Python Metabolism and Weight data

Thu 2025/03/27 Python python / diet / weight / samsung health / withings

Overview

• installation: see https://arrizza.com/setup-common.html

Summary

This project is to gather and aggregate data about activity and weight.

Get raw data

Withings Body+ scale

• Login to Withings Login
• click on Support in the menubar
• enter "export data"
• click on "Privacy - How can I export my data?"
• click on the provided link "... all of your data at once by clicking this link" "https://account.withings.com/export/user_select"
• click "Start my archive"
• After a few minutes, you will get an email with a link
• download the zip file into ~/Downloads
• extract the data
• copy these files to metabolism-weight/data
  • weight.csv

The weight.csv should have the following columns:

• Date
• Weight (lb)
• Fat mass (lb)
• Bone mass (lb)
• Muscle mass (lb)
• Hydration (lb)
• Comments

Samsung Galaxy 4 Watch

• on your phone, go to Samsung Health
• go to settings
• scroll to "Download personal data", click it
• download the data to your phone
• plug your phone to your PC via USB-C
• On "allow access to phone data?" click Allow
• The phone data should not be mounted on your PC
• go to Internal Storage/Download/Samsung Health
• there should be a file: samsunghealth_smartthings_[todays-date-time-stamp]
• copy these files from folder to metabolism-weight/data (note: very slow!)
  • com.samsung.shealth.activity.day_summary.[todays-date-time-stamp]
  • you can ignore the "calories_burned.details" file
• you can delete the older files

Run

./doit
./doit --verbose     # to see the daily data
./doit 2025          # only load data for the given year; defaults to 2025
./doit 2024
./doit --meta        # find maintenance caloric intake (see below "Maintenance calories")

Typical Output:

<snip>
49 2024-12-08 Sun  78,678   5,546  224.2   76.0  140.7 
50 2024-12-15 Sun  65,448   4,717  225.0   76.9  140.7 
51 2024-12-22 Sun  70,057   4,844  224.3   77.7  139.3 
52 2024-12-29 Sun  49,015   5,975  225.9   75.5  142.8 
Total steps   :  3,595,963
avg steps/day :      9,825
Total calories:    257,545
avg cal/day   :        704

The columns are:

• week number: from 1 to 53 (see isocalendar())
• date: this is weekly summary so it will be on Sundays
• steps: the number of steps for the week
• calories: the number of calories reported by Samsung Galaxy for the week; It takes 13 - 15 steps per calorie.
• weight: the value reported by the Withings Scale, in this case pounds
• fat: the value reported by the Withings Scale, in this case pounds
• muscle: the value reported by the Withings Scale, in this case pounds

The summary shows that in the last 4 weeks of 2024, I've gained 0.5 pounds of fat and added 2 pounds of muscle. which results in about 2.7 lbs weight. In 2024, I've walked 3.6 million steps and burnt just under 260,000 calories by walking and exercising. That turns out to be about 704 calories every day.

Note the weight, fat and muscle values are an average of values for the week.

Spreadsheet

A .xlsx spreadsheet is generated. There are three worksheets.

Weekly Summary worksheet

This shows the weekly summary on each Sunday of the year. The values are averages for the 7 days of the week.

Week Date	        Steps   Cals    Fat     Weight  Muscle	Score
1	2024-01-07 Sun	55,289	3,403	68.8	224.6	148.0	E -++
2	2024-01-14 Sun	54,528	3,541	72.9	221.8	141.5	F -+-
3	2024-01-21 Sun	70,260	4,532	72.9	220.8	140.6	C ++-
[snip]
41	2024-10-13 Sun	75,023	6,811	71.8	221.2	142.0	A +++
42	2024-10-20 Sun	77,305	5,733	71.7	221.0	141.9	C ++-
43	2024-10-27 Sun	59,554	3,688	71.3	221.7	143.0	B +-+

The "Score" column indicates a rough A through H classification. A and B are pretty good weeks, the rest are not as good. The three "+/-" shows whether the Fat, Weight and Muscle changed in a good way i.e. Fat went down, weight stayed the same or went down and Muscle stayed the same or went up.

Daily Summary worksheet

This shows the daily summary for each day of the year. The values are averages for all measurements taken during the day.

Date           Week	 Steps	Cal	    Fat   Weight  Muscle
2024-01-01 Mon  1    2,760    130    0.0     0.0     0.0
2024-01-02 Tue  1   10,603    623    0.0     0.0     0.0
2024-01-03 Wed  1   16,172  1,160   70.4   224.7   146.6
[snip]

Stats worksheet

This shows various status over the course of the year.

          Year Total	Avg. Week   Avg. Daily	  Min     Max
Steps      3,038,807     70,669.9     10,095.7   354.0	23,877.0
Calories     216,647      5,038.3        719.8    17.9   2,441.9
Weight           N/A          N/A        223.5   218.5     229.7
Fat              N/A          N/A         72.8    66.4      77.8
Muscle           N/A          N/A        143.2   134.7     150.0

• Year Total: total of all measurements for the entire year
• Avg. Week: the average of week measurements
• Avg. Daily: the average per daily measurements
• Min: the minimum measurement found
• Max: the maximum measurement found

Maintenance calories

Start with calories_in == calories_out.

calories_in == calories_out        # must balance
calories_in - calories_out == 0                    

Assume that calories can only be converted into fat or muscle or are used for exercise or maintenance.

• we don't want fat_gain since that represents too many calories_in
• we do want muscle_gain since that represents conversion of calories_in into muscle
• we do want calories_exercise since we want to spend more calories on exercising

0 ==        # still has to balance to zero 
   calories_in                    # from MyFitnessPal
   - fat_gain_lb * 3500 cal_lb    # assume fat loss is -ve value; from scale
   + muscle_gain_lb * 800 cal_lb  # assume muscle loss is +ve value; from scale
   + calories_exercise            # from watch data
   - calories_maintenance         # value to calculate

Now, solve for calories_maintenance:

# add calories_mainteance to both sides
calories_maintenance = 
  calories_in                   
  - fat_gain_lb * 3500 cal_lb     
  + muscle_gain_lb * 800 cal_lb
  + calories_exercise

Sum all of these and divide by the number of days in the data:

total_calories_maintenance = 
   sum_of(calories_in)
   - sum_of(fat_gain_lb * 3500 cal_lb)
   + sum_of(muscle_gain_lb * 800 cal_lb)
   + sum_of(calories_exercise)

daily_calories_maintenance = total_calories_maintenance / #days 

If there is data missing for specific days, ignore them. This will make them resolve to the average i.e. to daily_calories_maintenance.

Typical output:

$ ./doit --meta
<snip>
==== Calculating maintenance calories for 2025
     daily cal-in:    1393 exercise:     494 fat:     106 muscle:      -1 
     Maintenance calories:     1779 per day over days: 68

• Daily cal-in: the daily number of calories consumed
• Exercise : the daily number of calories recorded for exercise
• fat : the daily number of calories that turned into fat (+ve means fat gained, -ve means fat lost)
• muscle : the daily number of calories that turned into muscle (+ve means muscle gained, -ve means muscle lost)

Note that calories per day seems low. This is due to the inherent errors in the data provided.

• The exercise calories only count for number of steps, walking up stairs, etc. Sometimes these were part of actual exercise, but sometimes they are just part of NEAT (Non-Exercise Activity Thermogenesis). The average of all of this is a compromise. So consuming that amount of calories per day should be fine if your exercise/non-exercise routines stays consistent.
• The fat and muscle are gathered using a scale but those values at best are estimates. Those errors tend to be averaged out over many days worth of data.
• Using weight versus fat/muscle is more accurate, but it contains weight of water and poop which can also cause errors.
• errors in recording food intake is common. Either skipping food or under- over-estimating the caloric value of the food is very common. The key is to be consistent. So even if the recorded value is not accurate, then just eat consistently to the maintenance level.