Syndromic Surveillance of User Complaints

Description

Continuously monitors grouped 'syndromes' from free-text complaints and signals, using statistical aberration detection to catch emerging problems early.

Public health agencies monitor emergency department chief complaints, pharmacy sales and absenteeism, grouped into syndromes (e.g. respiratory, gastrointestinal), and flag unusual increases before diagnoses are confirmed. Aberration detection algorithms (Farrington, EARS C1-C3, CUSUM) compare current counts with expected baselines accounting for seasonality and day-of-week effects, controlling false alarms. For products, user complaints, support tickets, reviews and in-app feedback can be classified into syndromes (e.g. 'login trouble', 'billing confusion') with NLP, and monitored the same way. It provides early warning of emergent UX problems with explicit sensitivity and false-alarm trade-offs.

Attributes

Loss type
Unmigrated
Stage
Post-launch
Data
Mixed
Effort
Weeks

Steps

  1. Define syndromes as groups of related complaints and build a classifier (rules or ML) with validated precision.
  2. Assemble daily or hourly counts per syndrome, with denominators (active users).
  3. Choose an aberration detection algorithm suited to the data (Farrington for seasonal counts, EARS for short baselines).
  4. Calibrate thresholds to an acceptable false alarm rate using historical data.
  5. Set up alert triage: who investigates, with what playbook (e.g. outbreak investigation).
  6. Evaluate system sensitivity and timeliness against past incidents and refine syndromes.

Sources

  • Henning, What is syndromic surveillance? (2004), MMWR Supplement
  • Farrington et al., A statistical algorithm for the early detection of outbreaks of infectious disease (1996), JRSS A
  • Hutwagner et al., The bioterrorism preparedness and response Early Aberration Reporting System (EARS) (2003), Journal of Urban Health

Also known as

  • Syndromic surveillance
  • Aberration detection
  • Farrington algorithm
  • EARS

Categories

  • Epidemiology & Population Health