User Profile¶

UserProfile is the central domain model for all data collected during onboarding. It is stored as a JSON-serialized string in DataStore Preferences and used by the simulation engine, the AI context builder, and the Home screen.

Data Class¶

@Serializable
data class UserProfile(
    // Identity
    val displayName: String? = null,
    val ageYears: Int? = null,
    val biologicalSex: BiologicalSex? = null,
    val heightCm: Int? = null,
    val weightKg: Float? = null,

    // Manual lifestyle baselines
    val averageSleepHours: Float? = null,
    val averageDailySteps: Int? = null,
    val perceivedStressLevel: Int? = null,  // 1..10

    // Current-day overrides (Home screen)
    val currentSleepHoursOverride: Float? = null,
    val currentStepsOverride: Int? = null,
    val currentHeartRateOverride: Int? = null,
    val currentStressLevelOverride: Int? = null,

    // Habits
    val smokingStatus: SmokingStatus? = null,
    val alcoholDrinksPerWeek: Int? = null,
    val dietQuality: DietQuality? = null,

    // Bookkeeping
    val onboardingCompletedAtEpochMs: Long? = null,
)

All fields are nullable. null means "not yet provided" — never conflated with 0 or any other sentinel.

Field Groups¶

Identity¶

Basic demographic data. Used in the simulation engine (age contributes to long-term risk) and in the AI context prompt (age, sex, height, weight).

displayName is shown in the Home screen greeting only — it is not included in the AI context to preserve the "speaking as yourself" persona.

Manual Lifestyle Baselines¶

Collected during onboarding Step 3. Used as fallbacks when no wearable data is available:

averageSleepHours → fallback for SimulationInput.sleepHoursAvg7d
averageDailySteps → fallback for SimulationInput.stepsAvg7d
perceivedStressLevel → fallback for SimulationInput.stressAvg7d

Current-Day Overrides¶

Set from the Home screen's manual override dialogs. Take precedence over wearable data for the currently displayed day. Historical wearable records are never touched.

Override priority: currentOverride > wearable > baseline fallback > null

Habits¶

Always manual — no wearable measures smoking or diet. Fed directly into the simulation risk scores.

Enum	Values
`SmokingStatus`	NEVER, FORMER, OCCASIONAL, REGULAR
`DietQuality`	POOR, AVERAGE, GOOD, EXCELLENT
`BiologicalSex`	MALE, FEMALE, OTHER, PREFER_NOT_TO_SAY

Persistence¶

UserProfileDataSource serializes the entire UserProfile to a JSON string using kotlinx.serialization and stores it as a single DataStore Preferences key.

// Simplified read/write pattern
val profileFlow: Flow<UserProfile?> = dataStore.data
    .map { prefs -> prefs[KEY_PROFILE]?.let { Json.decodeFromString(it) } }

suspend fun save(profile: UserProfile) {
    dataStore.edit { prefs ->
        prefs[KEY_PROFILE] = Json.encodeToString(profile)
    }
}

Using JSON rather than Proto DataStore avoids the schema-compilation step, which matters at hackathon speed. The stored JSON is human-readable and inspectable with adb shell.