MoodBoss — when science works for your happiness. We turn biological rhythms and context into readable Stress %, Vital % and six core emotions. Below is what we use, how it works, and its limits.
MoodBoss is a self-awareness tool — not medical advice and not a diagnosis.
Inputs we use
- Rhythms: female cycle phases, male daily rhythm.
- Profile: age, BMI (optional).
- Context: time of day, basic habits (sleep/activity if you opt in via Apple Health / Google Fit), manual mood & symptom check-ins.
- Privacy: integrations are opt-in; export/delete supported.
Outputs
- Stress % (0–100): “dark share” of current affect (tension/over-activation).
- Vital % (0–100): subjective energy to act.
- Six emotions (0–95%): Joy, Interest, Calm, Anxiety, Sadness, Anger.
- Emo-weather: labels from Calm to Storm + 24–72 h windows.
How it works (in 4 steps)
1) Normalization
Unify heterogeneous inputs to a 0–1 scale with sex/age adjustments.
2) Rhythm layer
Phase/daytime patterns serve as priors that shape expected variation.
3) Emotion mixing
Each emotion = weighted blend of rhythm and context; weights calibrated on anonymized data.
4) Aggregate indices
Stress % ≈ normalized sum of tense components; Vital % ≈ tone + positive activation − fatigue.
Technically: Bayesian composition of priors & observations + exponential smoothing.
24–72 h forecast
- Model: rhythm-aware state-space with recent observations.
- Output: probability windows (calm / boost / storm) with alert thresholds.
- Use: schedule meetings, workouts, recovery and focus time.
Colored ticks show typical alert thresholds; your personal thresholds adapt over time.
Validation
- Time-series concordance with self-reports (emotions, sleep, fatigue).
- Rolling back-tests for forecast error (MAE / CRPS).
- Utility metrics: fewer last-minute reschedules, more focus minutes, better adherence.
Limitations & proper use
- Does not replace clinical evaluation or therapy.
- Early personalization is limited; accuracy improves with your check-ins.
- Acute events (illness, acute stress) may produce outliers not always predicted.
Privacy
- Encrypted at rest and in transit; export / delete available on request.
- Health/Fit integrations are optional; the website and pixels don’t receive medical data.
- See the Privacy Policy for details.
Glossary (short)
- Rhythms: periodic variations (circadian/cyclical) shaping baseline.
- Smoothing: reducing noise in time series for stable indices.
- Forecast windows: intervals more likely to intensify/relax emotions.