Q1: Known Unknowns (Internal)
Variables we can measure but lack data for
- Data Continuity: Users removing CGMs or failing to log data.
- Prediction Accuracy: AI missing individual metabolic variability.
- Engagement: Long-term retention with continuous monitoring.
- Interpretability: Users misunderstanding probabilistic outputs.
- Algorithmic Bias: Models trained on non-representative baseline data.
Q2: Known Unknowns (External)
Market variables with probabilistic outcomes
- Regulatory: Evolving FDA guidance on digital health / clinical AI.
- Physician Adoption: Providers rejecting patient-generated insights.
- Pharma ROI: Uncertainty in measurable returns from RWE.
- Employer WTP: Hesitation without immediate 12-month cost savings.
- Adverse Selection: Employers with the unhealthiest populations adopting first, skewing baseline data.
Q3: Unknown Unknowns (Internal)
Emergent second-order behaviors
- Anxiety-Driven Monitoring: Hyper-fixation on real-time data increasing cortisol and worsening metabolic health.
- Goodhart's Law Optimization: Users avoiding healthy stressors (like HIIT workouts) because they temporarily spike glucose.
- AI Dependency: Users losing touch with somatic signals (hunger/fatigue).
Q4: Unknown Unknowns (External)
Systemic market reactions
- The GLP-1 Rebound Liability: If Aurea routes a patient to a GLP-1, they lose weight, stop the drug, and metabolism crashes—who owns the predictive failure?
- Structural Shifts: Sudden changes in CMS reimbursement models.
- Competitive Disruption: Big Tech closing ecosystem APIs to prevent third-party RWE generation.