Hypothesis

AI‑enhanced CGM metrics of postprandial glycemic variability predict the magnitude of substrate‑switching (ΔRER) measured by indirect calorimetry during a mixed‑meal tolerance test.

Rationale

• Continuous glucose monitors now deliver high‑frequency (≥5 min) glucose traces that can be processed by machine‑learning algorithms to extract variability indices such as mean amplitude of glycemic excursions (MAGE), CONGA, and glucose rate‑of‑change (ROC) {1}.
• Postprandial glucose spikes trigger insulin surges that acutely inhibit lipolysis via AKT‑mediated phosphorylation of hormone‑sensitive lipase (HSL) and increase malonyl‑CoA through activation of acetyl‑CoA carboxylase (ACC), thereby suppressing carnitine palmitoyl‑transferase‑1 (CPT‑1)–mediated fatty‑acid oxidation {2}.
• Repeated oscillations in glucose and insulin cause intermittent re‑activation of fatty‑acid oxidation, preventing a smooth transition to fat burning and reducing the flexibility of the metabolic system.
• Consequently, individuals with higher postprandial glycemic variability should exhibit a smaller increase in carbohydrate oxidation (lower ΔRER) after a mixed meal, reflecting impaired metabolic flexibility.

Testable Prediction

In a cohort of n ≥ 60 healthy adults, the postprandial MAGE (calculated from CGM data collected during a 4‑hour mixed‑meal tolerance test containing 75 g carbohydrates, 30 g fat, 30 g protein) will be negatively correlated with ΔRER (RER_post‑meal − RER_fast) measured by indirect calorimetry (Pearson r < ‑0.3, p < 0.01).

Experimental Design

1. Participants: Recruit 60 adults (BMI 18.5‑30 kg/m²), stratify by sex and age.
2. Baseline: Measure fasting RER via 20‑min indirect calorimetry; attach Dexcom G7 CGM (15.5‑day wear) {2}.
3. Mixed‑Meal Tolerance Test (MMTT): Provide a standardized liquid meal (75 g CHO, 30 g FAT, 30 g PRO). Record CGM at 1‑min intervals for 240 min.
4. CGM‑Derived Metrics: Compute MAGE, CONGA1, and mean absolute glucose rate‑of‑change (mg/dL/min) using open‑source algorithms.
5. Indirect Calorimetry: Measure VO₂ and VCO₂ before meal (fasting) and at 90, 150, 210 min post‑meal; calculate RER and ΔRER as the maximal post‑meal RER minus fasting RER.
6. Statistical Analysis: Test correlation between each CGM variability metric and ΔRER; adjust for age, sex, BMI, and fasting insulin.

Falsifiability

If the correlation coefficients are not significantly different from zero (|r| < 0.1, p > 0.05) for all variability metrics, the hypothesis is falsified. Conversely, a robust negative correlation supports the notion that CGM‑derived glycemic variability serves as a proxy for the dynamic insulin‑malonyl‑CoA axis that gates fatty‑acid oxidation, thereby offering a low‑cost, wearable surrogate for metabolic flexibility.

Broader Impact

Validating this link would enable athletes, biohackers, and clinicians to assess metabolic flexibility using only a CGM device, closing the gap between glucose tracking and substrate oxidation measurement highlighted in recent reviews {3} and informing personalized nutrition or exercise prescriptions without the need for costly indirect calorimetry or breath analyzers.

Key takeaway: Postprandial glucose ‘noise’ is not merely measurement error; it reflects the physiological tug‑of‑war between insulin‑mediated lipolysis inhibition and the rebound of fat oxidation, shaping the body’s capacity to switch fuels.