loader

Will an XGBoost reveal the same risk factors as those used by a physician?

Photo by Alexander Sinn on Unsplash

Feature Selection

Fig 1

Baseline Model

Fig 2: Confusion matrix for the baseline model

Hyperparameter Tuning

colsample_bytree:1,
learning_rate:0.1,
max_depth:4,
min_child_weight:1e-05,
n_estimators:200,
objective:’binary:logistic’,
subsample:0.5

Hyperparameter Tuned Model

Fig 3: Confusion matrix for the hyperparameter tuned model

Validity of Model

Fig 4: The top feature used by the XGBoost model to classify heart disease. F score is the number of times the model used a specific feature to split upon

Age

Fig 5: The total number of people within each age group. The percentage is the number of people with CVD.

Non-Modifiable Risk Factors.

Table 1 from [10]
Fig 6: Average Maximum Heart Rate within each age group with heart disease present
Fig 7: Average Maximum Heart Rate within each age group with no heart disease present


Source link

Leave a Reply