[Feature] Expand the encrypted health profile with a structured health and lifestyle questionnaire

[lutzkind]

Summary

Expand HealthLog’s existing encrypted “About me” / self-context profile into an optional, structured, effective-dated health and lifestyle questionnaire that can personalize Coach guidance, contextual analysis, screening, and future adaptive scoring without duplicating data already stored elsewhere.

HealthLog already has a useful foundation: encrypted free-text self-context, chronic conditions, allergies/intolerances, Coach focus, profile age/gender, structured allergies, structured family history, and follow-up questions. The proposed feature should build on that architecture rather than replace it.

Problem

A measurement can have very different meaning depending on the tracked person’s context. Examples include:

• An outdoor worker versus someone who remains indoors during high heat or pollution.
• A shift worker versus someone with a consistent daytime schedule.
• A smoker, former smoker, or never-smoker.
• Someone who drinks frequently versus rarely.
• Someone with known asthma, migraine, sleep apnea, or exercise limitations.
• Someone following a particular diet or regularly consuming late meals.
• Someone whose primary goal is cardiovascular fitness versus medication adherence or sleep improvement.

Currently, much of this context can only be entered as free text. Free text is valuable, but it is difficult to validate, update over time, selectively disclose, analyze consistently, or use transparently in scoring and Coach explanations.

Proposed design

1. Optional structured questionnaire

Provide modular sections that users can complete gradually and skip freely.

Suggested sections:

Personal and medical context

• Existing chronic conditions and relevant medical history.
• Functional limitations or contraindications.
• Known sleep disorders.
• Existing structured allergies/intolerances and family history should be referenced rather than duplicated.
• Health goals and clinician-defined targets where available.

Tobacco and nicotine

• Never, former, occasional, or current use.
• Product type.
• Typical frequency and approximate quantity.
• Years used and quit date where applicable.
• Second-hand smoke exposure.

Alcohol

• Typical drinking days per week.
• Typical quantity per drinking day.
• Optional binge-frequency context.
• Whether the user wants alcohol-related pattern analysis.
• Optional standardized screening such as AUDIT-C/AUDIT, clearly identified as a screening tool rather than a diagnosis.

Exercise and movement

• Typical exercise days per week.
• Aerobic, resistance, flexibility, sport, and recreational activity.
• Sedentary work or physically demanding work.
• Exercise goals.
• Injuries or restrictions.

Diet and eating pattern

• General dietary pattern.
• Meal timing and typical late eating.
• Typical caffeine intake and timing.
• Food restrictions or intolerances through existing structured records where possible.
• Weight-management or nutrition goals.
• Optional high-level diet-quality questions without requiring a full food diary.

Sleep and schedule

• Typical bedtime and wake time.
• Shift work or rotating schedule.
• Regular naps.
• Snoring, known sleep apnea, or sleep medication.
• Bedroom noise, light, and temperature context.
• Typical travel frequency and social jet lag.

Occupation and environment

• Indoor, outdoor, or mixed work.
• Heat, cold, dust, fumes, solvents, noise, or other occupational exposure.
• Air-conditioning and ventilation.
• Air purifier use.
• Relevant pets or allergens.
• Commuting pattern.

Mental, social, and practical context

• Typical stress burden.
• Caring responsibilities.
• Social support.
• Constraints that materially affect recommendations.
• Preferred Coach focus, goals, and communication style.

Optional broader substance screening

• Standardized tools such as WHO ASSIST may be offered as explicit opt-in modules.
• HealthLog should not invent opaque proprietary risk scores where validated instruments already exist.

2. Structured data plus free text

Retain the existing “About me” free-text field. Structured answers and free text serve different purposes:

• Structured answers support consistent interpretation and transparent feature logic.
• Free text captures nuance, identity, priorities, and circumstances that a fixed form cannot.

The UI should show both and explain how each may be used by Coach or analytics.

3. Effective dating and history

Lifestyle facts change. A questionnaire answer should not overwrite history as though it had always been true.

Examples:

• Current smoker → quit on a particular date.
• Started shift work.
• Changed diet.
• Began exercising regularly.
• Moved home or changed work environment.

Support either revision records or validFrom / validUntil semantics so historical analysis uses the context that applied at the time.

4. Reuse existing authoritative records

Do not duplicate:

• Age, gender, or date of birth from the main profile.
• Medication lists and dose schedules.
• Structured allergies.
• Family-history records.
• Diagnoses or illness records where HealthLog already has an authoritative model.
• Daily lifestyle events proposed in [Feature] Add a unified contextual-factor timeline and structured lifestyle-event logging #613.

The questionnaire should reference or summarize those records and only store profile-level information that does not already have a canonical home.

5. Coach and analysis integration

• Inject only relevant structured profile facts into Coach prompts through the existing fenced self-context architecture.
• Preserve provenance: user-reported profile facts must be identified as self-reported.
• Do not infer unstated diagnoses or habits.
• Show users which profile facts influenced a recommendation or explanation.
• Let users disable selected sections from Coach use without necessarily deleting the underlying record.
• Use profile context to prioritize analysis, not to force a conclusion.

Examples:

• Prioritize air-quality and pollen context for a user who reports asthma or allergies.
• Consider shift work when interpreting irregular sleep timing.
• Avoid offering nicotine-reduction guidance to a never-smoker.
• Prioritize migraine-related pressure, alcohol, sleep, and hydration relationships for a user who asks the Coach to watch migraines.

Privacy and security requirements

• All sensitive questionnaire answers encrypted at rest using HealthLog’s existing encrypted-column conventions.
• Explicit opt-in and section-level skip controls.
• No requirement to complete the questionnaire to use HealthLog.
• Clear source labels: self-reported, structured record, imported, or measured.
• Full export, correction, and deletion support.
• Audit changes without placing plaintext health answers into logs.
• No raw profile content in telemetry or error reporting.
• Prompt injection protections equivalent to the existing self-context fence.
• Module or operator controls where appropriate for multi-user/self-hosted deployments.

UX principles

• Progressive completion rather than one long mandatory onboarding form.
• Save each section independently.
• “Prefer not to answer” where appropriate.
• Review/update reminders at a low frequency, not repeated nagging.
• Display when an answer was last updated.
• Allow Coach-generated clarification questions, but require explicit user confirmation before adopting an answer into the persistent profile.
• Explain whether a field is used for personalization, screening, correlation prioritization, or scoring eligibility.

Acceptance criteria

• Users can complete optional structured sections incrementally.
• Existing encrypted free text remains available.
• Answers are encrypted at rest and excluded from plaintext logs.
• Time-varying answers retain effective history instead of rewriting the past.
• Existing canonical medication, allergy, family-history, demographic, and event records are not duplicated.
• Relevant profile facts can be safely composed into Coach and briefing context with provenance and fencing.
• Users can see, edit, export, delete, and selectively exclude profile sections from AI use.
• Standardized screening modules, if included, are clearly identified and remain optional.
• Tests cover encryption, authorization, history/effective dates, prompt composition, redaction, export, and deletion.

Relationship to other work

• Daily and timed events belong in [Feature] Add a unified contextual-factor timeline and structured lifestyle-event logging #613 rather than this long-term profile.
• Personal pattern discovery should consume the profile as context but should not treat questionnaire answers as proof of causation.
• Adaptive Health Score eligibility may use profile information, but score design should remain a separate request.

[MBombeck]

Thank you for this, @lutzkind, and for the care you put into writing it up. Expanding the encrypted profile with a structured health and lifestyle questionnaire makes sense, especially for grounding the insights.

I want to be straight about timing: the near-term focus is on making the foundation rock solid, reliability, sync, and the quality issues you and others have surfaced, rather than new features. So I am capturing this as a roadmap item rather than starting on it right away. It is noted and valued, and I would rather build it well later than rush it now. Please keep the ideas coming.

[MBombeck]

v1.34.0 ships a first slice of this: smoking status, alcohol pattern, and shift or rotating-schedule work can now be recorded as dated, correctable entries in the encrypted health profile, each with its own switch for whether it reaches the Coach's prompt. This is not the full structured questionnaire the issue describes; it is the three facts that also feed the correlation engine most directly right now. Broader profile fields remain open. Refs #614.