How can healthcare feel
easier to navigate?
Over two months, I worked on how members arranged lab tests, prepared for visits, and understood their results.
Transcript
Humanolab was a two-month project. I worked on how members arranged lab tests, prepared for visits, and understood their results. My question was: how could healthcare feel easier to navigate? I designed and built changes to booking, lab orders, onboarding, and comparisons over time. I wanted each step to feel clear, without members having to ask someone what to do next.
Placeholder narration, not Olivier’s voice.Audio is unavailable. You can read the transcript instead.
Interface examples use synthetic data.
Lab scheduling
Laboratory scheduling, in the member’s hands
Connected location discovery, booking and appointment changes across web and mobile.
After · An independent, guided booking flow
Before
Assessment sequencing could block booking. Finding a location and managing changes could require staff help.
A member can choose a visit and return to reschedule. The assessment no longer has to be completed first.
Design decisions
- Treat a Synlab visit as a walk-in commitment, not a reserved capacity slot.
- Separate assessment completion from plan eligibility.
- Check membership, organization permission and active-plan ownership independently.
My role
I led the journey redesign and implementation across web, mobile, location search, booking, notifications and release verification. I later operated acceptance for the plan-eligibility safeguards.
What changed
Released a self-service scheduling journey with safeguards for active-plan eligibility.
Still to measure. Booking completion and staff assistance avoided are not yet measured. Faster access is the intended benefit, not a quantified result.
Source notes and release evidence
The original scheduler, independent booking and later eligibility guard are separate releases. The source comparison below covers the original scheduler only.
These interface demonstrations use archived components and synthetic data, not live member records.
Orders & catalog
From purchased plan to a trustworthy medical order
Connected versioned laboratory configuration, immutable orders and private delivery.
After · One traceable order workflow
Before
Plan contents, laboratory tests, document creation and operational follow-up lacked one versioned, auditable system.
The plan, the test and the partner’s ordering code have an explicit relationship. Issued orders keep the purchased version.
Design decisions
- Distinguish orderable laboratory tests from measured and calculated biomarkers.
- Preserve issued orders when catalog definitions change.
- Keep deliberate administrator resend controls and delivery history.
My role
I designed and implemented the laboratory data model and operational workflow, then owned production migration, catalog verification, end-to-end QA and release communication.
What changed
Members received clearer arrival and payment guidance.
Still to measure. Order-preparation time, corrections and manual follow-up have not been measured against an approved baseline.
Source notes and release evidence
The order release, governed catalog and arrival guidance shipped in several changes. The representative source comparison covers the medical-order release.
These interface demonstrations use archived components and synthetic data, not live member records.
First-time onboarding
A first session that leads to something useful
Reframed the first visit around booking, optional past results and a resumable assessment.
Before
After · A clear first milestone, then flexible progress
Before
Future-focused promises and rigid sequencing did not quickly show members the product’s current value.
The introduction moves from a broad longevity promise to what a member can understand and do now.
Design decisions
- Avoid scheduling a duplicate appointment.
- Keep historical-result upload optional.
- Preserve the shared upload contract and keep read-only admin previews out of onboarding.
My role
I defined the first-time member journey and implemented the web experience using the existing scheduler, shared upload/review flow, assessment and dashboard.
What changed
The onboarding implementation and follow-up fixes merged into main. The pull request records authenticated local acceptance and production-release approval.
Still to measure. First-session activation and completion gains are unmeasured. Production acceptance has not yet been confirmed for this redesign.
Source notes and release evidence
The redesign is integrated; production acceptance remains unconfirmed.
These interface demonstrations use archived components and synthetic data, not live member records.
Longitudinal results
Repeated lab tests become a history you can trust
Preserved report identity and connected dated comparisons to constrained AI explanations.
After · Each report keeps its place in time
Before
Repeated values could be overwritten or mixed across reports. Charts, scoring and interpretation lacked a consistent current snapshot.
A result belongs to a dated report. Members can compare observations, while stale results and interpretation limits remain visible.
Design decisions
- Separate raw observations from report-level comparison values.
- Require a measurement date when extraction cannot establish one.
- Refresh explanations when the underlying snapshot changes and constrain short-interval language.
My role
I defined and implemented the report/date/snapshot contract and member-facing explanation path, then operated migration reconciliation, authenticated acceptance and corrective follow-up.
What changed
Released dated comparisons, multi-report publication, asynchronous explanation states and immediate upload-history refresh.
Still to measure. Member understanding, repeat-testing engagement and clinical benefit remain unmeasured.
Source notes and release evidence
The first comparison covers report-safe longitudinal analysis. A subsequent pull request adds the member-facing AI explanation and further publication safeguards.
These interface demonstrations use archived components and synthetic data, not live member records.
Personal review only. External-use approval and final attribution review remain pending.