This is the repository containing all of the code for the Heart Rate Monitor Project. The code is largely broken up into 2 distinct components: Reader.py and Processing.py. This distinction is larely for the purpose of organizing the commands together under categories as you don't need to run both of them to get the desired output. To simply run the program use: python Processing.py
Code Function:
Reader.py - This function performs all the initial reading and checking for the validity of files in the current repository.
First it finds all files ending in .csv and groups them together. Then it sorts through which of these files are true .csv files.
Finally it sorts out if any of the remaining valid files have any abnormal lines. An abnormal line is one where maybe one of the data values is missing or there is a string (of words not a string like '5').
It creates a list for each file that makes it to the end of the Reader.py file and the list contains a value for each row of the file. If the value is 1 that row is fine to be passed along. If the value is 0 then that row is marked for removal later on in the process. Its ultimate output is a dictionary containing the file names that passed the checks as keys and the list mentioned above as the value.
Processor.py- This function performs the processing and the output for the assignment. First it calls Reader.py to sort through all of the files in the current folder. It takes the dictionary output and then parses through to get two lists: time and voltage. It uses the list passed along with the file names to toss out bad rows. It also notifies the user at this point if any and how many rows were thrown out and if the values recorded in the lists were inside the normal bounds for an ECG. Then it calculates the minimum voltage, maximum voltage, and the duration of the recording.
Next it takes the derivative of the input graph using a simple, discrete slope calculator. Because the QRS complex is typically much sharper than the surrounding features this amplifies the differences and allows for an easier time thresholding for beat detection. This method also eliminates some of the issues with drift of a signal because the slope in between two points will likely not shift enough to cause an error in a derivative threshold whereas you may run into problems thresholding the raw signal.
The beat detection function can be modulated by the user for a different threshold if desired. Currently the method detects the largest derivative value then sets the threshold at 50% of that value. This value of 50% can be adjusted from (0,1).
The next function is the heart rate detector function. This calculates the heart rate based upon an input number of minutes. It currently has a default of 1 minute but can be adjusted by the user to really anything.
The final function outputs a json file that contains the dictionary of all the calculations made by the rest of the file. It changes the numpy array back to a list (because json can't convert a numpy array into a json string, who knew?). The output is a file that is titled the same as the input without the .csv file extension. For example, test_data1.csv -> test_data1.