desc(glucose)) %>%
  filter(TenYearCHD == 1) %>%
  ggplot(aes(glucose, perc)) +
    geom_point()+
    geom_smooth()+
    theme_bw()+
    ylab("% TenYearCHD")