Statistical Process Improvement		 Georgetown University
 

Analysis of Watch Data

 

Assigned Reading

Presentations

  • Analyze heart rate while controlling for physical activity Slides► (Coming soon)

Assignments

Instruction for Submission of Assignments: Assignments should be submitted directly on Blackboard.  In rare situations assignments can be sent directly by email to the instructor. Submission should follow these rules:

  1. Include a summary sheet/page and compare your answers to answers posted

Question 1: Watch data include both heart rates and exercise level. Unfortunately heart data is not always accurate as exercise levels affect heart rate measures and interfere with the ability of the watch to alert for unusual watch rates.  Use a risk adjusted control chart to analyze heart rate. Compare observed heart rate to heart rate expected from exercise.  This analysis requires the following steps:

  • Data: Download and merge data from a watch, or use data available here.  Steps to download data from a watch are available on the web.
  • Data Frame: Merge the data into one data frame.  Plot the time series.
  • Analysis Time:  Select a timeframe for developing regression model (e.g., last week excluding yesterday) and a later timeframe (e.g., yesterday) for testing accuracy of the model.
  • Observations per Unit of Analysis: Decide what type of control chart you plan to create by deciding unit of analysis.  If you are looking at multiple observations per time period then risk-adjusted XBar chart is appropriate.  If you are looking at 1 observation per time period then Tukey chart done on expected values is appropriate.
  • Expected Heart Rate Values:  Regress heart rate on calorie consumption, steps, respiratory rate, body motion (e.g., sedentary .vs. active) and other available variables.
    • Verify that heart rate has a normal distribution by creating a histogram of the heart rate data.  If data is not normal come up with transformations of the data that will make the data more normal.
    • Report the R-square of the regression.  If low or moderate, include pairwise and triple-way interaction terms in the analysis.
    • Calculate expected heart rate
  • Create Control Chart:  Create a control chart.  If you have multiple observations for each unit of time, then you should use risk-adjusted control chart.  If you have a single observation per time period then use Tukey chart.  If the number of observations in each time period is not constant, then control limits should not be a straight line and should change based on number of observations.  In reporting the control chart, include information on type of chart, parameters of the control chart and provide the chart.

Here are some resources that might be useful. Please note that some student reports may not be complete. 

In the following chart of heart rate on September 30th 2022, heart rate is high at start of the day, drops in middle, and increases towards the end of the day. The increase in the end of the day is within expectation, as heart rate for the most part remains within expected heart rate limits.  This chart was created using risk adjusted XBar procedures

Heart Rate vs Expected Heart Rate

 

More
 
Copyright © 1996 b> Farrokh Alemi, Ph.D. Most recent revision 12/09/2022.  This page is part of the course on Statistical Process Improvement, this is the lecture on Introduction to the Course.