30 day rope skip challenge.

Step 1: Get raw data

library(DBI)
library(RSQLite) 
library(tidyverse)

# Setup a working directory to copy files to for processing.
workDir <- "~/HealthData/FitFiles/working/"
# setwd(workDir)

con <- dbConnect(SQLite(), "/Users/amrit/HealthData/DBs/garmin_activities.db")

# dbListTables(con)

acts <- dbReadTable(con, 'activities')

# Path of the SQLite activities database.
path <- "/Users/amrit/HealthData/FitFiles/Activities/"

# Get .fit filenames based on running activity names.
cardio <-
  acts %>%
  filter(sub_sport == "cardio_training") %>%
  arrange(desc(start_time)) %>%
  # Allow partial activity date matching.
  filter(str_detect(start_time, "2020-")) %>%
  select(start_time, activity_id) %>%
  mutate(activity_id = paste0(path, activity_id, ".fit"))
 
save(cardio, file = "~/HealthData/FitFiles/working/cardio.Rdata")

# Copy .fit files to working directory.
walk(paste("cp", cardio$activity_id, workDir), system) # same as lapply w/ no output printed

Step 2: Convert data from binary to human readable form

getFitOut <- function() {

library(reticulate)
py_capture_output(py_run_string("import fitparse

# Load the FIT file
fitfile = fitparse.FitFile(r.fitFile)

# Iterate over all messages of type \"record\"
# (other types include \"device_info\", \"file_creator\", \"event\", etc)
for record in fitfile.get_messages(\"session\"):

    # Records can contain multiple pieces of data (ex: timestamp, latitude, longitude, etc)
    for data in record:

        # Print the name and value of the data (and the units if it has any)
        if data.units:
            print(\"{}, {}, {}\".format(data.name, data.value, data.units))
        else:
            print(\"{}, {}, NA\".format(data.name, data.value))
"))
}

Step 3: Format data for plotting

getJumpData <- function(jumpFit) {

# Run python file created by @mcandocia on GitHub @ https://github.com/mcandocia/examples/tree/master/convert_fit_to_csv 
# to convert .fit data to .csv to allow for processing in R.
  
# source_python("jumps.py", envir = NULL)

jump <- read_csv(jumpFit, col_name = c('type', 'value', 'units'))

jump %>%
  filter(
    type == "unknown_110" |
    type == "Jumping Effect" |
    type == "Jumps" |
    type == "avg_heart_rate" |
    type == "max_heart_rate" |
    type == "avg_temperature" |
    type == "max_temperature" |
    type == "start_time" |
    type == "sub_sport" |
    type == "total_anaerobic_training_effect" |
    type == "total_training_effect" |
    type == "total_calories" |
    type == "total_elapsed_time"
    ) %>%
  select(type, value) %>%
  pivot_wider(names_from = type) %>%
  type_convert(col_types = cols(.default = col_double(),
                                start_time = col_datetime(),
                                sub_sport = col_character(),
                                unknown_110 = col_character()
                                )
               ) %>%
  rename(name = unknown_110,
         jumpingEffect = "Jumping Effect",
         jumps = Jumps
         )
}

passFit <- function() {
  getJumpData(getFitOut())
}

#datalist = list()
# Advice from: https://adv-r.hadley.nz/control-flow.html#for-family
# Make sure to preallocate the output container,otherwise the loop will be very slow.
datalist <- vector("list", length(cardio$activity_id))
# Use seq_along(x) instead of 1:length(x). It always returns a value the same length as x.
for (i in seq_along(cardio$activity_id)) {
  fitFile <- cardio$activity_id[i]
  dat <- passFit()
  datalist[[i]] <- dat
}

jumpData <-
  bind_rows(datalist) %>%
  filter(jumps > 0) %>%
  mutate(jpm = jumps / total_elapsed_time * 60)

# Determine number of days in a row with a 10 min. jump rope session. Multiple entries per day should be excluded.

sessions <- unique(format(jumpData$start_time, "%A, %B %d, %Y"))

library(glue) # for convenient string/var concatenation

pltJumpData <- 
  jumpData %>%
  select(-sub_sport, -name) %>%
  pivot_longer(-start_time)

ggplot(pltJumpData, aes(start_time, value)) + 
  geom_point() +
  geom_smooth(method = loess, se = FALSE) +
  facet_wrap(~ name, scales = "free_y") +
  labs(title = glue("Summary of jump rope session data over {length(sessions)} days"),
       subtitle = "Goal is min of 10 min. continuous jumping per session",
       x = "Date",
       y = "Value") +
  theme(axis.text.x = element_text(angle = 45, hjust = 1))

Total training effect vs. average heart rate

This plot shows that what really matters for a cardio workout is getting your heart rate up and keeping it up. Obviously, this is more achievable the faster you jump, but jump rate is also affected by coordination. This is why there is no clear trend or clustering by jump rate.

As for training effect, the interpretation of this variable is as follows:

Also, there is both aerobic and anaerobic training effect. The plot above (and further below) shows only the aerobic training effect since it is really the only significant one. Therefore, for a jump rope session to be a good cardio workout, we should target to have a training effect >= 2.0 – that is, a workout that will allow us to at least maintain our current level of fitness. However, ideally, >= 3.0 would be desirable for cardio gains.

Moving from the top left corner of the plot (higher heart rate and lower jump rate) toward the lower right corner (lower heart rate and higher jump rate) shows an increase in jumping efficiency over time (though not necessarily cardio fitness as discussed above)⁴.

Average heart rate vs. total jumps by date

Again, the trend is clear by presenting the data this way with the change in color indicating the change in jumping efficiency.

It also shows that:

• The first couple of days were not very productive and
• Jumping inefficiency decreases very fast!

From the very first plot in this article,⁵ it seems that there is about a 10 bpm decrease in average heart rate every 2-3 days in the first week! This seems to level out as jumping efficiency increases over time leading to a higher jump rate, which causes a slight bump up in average heart rate.

Data normalization

Recalling that a jumping effect of 100 is equivalent to jumping 20 minutes non-stop at a 120+ jpm pace, we can combine session duration and jump rate variables into a single convenient normalizing parameter. This would allow for removal of arbitrary 10 minute session duration and ~1000 jumps target encouraging growth. This parameter isn’t ideal though as it treats any jump rate above 120 jpm the same. More data is needed at higher jumping effect values (i.e., longer/faster sessions) to say anything about when cardio benefits start to show up.

Using recent hilly training runs to look for cardio gains

Since I’ve been running while doing this jump rope cardio fitness experiment, there is a good opportunity to use the running data as a basis to look for cardio gains.

Looking at recent running data around the Lalitpur foothills near Kirtipur and Tribhuvan University,⁶ we might be able to detect changes in performance.

Using previous plotting tools developed to analyze and compare series’ of running data, here is a quick plot of some training runs with some variables of interest.

A quick glance indicates a high degree of similarity between these runs although it did rain that first day likely providing a significant cooling effect and slight performance bump.⁷

Since the running performance index (RPI)⁸ data is so noisy, it is easier to report in tabular form.

Now back to the question:

Is jumping rope for 10 minutes a day for 30 days straight improving my cardio fitness?

Over the same course of time running, we see that the average RPI (or running economy), which is a function of heart rate and pace, is about the same for each run under similar conditions indicating that my performance isn’t really improving.

Long story short, this matches what my Garmin Connect VO₂^max data indicates since I started running in Kathmandu this summer with a constant value of 48 ml O₂/kg-min.

Here is what the Garmin Connect app says about VO₂^max:

VO₂^max is an indication of your cardiovascular fitness and should increase as your level of fitness improves. VO₂^max is the maximum volume of oxygen (in milliliters) you can consume per minute per kilogram of body weight at your maximum performance. Your compatible Garmin device uses heart rate and exercise data to estimate your VO₂^max.

And what is really great (from the Forerunner 935 Owner’s Manual):

Your device adjusts the VO₂^max values for heat and altitude, such as when you are acclimating to high heat environments or high altitude.

So accounting for differences in elevation gives added confidence in the Garmin-calculated VO₂^max.

2020-08-12

I ran 6.72 miles in morning; average heart rate was 147 bpm.

2020-08-13

While jumping rope today, I naturally started to move my feet differently. I began alternating my feet instead of jumping both at the same time early on in the 10 minute session. This resulted in much less of a bounce height. This also led to much more consistency while jumping with minimal disruptions of jump streaks. I had the realization that the average heart rate data reflected in the plot above is likely more indicative of my body learning to jump more efficiently rather than my cardio fitness improving.

I’m also wondering what the impact of me running relatively long distances recently has on the jump rope data. Does this introduce an element of fatigue or hysteresis that causes my average heart rate data on running days to be biased high when it otherwise wouldn’t be? Needless to say, the first day I went running (yesterday) my calves felt really tight (due to jumping rope every day for a week). However, I noticed that in the morning today, they actually felt better (less tight and sore), as if running helped loosen them up. This is counter to what I would expect though because typically running causes me to have tighter calves. I looked to the internet to try and explain what was going on and found an article mentioning that:

A good way to stretch your calves is during your workout: Walking or running uphill is a great calf activator and naturally forces your muscles to stretch while your heading up.

Since the recent running route is about half uphill, I might be getting some benefit from this.

2020-08-14

I ran 5.88 miles in morning; average heart rate was 149 bpm.

2020-08-18

Need to account for temperature effects.

2020-08-19

Jumped in rain, which served as a cooling effect; temperature conditions similar to yesterday.

2020-08-20

Fever of 100.6-100.9 F measured around 8 pm. Temperature recorded as normal (98.5 F) when woke up next morning around 7 am.

2020-08-23

Twisted ankle playing soccer; no visible swelling, but a significant strain.