Heart Rate Alarms

Heart Rate alarms use measured heart rate values to detect abnormal physiology patterns. The app supports three heart-rate checks that can run together.

How it works

flowchart TD
    A[Incoming heart rate value] --> B[Simple threshold check]
    A --> C[Adaptive threshold check around short rolling average]
    A --> D[Long rolling average threshold check]
    B --> E{Any HR check in ALARM?}
    C --> E
    D --> E
    E -->|Yes| F[Heart Rate algorithm ALARM]
    E -->|No| G[Heart Rate algorithm OK]

The Heart Rate detector is made of three separate checks that can run in parallel:

• Simple threshold: alarms if current heart rate is below minimum or above maximum.
• Adaptive threshold: compares current heart rate against a short rolling average plus or minus an offset.
• Rolling average threshold: alarms if the longer rolling average is outside configured limits.

Technical description of each algorithm

1. Simple threshold algorithm

This is the direct high/low bpm check.

• Input: current heart rate sample.
• Alarm condition: current HR < HRThreshMin or current HR > HRThreshMax.
• Best use: clear hard limits, for example very high tachycardia or very low bradycardia.

flowchart LR
    A[Current HR sample] --> B{HR < HRThreshMin?}
    B -->|Yes| C[Simple HR ALARM]
    B -->|No| D{HR > HRThreshMax?}
    D -->|Yes| C
    D -->|No| E[Simple HR OK]

2. Adaptive threshold algorithm

This tracks a short rolling average and creates a moving alarm band around it.

• Input: current HR and short-window average.
• Dynamic limits:
  • lower limit = adaptive_average - HRAdaptiveAlarmThresh
  • upper limit = adaptive_average + HRAdaptiveAlarmThresh
• Alarm condition: current HR outside the dynamic limits.
• Best use: detect rapid changes relative to recent baseline, not just fixed absolute values.

3. Average HR algorithm

This monitors a longer rolling average and compares that average against fixed limits.

• Input: long-window rolling average.
• Alarm condition: average HR < HRAverageAlarmThreshMin or average HR > HRAverageAlarmThreshMax.
• Best use: sustained physiological drift rather than brief spikes.

Example: rising HR during a seizure-like event

The table below shows how the adaptive threshold band moves upward as HR rises, while still detecting a rapid jump.

Assumed settings in this example:

• HRAdaptiveAlarmThresh = 20 bpm
• HRThreshMax = 140 bpm
• HRAverageAlarmThreshMax = 115 bpm

Time point	Current HR	Adaptive average	Adaptive band (avg +/- 20)	Simple threshold	Adaptive threshold	Average HR
t0 (rest)	80	80	60 to 100	OK	OK	OK
t1	90	82	62 to 102	OK	OK	OK
t2	104	86	66 to 106	OK	OK	OK
t3 (rapid rise)	124	92	72 to 112	OK	ALARM	OK
t4	132	100	80 to 120	OK	ALARM	WARNING trend
t5 (sustained high)	128	108	88 to 128	OK	borderline/OK	ALARM

flowchart LR
    T0[t0: HR 80, avg 80, band 60-100, all OK] --> T1[t1: HR 90, avg 82, band 62-102, all OK]
    T1 --> T2[t2: HR 104, avg 86, band 66-106, all OK]
    T2 --> T3[t3: HR 124, avg 92, band 72-112, adaptive ALARM]
    T3 --> T4[t4: HR 132, avg 100, band 80-120, adaptive ALARM]
    T4 --> T5[t5: HR 128, avg 108, band 88-128, average-HR ALARM risk]

Interpretation:

• Simple threshold reacts only when fixed min or max limits are crossed.
• Adaptive threshold reacts early to fast changes, then its band follows the trend.
• Average HR reacts later, but is strong for prolonged elevation.

User settings

Setting	What it changes
HRAlarmActive	Enables or disables Heart Rate alarms.
HRThreshMin	Lower bpm threshold for the simple heart-rate check.
HRThreshMax	Upper bpm threshold for the simple heart-rate check.
HRAdaptiveAlarmActive	Enables adaptive threshold mode.
HRAdaptiveAlarmWindowSecs	Rolling-average window length used by adaptive mode.
HRAdaptiveAlarmThresh	Offset above and below the adaptive rolling average that triggers alarm.
HRAverageAlarmActive	Enables long rolling-average threshold mode.
HRAverageAlarmWindowSecs	Window length for long rolling-average mode.
HRAverageAlarmThreshMin	Lower limit for the long rolling average.
HRAverageAlarmThreshMax	Upper limit for the long rolling average.
HRNullAsAlarm	If enabled, missing or invalid heart-rate readings are treated as alarm.
HrFrozenAlarm	Raises a fault warning if heart-rate values appear frozen for too long.

Practical tuning effect

flowchart LR
    A[Lower HRThreshMin] --> B[Fewer low-heart-rate alarms]
    C[Raise HRThreshMax] --> D[Fewer high-heart-rate alarms]
    E[Shorter adaptive window] --> F[More responsive to rapid changes]
    G[Longer adaptive window] --> H[More stable, less reactive]
    I[Tighter adaptive offset] --> J[More sensitive]
    K[Wider adaptive offset] --> L[Less sensitive]

Important data-source note

Heart-rate detection quality depends on watch sensor quality and data continuity. In practical use, Garmin devices generally provide the most reliable continuous heart-rate data.