Read raw data

The following example shows you how to read raw data as part of the common workflow.

Read data

Health Connect allows apps to read data from the datastore when the app is in the foreground and background:

  • Foreground reads: You can normally read data from Health Connect when your app is in the foreground. In these cases, you may consider using a foreground service to run this operation in case the user or system places your app in the background during a read operation.

  • Background reads: By requesting an extra permission from the user, you can read data after the user or system places your app in the background. See the complete background read example.

The Steps data type in Health Connect captures the number of steps a user has taken between readings. Step counts represent a common measurement across health, fitness, and wellness platforms. Health Connect lets you read and write step count data.

To read records, create a ReadRecordsRequest and supply it when you call readRecords.

The following example shows how to read step count data for a user within a certain time. For an extended example with SensorManager, see the step count data guide.

val response = healthConnectClient.readRecords(
    ReadRecordsRequest(
        HeartRateRecord::class,
        timeRangeFilter = TimeRangeFilter.between(startTime, endTime)
    )
)
response.records.forEach { record ->
    /* Process records */
}

You can also read your data in an aggregated manner using aggregate.

suspend fun readStepsAggregate(startTime: Instant, endTime: Instant): Long {
    val response = healthConnectClient.aggregate(
        AggregateRequest(
            metrics = setOf(StepsRecord.COUNT_TOTAL),
            timeRangeFilter = TimeRangeFilter.between(startTime, endTime)
        )
    )
    return response[StepsRecord.COUNT_TOTAL] ?: 0L
}

Read mobile steps

With Android 14 (API level 34) and SDK Extension version 20 or higher, Health Connect provides on-device step counting. If any app has been granted the READ_STEPS permission, Health Connect starts capturing steps from the Android-powered device, and users see steps data automatically added to Health Connect Steps entries.

To check if on-device step counting is available, verify that the device is running Android 14 (API level 34) and has at least SDK extension version 20:

val isStepTrackingAvailable =
    Build.VERSION.SDK_INT >= Build.VERSION_CODES.UPSIDE_DOWN_CAKE &&
        SdkExtensions.getExtensionVersion(Build.VERSION_CODES.UPSIDE_DOWN_CAKE) >= 20

If your app reads aggregated step counts using aggregate and doesn't filter by DataOrigin, on-device steps are automatically included in the total, and no changes are required for the June 2026 update.

Attribution change for on-device steps

Starting with the June 2026 update, steps tracked natively by Health Connect are attributed to a Synthetic Package Name (SPN), such as com.android.healthconnect.phone.jd5bdd37e1a8d3667a05d0abebfc4a89e.

Previously, built-in steps were attributed to the package name android. Historical step data recorded before June 2026 retains the android package name.

SPNs are device-specific and scoped on a per-application basis to protect user privacy:

  • Stable: The SPN for the current device is stable for your application.
  • Application-Scoped: Different applications on the same device see different SPNs for on-device step data.

Query for on-device steps

Because SPNs are scoped and device-specific, you must not hardcode SPN values. Instead, use the getCurrentDeviceDataSource() API to retrieve the SPN for the current device.

While on-device step counting requires SDK extension version 20 or higher, the getCurrentDeviceDataSource() API is available on Android 14 (API level 34) with SDK extension version 11 or higher.

The getCurrentDeviceDataSource() API is not yet available in the Health Connect Jetpack library. The following examples use the Android framework API instead:

import android.content.Context
import android.health.connect.HealthConnectManager

val healthConnectManager = context.getSystemService(HealthConnectManager::class.java)
val deviceDataSource = healthConnectManager?.getCurrentDeviceDataSource()
val currentDeviceSpn = deviceDataSource?.deviceDataOrigin?.packageName

If your app needs to read on-device steps, or if it displays step data broken down by source application or device, you must query for records where the DataOrigin is android or matches the device's SPN. If your app shows attribution for step data, use metadata.device to identify the source device for individual records. For on-device steps identified by an SPN in aggregated data, you can use device metadata such as model or manufacturer from DeviceDataSource for attribution, or use a generic label like "Your phone" for on-device steps.

The following example shows how to read aggregated on-device step count data by filtering for both android and the current device SPN:

import android.content.Context
import android.health.connect.HealthConnectManager
import android.os.Build
import android.os.ext.SdkExtensions
import androidx.health.connect.client.HealthConnectClient
import androidx.health.connect.client.records.StepsRecord
import androidx.health.connect.client.records.metadata.DataOrigin
import androidx.health.connect.client.request.AggregateRequest
import androidx.health.connect.client.time.TimeRangeFilter
import java.time.Instant

suspend fun readDeviceStepsByTimeRange(
    healthConnectClient: HealthConnectClient,
    context: Context,
    startTime: Instant,
    endTime: Instant
) {
    // 1. Check if SDK Extension 11+ is available for getCurrentDeviceDataSource()
    val isDataSourceApiAvailable = Build.VERSION.SDK_INT >= Build.VERSION_CODES.U &&
            SdkExtensions.getExtensionVersion(Build.VERSION_CODES.U) >= 11

    try {
        val healthConnectManager = context.getSystemService(HealthConnectManager::class.java)

        // 2. Safely fetch the package name only if API is available and data exists
        val currentDeviceSpn = if (isDataSourceApiAvailable) {
            healthConnectManager?.getCurrentDeviceDataSource()?.deviceDataOrigin?.packageName
        } else {
            null
        }

        val dataOriginFilters = mutableSetOf(DataOrigin("android"))

        // 3. Explicit null-safety check using .let
        currentDeviceSpn?.let {
            dataOriginFilters.add(DataOrigin(it))
        }

        val response = healthConnectClient.aggregate(
            AggregateRequest(
                metrics = setOf(StepsRecord.COUNT_TOTAL),
                timeRangeFilter = TimeRangeFilter.between(startTime, endTime),
                dataOriginFilter = dataOriginFilters
            )
        )

        val stepCount = response[StepsRecord.COUNT_TOTAL]

    } catch (e: Exception) {
        // Now this catch block only handles actual runtime exceptions, 
        // rather than Errors from missing methods.
    }
}

On-Device Step Counting

  • Sensor Usage: Health Connect utilizes the TYPE_STEP_COUNTER sensor from SensorManager. This sensor is optimized for low power consumption, making it ideal for continuous background step tracking.
  • Data Granularity: To conserve battery life, step data is typically batched and written to the Health Connect database no more frequently than once per minute.
  • Attribution: Steps recorded by this feature before June 2026 are attributed to the android package name in the DataOrigin. After this date, they are attributed to a device-specific SPN. See Attribution change for on-device steps.
  • Activation: The on-device step counting mechanism is active only when at least one application on the device has been granted the READ_STEPS permission within Health Connect.

Background read example

To read data in the background, declare the following permission in your manifest file:

<application>
  <uses-permission android:name="android.permission.health.READ_HEALTH_DATA_IN_BACKGROUND" />
...
</application>

The following example shows how to read step count data in the background for a user within a certain time by using WorkManager:

class ScheduleWorker(appContext: Context, workerParams: WorkerParameters) :
    CoroutineWorker(appContext, workerParams) {

    override suspend fun doWork(): Result {
        val healthConnectClient = HealthConnectClient.getOrCreate(applicationContext)
        // Perform background read logic here
        return Result.success()
    }
}
 
@OptIn(ExperimentalFeatureAvailabilityApi::class)
fun enqueueBackgroundReadWorker(context: Context, healthConnectClient: HealthConnectClient) {
    if (healthConnectClient
            .features
            .getFeatureStatus(
                HealthConnectFeatures.FEATURE_READ_HEALTH_DATA_IN_BACKGROUND
            ) == HealthConnectFeatures.FEATURE_STATUS_AVAILABLE
    ) {

        val periodicWorkRequest = PeriodicWorkRequestBuilder<ScheduleWorker>(1, TimeUnit.HOURS)
            .build()

        WorkManager.getInstance(context).enqueueUniquePeriodicWork(
            "read_health_connect",
            ExistingPeriodicWorkPolicy.KEEP,
            periodicWorkRequest
        )
    }
}

The ReadRecordsRequest parameter has a default pageSize value of 1000. If the number of records in a single readResponse exceeds the pageSize of the request, you need to iterate over all pages of the response to retrieve all records by using pageToken. However, be careful to avoid rate-limiting concerns.

pageToken read example

It is recommended to use pageToken for reading records to retrieve all available data from the requested time period.

The following example shows how to read all records until all page tokens have been exhausted:

val type = HeartRateRecord::class
val endTime = Instant.now()
val startTime = endTime.minus(Duration.ofDays(7))

try {
    var pageToken: String? = null
    do {
        val readResponse =
            healthConnectClient.readRecords(
                ReadRecordsRequest(
                    recordType = type,
                    timeRangeFilter = TimeRangeFilter.between(
                        startTime,
                        endTime
                    ),
                    pageToken = pageToken
                )
            )
        val records = readResponse.records
        // Do something with records
        pageToken = readResponse.pageToken
    } while (pageToken != null)
} catch (quotaError: IllegalStateException) {
    // Backoff
}
 For information about best practices when reading large datasets, refer to Plan to avoid rate limiting.

Read previously written data

If an app has written records to Health Connect before, it is possible for that app to read historical data. This is applicable for scenarios in which the app needs to resync with Health Connect after the user has reinstalled it.

Some read restrictions apply:

  • For Android 14 and higher

    • No historical limit on an app reading its own data.
    • 30-day limit on an app reading other data.

  • For Android 13 and lower

    • 30-day limit on app reading any data.

The restrictions can be removed by requesting a Read permission.

To read historical data, you need to indicate the package name as a DataOrigin object in the dataOriginFilter parameter of your ReadRecordsRequest.

The following example shows how to indicate a package name when reading heart rate records:

try {
    val response =  healthConnectClient.readRecords(
        ReadRecordsRequest(
            recordType = HeartRateRecord::class,
            timeRangeFilter = TimeRangeFilter.between(startTime, endTime),
            dataOriginFilter = setOf(DataOrigin("com.my.package.name"))
        )
    )
    for (record in response.records) {
        // Process each record
    }
} catch (e: Exception) {
    // Run error handling here
}

Read data older than 30 days

By default, all applications can read data from Health Connect for up to 30 days prior to when any permission was first granted.

If you need to extend read permissions beyond any of the default restrictions, request the PERMISSION_READ_HEALTH_DATA_HISTORY. Otherwise, without this permission, an attempt to read records older than 30 days results in an error.

Permissions history for a deleted app

If a user deletes your app, all permissions, including the history permission, are revoked. If the user reinstalls your app and grants permission again, the same default restrictions apply, and your app can read data from Health Connect for up to 30 days prior to that new date.

For example, suppose the user deletes your app on May 10, 2023 and then reinstalls the app on May 15, 2023, and grants read permissions. The earliest date your app can now read data from by default is April 15, 2023.

Handle exceptions

Health Connect throws standard exceptions for CRUD operations when an issue is encountered. Your app should catch and handle each of these exceptions as appropriate.

Each method on HealthConnectClient lists the exceptions that can be thrown. In general, your app should handle the following exceptions:

Table 1: Health Connect exceptions and recommended best practices
Exception Description Recommended best practice
IllegalStateException One of the following scenarios has occurred:

  • The Health Connect service isn't available.
  • The request isn't a valid construction. For example, an aggregate request in periodic buckets where an Instant object is used for the timeRangeFilter.

Handle possible issues with the inputs first before doing a request. Preferably, assign values to variables or use them as parameters within a custom function instead of using them directly in your requests so that you can apply error handling strategies.
IOException There are issues encountered upon reading and writing data from disk. To avoid this issue, here are some suggestions:

  • Back up any user input.
  • Be able to handle any issues that occur during bulk write operations. For example, make sure the process moves past the issue and carry out the remaining operations.
  • Apply retries and backoff strategies to handle request issues.

RemoteException Errors have occurred within, or in communicating with, the underlying service to which the SDK connects.

For example, your app is trying to delete a record with a given uid. However, the exception is thrown after the app finds out upon checking in the underlying service that the record doesn't exist. 		To avoid this issue, here are some suggestions:

  • Perform regular syncs between your app's datastore and Health Connect.
  • Apply retries and backoff strategies to handle request issues.

SecurityException There are issues encountered when the requests require permissions that aren't granted. To avoid this, make sure that you've declared use of Health Connect data types for your published app. Also, you must declare Health Connect permissions in the manifest file and in your activity.

Previous
Write data
Next
Read aggregated data