Ways of Healing
An interactive field atlas of culture, illness, and care.
Field Atlas · Medical Anthropology · Exhibition UX · AI Collaboration · Senior Capstone
- Problem
- Before this course, medicine meant modern medicine to me. When I, a family member, or a pet got sick, I looked for the newest Western treatment and nothing else. The readings showed how differently cultures understand illness and care, and that judging a traditional practice without its historical and cultural context is not a fair judgment. The material was large and scattered, and I wanted to share it in a form that was structured, interactive, and interesting to someone outside the course.
- Process
- Paper sketches and AI prototypes helped me compare three homepage directions. With Claude and Codex, I developed the atlas through three rounds of source review and two rounds of user feedback.
- Result
- A working atlas brings 17 case stories, 10 concepts, and 57 sources into one place, with routes for browsing, comparison, and further reading. User feedback informed revisions to typography, navigation, and reading controls.
Project Summary
How the project started
The idea came from a medical anthropology course I took in 2024. I had not expected the readings to change how I thought about medicine, but they did: practices I would once have dismissed made sense once I saw the history, geography, and beliefs around them. Keeping up with the reading was difficult, though, and I imagined a website where someone could begin with a story that interested them and explore the concepts around it.
The idea became my senior capstone: an atlas of published accounts of illness and care, including how families understand a diagnosis and how social conditions affect access to treatment. The stories come from existing research; I did not conduct the fieldwork behind them.
My earlier project, Where Dragons Dwell, gave me experience organizing cultural material through an interactive map. For Ways of Healing, I needed to reconsider the tone and presentation for stories involving illness, trauma, and poverty.
Audience & Design
Choosing the audience and visual direction
Early critique helped me narrow the primary audience to culturally curious adults who were new to medical anthropology, which is who I had been before the course. Some might approach the subject through caregiving or experiences within a multicultural family. Educators were a secondary audience who could use the atlas to introduce a discussion.
That priority kept the atlas focused on introducing the subject. Portfolio reviewers would also see it, but their interest in the design process would not determine the content presentation.
Three homepage sketches gave me options to compare:
- A flat world map with an index and filters.
- An image-led gallery.
- A rotating vector globe.
After prototyping each direction with AI, I chose the globe. The sketches also established the starting layouts for the case pages, Comparison Lab, and Bookmark Workshop.
Field notebooks and museum archives informed the warm cream backgrounds, terracotta, olive, and deep navy. Case files, stamps, and paper-like cards gave the pages a consistent structure, with limits on decoration appropriate to the subject matter.
Atlas Experience
Exploring the material
The globe introduces the 17 cases geographically. Visitors can rotate it, select a case, or use the region and theme filters to narrow what they see. A case index provides another way to browse.
Five routes let visitors follow their interests: cases, concepts, themes, comparison, and the Atlas Guide.
Reading a case and its concepts
Each case story introduces the people and setting before explaining the cultural questions involved. References remain available alongside the material.
I divided longer explanations into smaller sections using flip cards, expandable content, and questions and answers. This gave visitors some control over how much detail they opened at once.
The ten core concepts provide another route through the collection. Each concept has a plain-language explanation and connections to relevant cases, so readers can move between an idea and an example.
Comparing cases
The Comparison Lab uses a filing-cabinet interface. Visitors can pull two or three cases onto a shared table and compare them using six questions:
- Where is the cause of illness located?
- How is healing carried out?
- How are body, spirit, and society connected?
- Who is recognized as a healer?
- What role do family and community play?
- What is at stake for the people involved?
The comparison can also be printed. Using the same questions across cases gives visitors a consistent structure for examining their differences.
Asking the guides
The AI features are connected to a live language model, with two distinct roles:
- Atlas Guide helps visitors find relevant cases and explore questions across the collection.
- Case Guide focuses on the material within an individual case.
I scoped both guides to educational questions about the atlas. They are designed to decline requests for personal medical advice and direct those questions to a qualified professional.
Making a bookmark
The Bookmark Workshop offers three daily draws. Visitors can choose a bookmark, add an illustration and their name, then print it.
This souvenir corner gives visitors a small activity after reading and something to take away.
The wandering figures
Twelve cut-paper figures move along the bottom of the pages. Their identities come from what they carry, such as a basket, lantern, or book. Clicking a figure pauses it and reveals a short line.
Visitors can turn the figures off while reading.
My Role
My role and AI collaboration
This project was built in collaboration with AI; the design decisions, content, and direction are mine. Claude was my primary AI collaborator, and I also worked with Codex. I did not write the application line by line.
I defined the concept, content scope, visual direction, cultural-sensitivity standards, and page interactions.
AI tools helped synthesize the source material and draft the site’s copy. They generated illustrations under my direction, implemented the front end, and carried out citation checks using criteria I defined.
Each page went through individual review and revision before I approved it.
Alongside content review, I handled performance testing, API credentials and environment variables, version control, and deployment to Vercel.
Accuracy
Checking the content and illustrations
The initial content drafts drew on roughly 45 medical anthropology sources. Reviewing them revealed problems that would have been easy to miss in polished prose, including unsupported statistics, incorrect reference titles, and misattributed claims.
Three rounds of source review followed four requirements I set for the AI checks:
- Remove figures or descriptions that could not be supported by a source.
- Trace claims to the supplied readings or another cited academic source.
- Check titles and authors, and use page numbers only when they could be verified.
- Confirm that each reference supported the claim attached to it.
The reviews removed invented statistics and corrected numerical and citation errors. I kept the wording general where a specific figure could not be confirmed. The resulting bibliography contained 57 works, collected on the Sources & Safety page.
The illustrations needed the same scrutiny. I reviewed all 51 individually. In one early HIV illustration, the AI had drawn a coronavirus, so I rejected it.
I also set visual rules for sensitive material: no clearly depicted faces, wounds, or human remains, and no use of suffering as decoration. These requirements guided both the image prompts and my decisions about which images to keep.
User Research
What the research changed
Feedback survey
In June 2026, eight people completed a feedback survey. The reported ratings were:
| Area | Rating out of 5 |
|---|---|
| Visual style | 4.5 |
| Ease of navigation | 4.0 |
| Text clarity | 3.4 |
| Overall experience | 4.4 |
Text clarity received the lowest rating, and the written comments also pointed to readability. I increased type sizes, reduced the use of the handwritten typeface, and moved How to Explore out of the global navigation while keeping it available as a secondary resource.
Moderated usability session
On June 30, 2026, I conducted a moderated think-aloud session with Kevin, a 40-year-old engineer who visits museums.
His session revealed specific points of confusion and suggestions to weigh against the atlas’s existing structure:
| Feedback or observation | My decision | Status |
|---|---|---|
| Some case and concept pages still showed an old Explore navigation link. | Remove the leftover link. | Applied |
| Both AI assistants were called Field Guide, making their purposes difficult to distinguish. | Rename them Atlas Guide and Case Guide, with a scope note for each. | Applied |
| The moving figures became distracting during reading. | Add a visible on/off control that remembers the visitor’s choice, alongside reduced-motion support. | Applied |
| Labels such as Across the Atlas were unclear. | Use more direct wording, including Related Cases. | Applied |
| A suggestion was made to rename By Theme as By Case. | Keep the distinction between thematic browsing and the case index. | Not adopted |
| A concept filter was suggested for the homepage. | Defer it until further testing could establish whether another filter would help. | Deferred |
On smaller screens, scroll sideways to see all columns.
The session notes recorded Kevin’s wish to begin exploring without reading instructions first, supporting the earlier decision to make the how-to page a secondary resource.
The survey and session informed these revisions. A follow-up study would be needed to measure whether the changes improved readability or navigation.
Outcome
Delivery and reflection
Ways of Healing was deployed as a working site built with HTML, CSS, and JavaScript, using D3 for the globe. A shared content catalog supplies the case and concept information, allowing the same material to be reused across the atlas.
The atlas is the introduction I would have wanted before the course: one story at a time, with the context that makes the practice understandable and the sources that let a reader check it.
Testing helped me judge the exhibition style by how people used it. I could retain its visual character while changing labels and controls that interrupted reading.
Content review demanded the same attention: a citation or illustration could change what a visitor took from a page. Those judgments remained my responsibility throughout the AI-assisted build.
Explore Ways of Healing
Browse the cases, try the Comparison Lab, or make a printable bookmark.