Continuous Glucose
& Glycemic Variability

The mean of all valid sensor readings — the level around which everything else varies, and the input to GMI.

Absolute glucose variability. Read alongside CV (consensus prefers CV because SD scales with the mean).

Average absolute rate of change per hour — how briskly glucose moves between samples.

Share of the reporting period with valid sensor data. Consensus guidance wants ≥70% over 14 days for the metrics to be representative.

Total days of active sensor wear in the analysis. Some indices (ADRR, MODD) require at least two days; the consensus AGP window is 14 days.

Minutes of suspected nocturnal compression (lying on the sensor), held out of the time-below-range accounting so a false low does not inflate hypoglycemia. A genuine sharp insulin hypo (sustained sub-70, gradual descent + Somogyi rebound) is now disambiguated from a positional artifact and survives into the nocturnal-hypo count; only a near-vertical, single-cell drop-and-recover is flagged here.

A direct data-quality confidence multiplier reflecting coverage, gaps and flagged artifact.

Count of detected nocturnal sub-70 mg/dL episodes — the most clinically consequential lows, and an input to the hypo↔QTc fusion signal.

Number of slope-detected glucose excursions exceeding one SD — the raw count feeding MAGE.

The primary CGM target: percentage of time glucose sits in 70–180 mg/dL. Each 5-point gain in TIR is clinically meaningful; the consensus goal for most adults is >70%.

An estimated HbA1c-equivalent computed from mean glucose. A modelled estimate that can differ from a lab HbA1c by ±0.3% or more — see GMI vs Lab.

The ADAG study’s mean-glucose-to-A1c regression — an alternative estimate to GMI using a different coefficient set.

Glucose variability normalised to the mean — the consensus stability metric. Above 36%, hypoglycemia risk rises sharply.

Total time below 70 mg/dL — overall hypoglycemia exposure (TBR level 1 + 2 combined).

Total time above 180 mg/dL — overall hyperglycemia exposure (TAR level 1 + 2 combined).

Time in the tighter 70–140 mg/dL band — a more demanding target introduced in the 2023 trial-metrics consensus, relevant for near-normal glycemia goals.

Time in 54–69 mg/dL — level-1 hypoglycemia.

Time below 54 mg/dL — clinically significant level-2 hypoglycemia.

Time in 181–250 mg/dL — level-1 hyperglycemia.

Time above 250 mg/dL — level-2 hyperglycemia.

Average size of glucose swings exceeding one SD — the classic variability index, sensitive to large meal/treatment excursions.

Average absolute difference between glucose at the same clock time on consecutive days — day-to-day reproducibility. Needs ≥2 days.

SD of the differences between readings a fixed interval apart (commonly 1, 2 or 4 h) — intra-day variability at a chosen timescale.

A single index combining mean level and variability into one number.

A risk score weighting each reading by its log-scaled deviation from euglycemia — penalises both highs and lows.

A symmetric daily risk index capturing both hypo- and hyperglycemic excursions equally. Needs ≥2 days of data.

A risk-weighted measure of hypoglycemia exposure — emphasises how far and how often glucose drops low.

A risk-weighted measure of hyperglycemia exposure — the upper-range counterpart of LBGI.

Variability as the extra path-length the glucose trace travels versus a flat line — a geometric variability measure that captures both amplitude and frequency.

The pre-breakfast glucose rise from the 03–06h nadir — a circadian counter-regulatory pattern that can drive morning highs.

The rate-corrected QT interval imported from ECGDex when available — paired with nocturnal hypoglycemia to flag the hypo↔QTc pattern below.

A 0–100 stability headline blending time-in-range, CV and hypoglycemia exposure.

A directional insulin-resistance risk band from glycemic level/variability and autonomic inputs.

A composite autonomic-risk signal combining ECGDex overnight slope, surges and coupling with glycemic context.

A composite variability index blending CV, MAGE and the dawn rise — a single fusion input for the Integrator.

A directional composite flagging the coincidence of nocturnal hypoglycemia and prolonged QTc — the “dead-in-bed” risk pattern. A screening signal, not a diagnosis.

The gap between GMI and a lab HbA1c when one is entered. GMI reflects recent mean glucose; lab A1c reflects red-cell glycation over months — a delta is expected, not an error.

Sensor mean glucose versus the estimated average glucose implied by a lab A1c — a sensor-calibration sanity check.