Alert Fatigue Detector¶

Analyze safety alert streams to detect operator fatigue — when too many alerts lead to desensitization and missed critical events.

Fatigue Indicators¶

Indicator	What it measures
Volume Overload	Alerts/hour exceeding sustainable threshold
Repetition Noise	Same alert type firing repeatedly
Severity Inflation	Too many critical/high alerts diluting urgency
Ack Lag Growth	Acknowledgment time trending upward
Suppression Rate	Percentage of alerts auto-suppressed
Off-Hours Load	Alerts during nights/weekends (burnout risk)

Fatigue Levels¶

Healthy (0-19): Alert hygiene is good
Mild (20-39): Early signs of noise
Moderate (40-59): Action recommended
Severe (60-79): Operator burnout likely
Critical (80-100): Immediate intervention needed

CLI Usage¶

# Simulate and analyze 72 hours of alerts
python -m replication fatigue-detect simulate --hours 72 --rate 40

# Analyze alerts from file
python -m replication fatigue-detect analyze --alerts alerts.json

# Get recommendations
python -m replication fatigue-detect recommend --alerts alerts.json

# Generate HTML report
python -m replication fatigue-detect report --output fatigue.html

Programmatic API¶

from replication.fatigue_detector import FatigueDetector, FatigueThresholds, AlertEvent

detector = FatigueDetector(thresholds=FatigueThresholds(max_alerts_per_hour=30))
detector.ingest(alert_events)
result = detector.analyze()

print(f"Score: {result.score}/100 — {result.level}")
for ind in result.indicators:
    print(f"  {ind.name}: {ind.score} — {ind.detail}")
for rec in result.recommendations:
    print(f"  → {rec}")

Simulation¶

The built-in simulator generates realistic alert streams with fatigue onset:

from replication.fatigue_detector import simulate_alerts

events = simulate_alerts(hours=72, base_rate=20, fatigue_onset_hour=24, seed=42)
# After fatigue_onset_hour: ack delays grow, suppression increases

HTML Reports¶

Generate self-contained HTML reports with score cards, indicator bars, recommendations, and statistics.