Machine Learning/Predictive Analysis for Recurring Myocardial Infarction and Cardiac Arrest
Project: Program, in Matlab, to calculate the probability of patients incurring or recurring myocardial infraction (MI) or cardiac arrest (CA) based on training data including patient history file and normal ECG signals.
- Train data includes ECG signals with patients with no history of MI or CA,
- Further categorization based on patient history including gender, smoker, hypertension, previous MI/CA conditions
- Normal patients' training data: normalization and undergoes Fourier transform (FTT) to detect ECG signals peaks and establish a pattern for PRQST positions and amplitudes
- Cross-correlate signals between normal patients and patients with history of CA and MI to differentiate and map patterns due to history (smoker, hypertension)
- Establish a threshold for patients to start incurring/recurring MI or CA
- Build a GUI for easy user- friendly interface
Some examples:
![ml ca ma 4](https://brianchami.com/wp-content/uploads/2019/11/Screen-Shot-2019-11-25-at-1.20.13-AM-1024x900.png)
ECG signals for a known patient: normal male.
![ml ca ma 5](https://brianchami.com/wp-content/uploads/2019/11/Screen-Shot-2019-11-25-at-1.20.24-AM-1024x898.png)
ECG signals for a known patient: normal female.
![ml ca ma 6](https://brianchami.com/wp-content/uploads/2019/11/Screen-Shot-2019-11-25-at-1.20.36-AM-1024x893.png)
ECG signals for an unknown patient. Prediction: There’s a high chance that the patient will incurring/recurring MI or CA
![ml ca ma 7](https://brianchami.com/wp-content/uploads/2019/11/Screen-Shot-2019-11-25-at-1.20.53-AM-1024x900.png)
ECG signals for an unknown patient. Prediction: There’s a very low chance that the patient will incurring/recurring MI or CA
Graphic User Interface:
![ml ca ma 2](https://brianchami.com/wp-content/uploads/2019/11/Screen-Shot-2019-11-25-at-1.59.00-AM-1024x479.png)
GUI with some examples, including options to choose risk factors, gender, and whether to normalize the signals.