We’re going to analyse the “WebMD Symptom Checker with Body Map” as a recommender system.
WebMD’s Symptom Checker system allows you, the user, to verbally describe your symptoms to select tags of symptoms. You can also click on parts of the body to get a list of suggested symptoms that affect that part of the body to select from additional symptoms. As you compile symptoms it offers potential diagnoses with a likeihood or relevance meter for you to take into consideration. Any of the diagnoses can be selected to see additional articles about the condition.
Here is a blog article that roughly describes Scenario Design from Forrester (2004): forrester.com/forrester-introduces-a-disciplined-approach-to-customer-experience
I would describe it as going from a product/company-centric viewpoint to a user-centric view point. By viewing your product from the lens of a user you are able to better design the product to fit the user’s needs. Scenario Design identifies facets of the user experience, or clusters similar user experiences so that can be individually improved.
The three main questions asked in Scenario Design Analysis are:
- Who are your target users?
- What are their key goals?
- How can you help them achieve their goals?
This means to me
1) describe multiple people that represent the different kinds of
customers who will experience your product
2) identify what they want to accomplish
3) come up with design options that can help them satisfy their
goals
The Users
User 1 (1) Hank’s foot hurts, and (2) he wants to know the range of possible issues that he could be dealing with. (3) By clicking on the foot of the ‘Body Map’ Hank is given a list of tagged symptoms to help him articulate, in a way that’s meaningful to our search engine and metrics, what’s wrong.
User 2 (1) Iris has funny rashes, sore joints and is feeling very overwhelmed. (2) She wants to know the range of possible issues that she could be dealing with. (3) By offering her a field to describe her symptoms, we’re able to offer her a series of tagged symptoms that she can choose from to create a picture of her health that’s meaningful to our search engine.
We should always consider scenario design for internal users as well.
User 3 (1) Janic attributes metadata and formatting to articles once they are ready to upload. (2) He needs to make sure that any new articles interface with the Symptom Checker. (3) We design a process so that authors submit the symptom tags and body map tags and have those checked for reasonableness during the article upload setup.
User 4 (1) Kaya is curious about medicine and (2) wants to learn more about specifically gastrointestinal issues. (3) Our user design doesn’t address this at the stage of Kaya first interacting with the Symptom Checker, however once she finds one article about gastrointestinal issues, she will be recommended related articles after at the end of her first article. We could consider creating a medical library interface for people to learn and reference academically.
User 5 (1) Lars is going on a multi-country trip to South America and (2a) wants to improve his medical literacy before potentially getting sick in a foreign country and (2b) once he is in country he may need to look up issues. (3a) We don’t have a strong user experience for Lars. We should consider creating multiple field medical guides which can be hiking or country-specific. Each could be an organized aggregation of our standalone articles. (3b) Once he’s in country he may not get suggestions appropriate to where he is at. Maybe we can consider using IP address or requesting location information as part of the Symptom Checker, to improve the types of articles we suggest. Additionally if we extract the user interaction into a realtime database of symptoms by location we could partner with the CDC to track health trends.
We’ve imagined the abstract, underlying structure of the Symptom Checker and suggestions for new features in the descriptions above however we haven’t described the match-meter.
With each suggested condition which matches the supplied symptoms there is a four-bar meter indicating the strength of the match from low to fair to good to high.
Another internal user profile, User 6, could be (1) Maya, who is in charge of the model, who (2) wants to check that the match algorithm is providing meaningful suggestions. (3) Maya could use metrics like length of time spent reading the suggested article or if they left the site or continued to read articles about other conditions.
I read the article WebMD submitted to PR Newswire in 2018, prnewswire.com/webmd-launches-redesigned-state-of-the-art-symptom-checker. The symptom checker was redesigned to take into consideration feedback from a 2015 British Medical Journal article that pointed out the flaws in online symptom checkers at the time. It also added the ability to type in general symptoms which then suggests a list of tagged symptoms to choose from because users had difficulty picking a part of the body map for generalized symptoms like fever.
But the main innovation continues to be the mapping of symptoms and diagnoses to tags. From the article above, “WebMD has made this possible by mapping clinical symptomology and diagnoses into easy-to-understand language that will be accessible to millions of users.”
WebMD has the opportunity for additional products related to their Symptom Checker:
They can create a medical libary interface for people to learn and reference academically. This could be expanded by offering WebMD certificates in learning specific to general topics. A prime first product motivator might be expectant parents who want to prep for being a medical advocate for the pregnant mother and baby to be.
They can create field medical guides for activities (hiking, ultra marathons, ice climbing) or for specific countries and regions. Travel medicine information could include recommended medical steps before travel such as a Typhoid booster and considerations in country such as insect management, altitude management, and recognizing the disease risk factor profile unique to that country or region.
They can build out a database based on location, symptoms described and when, to create epidemiological models. Not only would location improve the suggested articles, it would expand WebMD’s revelance to global populations and increase the user base. Also an epidemiological model based on symptoms could contribute to academic papers or increase the ability for the CDC to leverage relationships with tech firms to accomplish their mission.
To facilitate collection of location information WebMD can add location to the Symptom Checker questions, or automatically pull IP Address information and ask if the location is correct or if the person had recently travelled else where.