Atlas lets you explore your Apple Health data.
Install Atlas using pip:
pip install atlas-dbUpgrade Atlas using pip:
pip install atlas-db --upgradeUninstall Expanse (old name) using pipx:
pipx uninstall expanseNote
Here is a Quarto notebook with example code and SQL queries.
The notebook uses Vega-Altair and Clickhouse (chDB) to explore Apple Health time series data in a .parquet file. The .parquet file was generated by Atlas from an Apple Health export.xml file.
First we create the .parquet file from the export.xml file.
atlas parquet export.xml -o ah.parquetWe can explore the data in many ways.
It is just a table/dataframe/parquet file with 5 columns.
But here we'll use clickhouse local:
clickhouse localLet's take a look at the table.
DESCRIBE TABLE `ah.parquet`┌─name────┬─type────────────────────┬─default_type─┬─default_expression─┬─comment─┬─codec_expression─┬─ttl_expression─┐
│ type │ Nullable(String) │ │ │ │ │ │
│ start │ Nullable(DateTime64(6)) │ │ │ │ │ │
│ end │ Nullable(DateTime64(6)) │ │ │ │ │ │
│ created │ Nullable(DateTime64(6)) │ │ │ │ │ │
│ value │ Nullable(String) │ │ │ │ │ │
└─────────┴─────────────────────────┴──────────────┴────────────────────┴─────────┴──────────────────┴────────────────┘
What kind of "types" do we have and how many?
SELECT
type,
COUNT(*) AS count
FROM `ah.parquet`
GROUP BY type
ORDER BY count DESC┌─type───────────────────────────┬──count─┐
│ ActiveEnergyBurned │ 879902 │
│ HeartRate │ 451854 │
│ BasalEnergyBurned │ 289031 │
│ DistanceWalkingRunning │ 260500 │
│ StepCount │ 217384 │
│ PhysicalEffort │ 69747 │
│ AppleExerciseTime │ 61363 │
│ AppleStandTime │ 58309 │
│ EnvironmentalAudioExposure │ 44535 │
│ SleepAnalysis │ 36599 │
│ WalkingStepLength │ 28281 │
│ WalkingSpeed │ 28281 │
│ RespiratoryRate │ 27829 │
│ AppleStandHour │ 25877 │
│ FlightsClimbed │ 22690 │
│ WalkingDoubleSupportPercentage │ 21900 │
│ WalkingAsymmetryPercentage │ 13820 │
│ HeartRateVariabilitySDNN │ 11961 │
│ OxygenSaturation │ 4912 │
│ StairDescentSpeed │ 4718 │
│ StairAscentSpeed │ 4249 │
│ DistanceCycling │ 2890 │
│ TimeInDaylight │ 2403 │
│ HeadphoneAudioExposure │ 2323 │
│ RestingHeartRate │ 1399 │
│ WalkingHeartRateAverage │ 1176 │
│ DistanceSwimming │ 455 │
│ SwimmingStrokeCount │ 455 │
│ AppleSleepingWristTemperature │ 442 │
│ RunningSpeed │ 391 │
│ VO2Max │ 366 │
│ RunningPower │ 173 │
│ DietaryCaffeine │ 171 │
│ AppleWalkingSteadiness │ 138 │
│ SixMinuteWalkTestDistance │ 122 │
│ HeartRateRecoveryOneMinute │ 76 │
│ RunningVerticalOscillation │ 74 │
│ RunningGroundContactTime │ 67 │
│ RunningStrideLength │ 54 │
│ MindfulSession │ 34 │
│ HighHeartRateEvent │ 18 │
│ AudioExposureEvent │ 14 │
│ BodyMass │ 14 │
│ Height │ 5 │
│ Fatigue │ 1 │
│ HKDataTypeSleepDurationGoal │ 1 │
└────────────────────────────────┴────────┘
What's our total step count?
Note
The value column is type Nullable(String) so we have to cast toFloat64 to sum up the step values.
SELECT sum(toFloat64(value))
FROM `ah.parquet`
WHERE type = 'StepCount'┌─sum(toFloat64(value))─┐
│ 30295811 │
└───────────────────────┘
30.295.811 (30.29 million) steps. That's a lot of steps!
- open the Apple Health app on iOS
- tap on your profile picture (or initials) at the top right
- tap on Export All Health Data
- tap on Export
- wait a few seconds to a few minutes (~3min for 10 years of data)
- get the export.zip archive via Airdrop to a Mac (or save to Files)
Note
The export.xml file is in the export.zip archive.
You can expand the export.zip file by double-clicking on it.
This creates a directory named apple_health_export and in it is the export.xml file.
See: Apple Support on how to export Apple Health and Fitness in XML format
atlas parquet export.xml
- turn export.xml into a simple parquet file
