Bi-Polar Well-Being Forecast is a personal self-tracking tool designed to help you monitor daily mood swings, sleep patterns, symptoms, and medication adherence over time.

What it does: It helps you recognize objective patterns, spot early warning signs (prodromal shifts) before a full mood episode develops, and organize clear observations you can share with your psychiatrist or therapist.

What it does NOT do: It does not make clinical diagnoses, provide medical advice, recommend treatment or dosage changes, or replace professional psychiatric care.

When you request an AI evaluation, three sources of information from your account are combined:

• Personal Profile (My profile): Provides baseline context—such as your bipolar diagnosis (Type I, Type II, or Undiagnosed), age, gender, weight/height, alcohol intake, and typical physical activity level—so trends are interpreted in context.
• Medication History (Enter / alter medications): Tracks your current prescribed medications, dosages, and any schedule updates over the last 21 days.
• Daily Check-In Reports (Log daily check-in): Captures your day-to-day overall feel (1–5), sleep hours, yesterday's medication adherence, active manic/hypomanic or depressive symptoms (rated 1–5), and personal notes.

Combining these three streams allows the evaluation to highlight meaningful correlations—such as how multi-day sleep changes, missed medication doses, or lifestyle factors relate to shifts in mood and energy.

Recognizing mood phases in bipolar disorder depends on both symptom duration and changes from your personal baseline:

• Clinical Duration Thresholds: A 21-day window covers the full 14-day threshold used clinically to identify depressive episodes, the 7-day threshold for mania, and the 4-day threshold for hypomania.
• Baseline vs. Recent Shift: Comparing weeks 1–3 against the most recent 3–7 days allows the analysis to distinguish your normal baseline from early prodromal shifts (for example, a gradual drop in sleep or rising restlessness over several days).

Living with bipolar disorder can make strict daily consistency difficult—especially during depressive or elevated phases—so you do not need a perfect 21-day streak to use the forecast.

• Minimum bar (5 check-ins in 21 days): A single day is just a snapshot. At least 5 check-ins are needed to establish a minimal direction of change across the 3-week window.
• Recommended (10+ check-ins in 21 days): Logging at least every other day (~10+ entries) provides enough coverage to reliably compare your earlier baseline against your last 3–7 days.

• One check-in per day: Recording one summary per day (ideally around the same time each day) keeps your timeline balanced so a single difficult hour does not skew your multi-week trend.
• Up to 3 evaluations per 7 days: Bipolar mood trajectories and sleep patterns evolve over multiple days rather than hours. Spacing out evaluations ensures each forecast reflects genuinely new check-in data.

Your data is protected through strict containerization and enterprise cloud isolation:

• Containerized Platform & Enterprise AI Isolation: The application runs in an isolated container environment, and AI evaluations are processed via Google Cloud Vertex AI Enterprise (EU region). Under enterprise data governance, your inputs and outputs are strictly isolated within the project container and are never used to train Google's foundation models.
• Encrypted Password Storage: Your password is never stored in plain text—it is hashed using Argon2, the modern cryptographic standard for password security.

Yes. When you click Create new evaluation, the application strips your personal identifiers before building the prompt:

• Your username, email address, and exact date of birth are never sent to the LLM.
• Only de-identified clinical context is included (such as your calculated age in years, gender, lifestyle baselines, medication names/dosages, and your 21-day symptom and sleep logs).
• Privacy Tip: Avoid typing real names, phone numbers, or addresses into your free-text daily Notes field so your logs remain 100% de-identified.

No. Your profile details, medication logs, daily well-being entries, and AI evaluations are strictly private to your account. Your personal data is never sold, shared with third parties, or used for advertising or marketing.

Memory recall during appointments can be challenging—especially when looking back over several weeks. You can review your My status reports table or bring the "Key Observations for Your Care Team" section from your latest evaluation to your appointment to discuss sleep-to-mood correlations, adherence patterns, and symptom shifts with your doctor.

This application is a self-reflection tool and cannot provide emergency monitoring or crisis intervention.