Python EDA looking at the physiological factors contributing to heart attacks for over 300 patients. After looking at multiple ML predictive models, able to predict the likelihood of a heart attack with up to 92%
Link to Heart Attack Analysis project: https://github.com/petermartens98/Heart-Attack-Analysis-and-Predictive-Models
The objective is to build a regressor model which can predict the heart rate of an individual.
This code defines a neural network model using Keras to predict the heart rate of an individual based on signals measured using ECG. It uses a training dataset to train the model and a testing dataset to evaluate its performance.
The first two functions, plot_corr_heatmap and plot_distribution, are helper functions to visualize the correlation matrix and distribution of the dataset respectively. The dataset is then split into train and test sets, and the target variable is set to "HR", which represents the heart rate.
The input data is normalized using StandardScaler to scale the values to have zero mean and unit variance. The model architecture consists of three fully connected (dense) layers with ReLU activation function, with each layer followed by a dropout layer to prevent overfitting. The output layer is a single node with linear activation function, as this is a regression problem.
The model is then compiled using the Adam optimizer and mean squared error (MSE) as the loss function, with mean absolute error (MAE) as the metric to monitor during training. The model is trained for 100 epochs with a batch size of 32 and a validation split of 0.2.
The performance of the model is evaluated on the test set using the evaluate function, which computes the test loss and MAE. The y_pred variable is used to store the predicted values for the test set, and a scatter plot of actual versus predicted values is created to visualize the model's performance.
The final section of the code uses scikit-learn's GridSearchCV to perform hyperparameter tuning on the model. It creates a parameter grid with different combinations of activation functions, dropout rates, and weight decay values, and searches over these hyperparameters to find the combination that results in the best performance. The create_model function is defined to create a Keras model with the given hyperparameters, and the KerasRegressor object is used to wrap this model for use in scikit-learn.
The data comprises various attributes taken from signals measured using ECG recorded for different individuals having different heart rates at the time the measurement was taken. These various features contribute to the heart rate at the given instant of time for the individual.
There are total of 6 CSV files with the names as follows: time_domain_features_train.csv - This file contains all time domain features of heart rate for training data frequency_domain_features_train.csv - This file contains all frequency domain features of heart rate for training data heart_rate_non_linear_features_train.csv - This file contains all non linear features of heart rate for training data
time_domain_features_test.csv - This file contains all time domain features of heart rate for testing data frequency_domain_features_test.csv - This file contains all frequency domain features of heart rate for testing data heart_rate_non_linear_features_test.csv - This file contains all non linear features of heart rate for testing data
Following is the data dictionary for the features you will come across in the files mentioned:
MEAN_RR - Mean of RR intervals
MEDIAN_RR - Median of RR intervals
SDRR - Standard deviation of RR intervals
RMSSD - Root mean square of successive RR interval differences
SDSD - Standard deviation of successive RR interval differences
SDRR_RMSSD - Ratio of SDRR / RMSSD
pNN25 - Percentage of successive RR intervals that differ by more than 25 ms
pNN50 - Percentage of successive RR intervals that differ by more than 50 ms
KURT - Kurtosis of distribution of successive RR intervals / "tailedness"
SKEW - Skew of distribution of successive RR intervals
MEAN_REL_RR - Mean of relative RR intervals
MEDIAN_REL_RR - Median of relative RR intervals
SDRR_REL_RR - Standard deviation of relative RR intervals
RMSSD_REL_RR - Root mean square of successive relative RR interval differences
SDSD_REL_RR - Standard deviation of successive relative RR interval differences
SDRR_RMSSD_REL_RR - Ratio of SDRR/RMSSD for relative RR interval differences
KURT_REL_RR - Kurtosis of distribution of relative RR intervals
SKEW_REL_RR - Skewness of distribution of relative RR intervals
uuid - Unique ID for each patient
VLF - Absolute power of the very low frequency band (0.0033 - 0.04 Hz)
VLF_PCT - Principal component transform of VLF
LF - Absolute power of the low frequency band (0.04 - 0.15 Hz)
LF_PCT - Principal component transform of LF
LF_NU - Absolute power of the low frequency band in normal units
HF - Absolute power of the high frequency band (0.15 - 0.4 Hz)
HF_PCT - Principal component transform of HF
HF_NU - Absolute power of the highest frequency band in normal units
TP - Total power of RR intervals
LF_HF - Ratio of LF to HF
HF_LF - Ratio of HF to LF
SD1 - Poincaré plot standard deviation perpendicular to the line of identity
SD2 - Poincaré plot standard deviation along the line of identity
Sampen - sample entropy which measures the regularity and complexity of a time series
higuci - higuci fractal dimension of heartrate
datasetId - ID of the whole dataset
condition - condition of the patient at the time the data was recorded
HR - Heart rate of the patient at the time of data recorded
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent