4. Case Study — Diagnosis

Multi-channel time-series matching decides correctly most often — and fastest

PipelineCorrectWrongNeverPrecisionCoverageMedian latency
Knowledge graph93180.750.4019 s
Time-series matching, single-channel21091.000.7019 s
Time-series matching, multi-channel25051.000.8314 s
Bayesian network140161.000.4721.5 s
Knowledge graph
9
3
18
1
Time-series, single
21
9
1
Time-series, multi
25
5
1
Bayesian
14
16
1
correctwrongnever confidentno detection

The gate produced no wrong decisions for three of four pipelines.

Errors remain only where the evaluation itself is ambiguous (overlapping symptoms).