Findings:

• High interpersonal variability in post-meal glucose observed in an 800-person cohort
• Using biometric and microbiome data enables accurate glucose response prediction
• Prediction is accurate and superior to common practice in an independent cohort
• Short-term personalized dietary interventions successfully lower post-meal glucose