Model card

    What Myelina's energy-pattern engine does — and, just as important, what it doesn't.

    What it does

    Myelina reads your own check-ins, reflections, and (if you connect one) your wearable to surface patterns in your energy that you can plan around — a likely afternoon window to protect, a habit that tends to move with better mornings. It observes; it does not diagnose.

    What it uses

    • Your daily energy check-ins and evening reflections.
    • Medications you choose to log (name, dose, timing).
    • Optional wearable imports: sleep, HRV, step count.
    • Optional weather signal for your saved location.

    What it doesn't use

    • Any other user's data. Patterns are personal, not population-wide.
    • Insurance, employment, or credit signals.
    • Third-party ad or tracking profiles.

    How we keep it honest

    • 60-day minimum. Correlations only surface after ~60 days of your own data, so a good week can't masquerade as a pattern.
    • Plain-language confidence. We show a 95% confidence range around every correlation, and describe it in English (e.g. "somewhere between a small and a moderate link").
    • Multiple-testing correction. When we test many possible patterns at once we apply the Benjamini–Hochberg false-discovery-rate correction, so a handful of coincidences don't get reported as real.
    • Illustrative mockups are labelled. Any sample energy forecast on the marketing site is marked "Illustrative only" and does not come from your data.

    When to ignore it

    • New or changing symptoms — call your neurology team, not Myelina.
    • Medication decisions of any kind. Myelina will never suggest deferring an appointment, a dose, or a prescription refill.
    • A day that doesn't feel like your data. Your body is the ground truth; the model is a second opinion at best.

    Citations

    The engine's assumptions draw on published MS-fatigue research, including:

    • National MS Society — prevalence and symptom prevalence data (NMSS, 2024).
    • Kratz AL et al. (2017) — day-to-day variability of fatigue in multiple sclerosis.
    • Blikman LJ et al. — energy conservation management for MS-related fatigue.
    • Chen MH et al. (2025) — wearable signals and fatigue prediction in MS.

    Questions or corrections

    Email hello@myelina.health. We update this page whenever the engine's rules change and log the change in the changelog.