Sub-System Trend Sensitivity Index (STSI)

GMI-AVI-028AviationIntegration-Required

Definition

Rate at which emerging reliability trends at the sub-system or component level are detectable at the standard ATA chapter level of analysis: measuring whether the reliability program's standard aggregation granularity can see developing problems before they escalate.

The Measurement Gap

The measurement gap STSI addresses: reliability programs built on ATA chapter-level trending were designed before AI-enabled analytics made sub-system-level trending computationally practical. Standard reliability reports show pilot report rates, removal rates, and delay rates by ATA chapter, which is the right level for communicating to flight operations and for tracking against established alert thresholds. But it's not the right level for early detection of developing problems, which often start in a single sub-system or even a single component position before propagating to a pattern visible at the chapter level.

An AI-powered reliability analysis system that flags trends operates on the data it's given. If the system is fed ATA chapter-level aggregated data, it will detect chapter-level trends, but it may miss sub-system trends that represent the early warning signal of what will eventually become a chapter-level problem. The system isn't failing; the aggregation is limiting its sensitivity.

The decision STSI supports: should the reliability program add sub-system-level trending for specific chapters where STSI is low, or does the current ATA chapter-level analysis provide sufficient detection sensitivity? STSI answers this empirically by measuring how often significant sub-system trends are producing visible chapter-level signals. If STSI is below 0.50 for ATA 32 (landing gear), the reliability team is missing half of the significant landing gear sub-system trends before they become chapter-level alerts. That's an information gap with safety consequences: one the reliability program doesn't know exists without STSI.

Formula

Plain language: For each ATA chapter, compute the reliability trend at the chapter level and at each sub-system level within the chapter. Identify sub-system trends that are statistically significant (rate increasing at a level that would trigger a reliability alert) but invisible at the chapter level. The ratio of detectable sub-system trends to total significant sub-system trends is the sensitivity index.

Notation:
STSI = Sub_system_trends_detectable_at_chapter / Total_significant_sub_system_trends

Variables:

  • Total_significant_sub_system_trends = count of sub-system-level reliability trends that are statistically significant (e.g., rate increasing at > 2 standard deviations from baseline over the measurement period) (count)
  • Sub_system_trends_detectable_at_chapter = count of those significant sub-system trends that also produce a statistically significant signal at the ATA chapter level (count)

Units: 0 to 1. 1.0 = every significant sub-system trend is visible at the chapter level: chapter-level trending is sufficient. Lower = sub-system trends are masked by chapter-level aggregation.

Trend detection criteria: define "statistically significant" consistently for both sub-system and chapter level. Use the same statistical test (e.g., Mann-Whitney or z-test for rate change), the same significance threshold (e.g., p < 0.05), and the same lookback window. Inconsistent criteria will produce misleading STSI values.

Focus chapters: STSI is most valuable for ATA chapters with many sub-systems where masking is most likely: ATA 32 (Landing Gear), ATA 34 (Navigation), ATA 27 (Flight Controls), ATA 73–80 (Engine sections when tracked by sub-system). Chapters with few sub-systems (ATA 49: APU) will show naturally high STSI because the chapter and sub-system are nearly the same granularity.

Proposed formula, pending validation pilots.

At a Glance

Category
Maintenance Program Integrity
← Reliability Alert Operational Value Index (RAOVI)Task Deferral Compounding Index (TDCI) →

Assess Your Organization

The AI-GPS Assessment evaluates your AI governance maturity and recommends KPIs like this one for your implementation roadmap.

Take the Free AI-GPS Assessment