Heart rate monitoring via mobile app

Heart Rate Monitoring Implementation via Mobile Application

Heart rate measurable three ways via phone: built-in camera (PPG via flash), Bluetooth GATT with external sensor (chest strap, smartwatch), or reading from platform storage (HealthKit, Health Connect — data from Apple Watch / Wear OS). Approach choice determines accuracy, latency and UX.

PPG via Camera: How it Works and Why Difficult

User places finger on camera with flash. Capillaries pulse — changes reflected red light. Analyze green channel (more hemoglobin sensitive) of each frame.

// iOS: AVCaptureVideoDataOutputSampleBufferDelegate
func captureOutput(_ output: AVCaptureOutput,
                   didOutput sampleBuffer: CMSampleBuffer,
                   from connection: AVCaptureConnection) {
    guard let pixelBuffer = CMSampleBufferGetImageBuffer(sampleBuffer) else { return }

    CVPixelBufferLockBaseAddress(pixelBuffer, .readOnly)
    defer { CVPixelBufferUnlockBaseAddress(pixelBuffer, .readOnly) }

    let width = CVPixelBufferGetWidth(pixelBuffer)
    let height = CVPixelBufferGetHeight(pixelBuffer)
    let bytesPerRow = CVPixelBufferGetBytesPerRow(pixelBuffer)
    guard let buffer = CVPixelBufferGetBaseAddress(pixelBuffer) else { return }

    // Average green channel over center area
    var greenSum: Int64 = 0
    let startX = width / 3
    let startY = height / 3
    for y in startY..<(height * 2 / 3) {
        for x in startX..<(width * 2 / 3) {
            let pixel = buffer.advanced(by: y * bytesPerRow + x * 4)
            greenSum += Int64(pixel.load(fromByteOffset: 1, as: UInt8.self))
        }
    }
    let avgGreen = Double(greenSum) / Double((width / 3) * (height / 3))
    processPPGSample(avgGreen)
}

Then — FFT or Peak Detection on time series. At 30 fps, HR range 40–200 bpm → frequency 0.67–3.33 Hz. FFT on 5-second window (150 samples) gives sufficient resolution.

PPG problems:

• Finger movement = artifacts 10× larger than pulse signal
• Ambient light brightness affects baseline (needs normalization)
• Accuracy ±5–10 bpm under ideal conditions, ±20+ on movement
• iOS limits flash brightness — weak signal on some models

PPG suitable for "measure at rest, hold finger 30 seconds". For real-time monitoring during movement — need external sensor.

Bluetooth GATT: Heart Rate Profile

Standard BLE profile for pulse sensors (Polar H10, Wahoo TICKR, Garmin HRM):

• Service UUID: 0x180D (Heart Rate)
• Characteristic UUID: 0x2A37 (Heart Rate Measurement)

Data arrives via Notification. First byte — flags: bit 0 determines format (UINT8 or UINT16), bit 4 — RR-intervals presence.

override fun onCharacteristicChanged(
    gatt: BluetoothGatt,
    characteristic: BluetoothGattCharacteristic,
    value: ByteArray
) {
    val flag = value[0].toInt()
    val isUint16 = flag and 0x01 != 0
    val heartRate = if (isUint16) {
        ((value[2].toInt() and 0xFF) shl 8) or (value[1].toInt() and 0xFF)
    } else {
        value[1].toInt() and 0xFF
    }

    // RR-intervals (if present) — for HRV
    if (flag and 0x10 != 0) {
        var offset = if (flag and 0x08 != 0) 4 else 3 // skip energy if present
        while (offset + 1 < value.size) {
            val rrRaw = ((value[offset + 1].toInt() and 0xFF) shl 8) or (value[offset].toInt() and 0xFF)
            val rrMs = rrRaw * 1000 / 1024 // convert from 1/1024 s to milliseconds
            rrIntervals.add(rrMs)
            offset += 2
        }
    }
}

RR-intervals — basis for HRV (Heart Rate Variability) calculation. If product claims stress monitoring or recovery — can't do without RR.

Reading from HealthKit / Health Connect

For apps not doing direct measurement, only displaying:

iOS:

let query = HKAnchoredObjectQuery(
    type: HKQuantityType(.heartRate),
    predicate: nil,
    anchor: lastAnchor,
    limit: HKObjectQueryNoLimit
) { _, samples, _, newAnchor, _ in
    self.lastAnchor = newAnchor
    let bpmValues = (samples as? [HKQuantitySample])?.map {
        $0.quantity.doubleValue(for: HKUnit(from: "count/min"))
    } ?? []
}
healthStore.execute(query)

Data arrives with sources: Apple Watch Series 4+ gives heart rate every 5–15 minutes at rest, every second during workout.

Visualization

For realtime HR graph: ring buffer of last N values, update once per second. On iOS — Charts (DanielGindi) or Swift Charts (iOS 16+). On Android — MPAndroidChart or Compose Canvas with custom drawing.

Heart rate zones calculated from max heart rate (220 minus age or Karvonen formula with resting heart rate):

Timeframes

PPG measurement via camera — 2–3 weeks (with algorithm part). Bluetooth GATT integration — 1–2 weeks. Reading from HealthKit/Health Connect with visualization — 5–8 days.