I chose the max30100 because its features an integrated pulse oximetry and heart-rate monitor sensor. Unfortunately Maximi provide little documentation or application notes covering how best to determine heart rate or spo2 values using the data returned from the sensor. The sensor hosts an IR and RED LED that can be adjusted to provide pulses for spo2 and heart rate measurements. As a trade off between adequate data samples, i2c bus speed and post measurement hr/spo2 calculation processing the max30100 was configured to return 100 IR and RED values per second. Below is a graph showing raw data values gathered from the NEO at a 100 readings per second.
For spo2 calculation we have taken a simplified approach of taking the AC components of both signals and determining the ratio. The ratio is referenced in an memory table containing empirical sp02 values. Hence the spo2 isn't clinically accurate, for greater accuracy the table would normally be based on experimental measurements from healthy patients cross referenced against clinically accurate readings.
As shown in the graph the IR values are smoother than RED values possible due to secondary emissions from the RED LED. For this reason the IR value are normally used to determine the heart rate. The heart rate is calculated by feeding the IR values in to a first order 6Hz low pass filter which in turn are used to calculate the time interval between 2 peaks. Sample output of applying the low pass filter is shown below ignore the graph labels, top is IR values, bottom is low pass filter.
I also hooked up a SSD1306 oled display to the same i2c bus so that the calculated heart rate and spo2 values are displayed. The main challenge of this exercise has been to be ensure the code running on the M4 is as efficient as possible because there are many time critical elements such reading data samples, hr/spo2 calculation and display updates which can interfere with the output results. As with my previous examples on the imx6sx this was developed using the FreeRTOS sdk.