From hidden books to an online medical encyclopedia

Problem

Late 2018, a product manager from the Springer Nature's German Online Professional Medical Education division came to the domain I was Lead UX for, with the following proposition:

Inside Springer Nature's content database is a set of medical reference books that are only available in print, and are unlike any other medical reference in that:

- There are 9000 of chapters
- Each one about a syndrome, organ, or major process, like pregnancy or diabetes.
- Each is curated continuously by the German-speaking highest clinical authority in that field to reflect the latest insights, treatments, and breakthroughs.
- And written to be educational for clinicians who are new to the topic.

Could we make this whole set of topics available online, cross-linked, searchable, smart, as educational and navigable as Wikipedia?‍

The team immediately started investigating. Our Search specialist investigated the search and automating linking in-text keywords to meaningful basic definitions in other chapters.  The CMS specialists explored how to pull the texts out of their deep storage into a higher-performance environment while still always remaining up-to-date.

Process

Cooperation

I started working on the first target scenario: new residents in hospitals trying to quickly become informed in a hallway on a topic they studied a while ago that suddenly became relevant. We had the fortune the Product Manager knew her audience really well, and had no problem pushing the envelope. For example, her guiding line was to get to content as quickly as possible We could dispense with product logos at the top, big menu options, or pre-ambles like the complete set of authors. ‍

My first thought was that residents would want to see options of where to go for related information at all times; instead the product manager wanted to strip all that furniture down to focus only on content and let the in-text links do the work. I liked the risk she wanted to take, so we kept stripping things down and down to the simplest structure: text with links.

Protoype and first lessons: wall of text

I made a prototype in Axure for user-testing with clinicians in an office setting. I did not conduct that test but studied the tapes and the conclusions: clinicians could find the topic they needed, but had a hard time with the articles. Just putting one of those chapters on their screen, especially on mobile, was completely overwhelming: it felt like an endless wall of text. 'Intimidating' was the word used.‍

This is the problem we would run into a lot: the texts were written for paper, not screen, and there is no way for our team to rewrite that. Could we make this more manageable?

Creating clarity

I tried multiple ways to make the wall of text less intimidating, but in the end the designs we all liked most, based on the testing, is to do what Wikipedia is now also doing, which is to roll up every section on the page under a title. ‍

However, often unrolling a long section would immediately create that intimidating wall of text again, making it hard for a clinician to find what they were looking for. Therefore I designed that expanding a top level sections would not show the full wall of text, but only the levels and sub-levels of headings so the user could make a more informed choice. It takes only one click to see the whole structure of the section, greatly clarifying the text.

Search

Another challenge was search. We knew we had to guide our users as quickly as possible to the right terms — and that for 9000 terms it really would be guiding. We settled on a Search field that would auto-complete suggestions as the user typed, but then the next question was, suggestions of what? Showing only titles of chapters would be enough most of the time, but sometimes would miss specific medical contexts the user was interested in.

The Product Manager and I theorized that full-text clinical search would only be interesting for clinicians that already knew the basics of a topic and were looking deeper, so we opted to only show full text search results in the type-ahead after the chapter titles, and to delineate both options clearly.

We opted to not show the quoted text in the suggestions, but the title of the book the result was in, since the title alone usually gives enough context on how the subject will be discussed. We did, of course have an option to see the text quote.

Results

There were other issues to conquer, like putting paywalls in while allowing the results to still be crawlable for SEO, and creating the mechanism by which new tomes could be added and maintain existing cross-linking. Eventually product management made the decision to call the product Emedpedia and not have it be a stand alone product but to fold it into Springer Nature's German medical Professional Education portal.

The response was instantaneous: the hospitals and medical schools that buy subscriptions to the portal for their clinicians and students let us know they saw this as a unique product that made them even more happy to be with Springer Nature, and sales reported the product featured prominently in conversations with new prospects. Usage statistics soon reached their initial goals, and enthusiasm was high enough to warrant adding more reference books from the starting 18 to the current 33, and to maintain work on the stand-alone mobile apps for the last 4 years now.
‍