Inter-Event Interval Estimator¶
Interval estimator — instantiates Event-Rate Magnitude Encoding
Reads rate from the time between consecutive events, so a single short gap already signals a high rate — the fastest, twitchiest estimate.
Inter-Event Interval Estimator works in the time-between domain rather than the count domain. Instead of tallying how many events fall in a window, it measures the gap between two successive events and inverts it: a rate is one event per interval, so the estimate is simply the reciprocal of the latest gap. Its defining trait is that it needs only two events to produce an estimate and updates on every new event, giving the lowest-latency rate any mechanism here can offer — but paying for that speed with the highest variance, because each estimate rests on a single noisy measurement. Because it lives in the interval domain, a run of unusually short gaps is not just a high number but a directly readable signature of a burst, which is what lets this mechanism flag bursts that a windowed count would average away.
Example¶
An ECG monitor at a bedside displays an instantaneous heart rate, and it does not count beats in a minute — it measures the R-R interval, the time between two consecutive R-waves, and reports 60 divided by that gap in seconds. A gap of 0.75 s reads as 80 bpm; the very next beat arriving 0.60 s later reads as 100 bpm. The number can change beat to beat, which is exactly what a clinician watching for an arrhythmia wants: they see the rhythm's fine texture, not a smoothed one-minute average that would hide it.
That sensitivity is also where the mechanism earns its burst-detection role. When a run of premature beats arrives — several R-R intervals suddenly collapsing to a fraction of a second — the estimator does not merely nudge the average up; it registers a cluster of very short intervals in a row, the direct fingerprint of a paroxysmal burst. The same run, poured into a one-minute count, would lift the minute's rate by a handful of beats and vanish. The cost of all this immediacy shows on the other side: a single dropped or spuriously detected R-wave doubles or halves one interval, throwing a wild instantaneous value that a count would have quietly absorbed.
How it works¶
- Time the gaps, don't count the events. On each new event, take the interval since the previous one; that single measurement is the estimator's raw input.
- Invert to a rate. The instantaneous rate is one event per interval — the reciprocal of the latest gap, scaled to the reporting units.
- Optionally pool a few intervals. Averaging the last handful of gaps (or their reciprocals) trades some of the beat-to-beat responsiveness for lower variance — a small, local smoothing that stays in the interval domain.
- Flag short-interval runs. A consecutive stretch of gaps below a threshold is annotated as a burst, surfacing clustering that a count over a window would flatten into a mild average.
Tuning parameters¶
- Intervals pooled — one gap gives maximum responsiveness and maximum noise; averaging the last k gaps steadies the estimate but reintroduces lag proportional to k.
- Reciprocal-vs-interval reporting — whether to average the intervals then invert, or invert then average; the two differ systematically because the reciprocal is nonlinear, and the choice biases the result.
- Burst threshold — the gap length below which consecutive intervals count as a burst, and how many in a row are required. Tighter thresholds flag more, at the cost of false burst calls on ordinary jitter.
- Artifact rejection — bounds on a plausible interval, so a single missed or double-counted event is discarded rather than inverted into an absurd rate.
When it helps, and when it misleads¶
Its strength is immediacy and burst sensitivity: it reacts on the very next event, needs almost no history or memory, and makes clustering directly legible instead of averaging it out. Where the timing texture of events matters — arrhythmia detection, machine chatter, packet micro-bursts — it sees structure the counting siblings smooth over.
Its failure mode is fragility. Because the estimate is the reciprocal of a single noisy interval, one missing or spurious event corrupts it completely, and — more insidiously — averaging reciprocals systematically overstates the rate[1], since the mean of one-over-the-intervals is not one-over-the-mean-interval. The classic misuse is treating the twitchy instantaneous value as a stable magnitude and reacting to every jitter, or averaging inverted intervals without correcting the bias and reading a phantom elevation. The guarding discipline is to reject implausible intervals before inverting, to pool a small number of gaps when a steadier read is needed, and to be explicit about whether you averaged intervals or their reciprocals.
How it implements the components¶
rate_decoder— the decode is a reciprocal: rate equals one event divided by the measured inter-event interval, updated on every event rather than at a window boundary.burst_annotation— a run of consecutive short intervals is flagged as a burst, turning event clustering into an explicit annotation instead of an averaged-away bump.
It does not implement observation_window — that is Fixed-Window Event Count, its nearest twin, from which it differs by needing no window at all: it reads rate from the gap between just two consecutive events rather than counting events over a fixed interval. It also leaves latency_precision_tradeoff_rule to windowed siblings, since it takes maximum responsiveness by construction rather than tuning a window to balance the two.
Related¶
- Instantiates: Event-Rate Magnitude Encoding — Inter-Event Interval Estimator gives the archetype its fastest, interval-domain decoder and its native burst detector.
- Sibling mechanisms: Fixed-Window Event Count · Rolling-Window Rate Estimator · Adaptive Window Widening · Exponential Leaky Integrator · Poisson Rate Model · Spike-Rate Readout · Pulse-Density Modulation · Rate Saturation Clamp · Anti-Aliasing Bin Selection
Editorial Notes¶
Form Classification¶
Form family: Analysis, Modeling & Optimization
Rationale: Inter-Event Interval Estimator operates as a computation, comparison, model, or analytic representation used to infer, estimate, or choose because it reads rate from the time between consecutive events, so a single short gap already signals a high rate — the fastest, twitchiest estimate
Independent corroboration: The frozen evidence defines Inter-Event Interval Estimator as 'Reads rate from the time between consecutive events, so a single short gap already signals a high rate — the fastest, twitchiest estimate', so its operative form is Analysis, Modeling & Optimization.
Review outcome: Independent reviewer agreement; high confidence.
Origin Attribution¶
Primary origin: Statistics & Experimental Design
Origin pattern: Convergent development
Present-day reach: Multi-domain
Rationale: Estimating event rate from waiting times is a standard statistical point-process and renewal-process construction.
Related originating lineages:
- Engineering & Design — Signal-processing and instrumentation practice contribute low-latency interval measurement and artifact rejection.
- Mathematics — Reciprocal transformations and Jensen's inequality materially explain the estimator's bias.
- Medicine & Healthcare — Instantaneous heart-rate estimation from R-R intervals is a mature clinical implementation.
- Neuroscience — Inter-spike interval coding materially provides a characteristic fast-rate estimator in biological sensing.
Review resolution: Both independent reviews place the primary lineage in statistics_experimental_design. The queued differences (alternate_origin_disagreement) concern secondary metadata rather than primary provenance. The final retains mathematics, neuroscience, engineering_design, medicine_healthcare only where a reviewer supplied a formative-lineage rationale; this does not convert downstream applicability into origin. origin_mode=convergent because the reviewers document independently established or materially co-developing traditions. domain_reach=multi_domain records application breadth separately from provenance.
Review outcome: Reconciled after independent review; high confidence.
References¶
[1] Jensen, J. L. W. V. "Sur les fonctions convexes et les inégalités entre les valeurs moyennes". Acta Mathematica 30, 175–193 (1906). For positive varying intervals, Jensen’s inequality establishes that the mean reciprocal exceeds the reciprocal of the mean. registry ↩