About Ada

Ada Is a mobile app created to bring the future of personalized healthcare to everyone. It uses a chatbot with AI technology to decipher the user’s possible diagnoses and its severity.

Ada's goals

To empower patients to make more informed decisions about their health.

To be proactive in our healthcare without adding the need for greater human doctor resources.

This portfolio was created to display how I would discover and execute problems for an app similar to Ada.

My propositions

See "Solutions" for more details

Body map/Avatar 

Content focused visuals

Symptom tracker

Contents

Research & Discovery

Investigation & Analysis

Concept Development

Solutions

Research & Discovery

Ada's current features

Before I started this project Ada had a few main features that the app revolved around. The primary feature was of course the symptom diagnosis. Secondary features included symptom tracker and a user profile.

Symptom diagnosis

Step 1.

The first step to the diagnosis assessment involves the user typing in their symptom and selecting the most relevant result.

Step 2.

Upon choosing a symptom the user is asked a series of questions. With each question the AI gets closer to determining a diagnosis.

Step 3.

Once the assessment is complete, a list of possible diagnosis appear in order of most likely to least likely.

The symptom tracker is used to track the users recovery. The user chooses their symptom they want to track, then select the severity. The tracker history is also available to scroll through.

Symptom tracker

Ada has a section with the users personal information which is used to help narrow down the users diagnosis during assessment.

User profile

User research

In order to reduce bias, I sought out a diverse range of participants. Examples of variables I focused on were; age, symptoms, professions.

Initial focus points

Whether symptoms were in the past or ongoing


How the user would describe their symptoms
 

Users’ reactions during Ada assessment
 

Users’ input on possible improvements

During the user tests I gave the simple task of completing an assessment of a current or past problem. While the user navigated through the app I observed and took notes.

Remote Research


When researching some participants who work in the medical field, remote research testing methods needed to be set in place.

I chose to create a simple page for the participants to fill out while I spoke to them on the phone.

I feel I should have approached this in a more observant way. That said some interesting results still arose.

Investigation & Analysis

Insights

While reviewing the findings from the research I took note of insights found. After the notes were taken I grouped insights based on similarities.

I decided it would be a good idea to create a 'how might we question' for each group of insights.*

* ‘How Might We Questions’ is a design-thinking method where statements are composed to provoke brainstorming.

Goals

To increase the users’ experience via clarity during assessment.


To optimise existing features to a higher potential.

Questions

User journey

I wanted to map the journey of all possible users to get a clear understanding of possible areas of pain points by placing the ‘How Might We’ notes on the map.

The outcome showed clear areas of pain points.

Concept Development

Initial concepting

Once the map was complete I gave myself 15 minutes to note some initial concepts I had for various stages of the users’ journey.

Notable inspiration

Before continuing with more concepting I decided to check other designs which could spark inspiration for possible solutions. When researching relatable product interfaces, I gathered and evaluated multiple examples of designs that would combat some of the problems found in during the research.

Cuberto Hyperloop

With this Hyperloop ticket concept design, when the first route selection was chosen the map would zoom in to the next option (finer detail).

Tesla Motors

Tesla Motors’ car interface shows the anatomy of the car along with technical information. This clearly shows the information for each part, allowing the user to feel more comfortable and safe.

Cuberto seat selection

Airline seat selection concept design shows a simple and understanding interface. As you scroll through the plane seat layout, the information on the left changes relative to the plane section.

Concept & evaluation

I went into more detailed concepts and follow-ons from some of the initial concepts. I followed this up with critical evaluations and further iterations. After multiple iterations it became apparent that there were three directions I wanted to investigate more. 

Body map/Avatar

Symptom tracker

Content focused visuals

To narrow down which solution I wanted to focus on with each direction, I performed an exercise called 'crazy 8'. With this exercise, you give yourself 8 minutes to try fill eight sections of a page, each section containing a different iteration.

Body map/Avatar

Symptom Tracker

Content focused visuals

Solutions

Symptom tracker notification

Original symptom tracker user flow

Problems found

There was a lot of doubt about whether the symptom tracker would be used regularly.


When symptoms are asked to be tracked they are frequently forgotten unless pain is present. This will give biased results for the doctor.

Area on user journey

How does this concept tackle its challenges?

This concept of mine will help increase use of the symptom tracker and take advantage of it's potential to be a valuable resource for doctors. The goal here is to not only track the patients symptoms when they are bad but also when they are good.

Extra growth benefits; this will not only increase user retention, but will also increase user acquisition through acquiring doctors then possibly more patients.

Doctors journey

The doctor can select their patient’s notification frequency in relation to the required amount of data about their patients recovery.

The doctor can check each patients symptom history and/or change notification frequency. This area will only store symptom tracking data.

Patient journey #1

Having easy, interactive notifications will increase users engagement.

Having a graph relative to time will give a better indication to patient’s speed of recovery.

Patient journey #2

Voice logs have the ability to create short, fast and specific notes. This can mean more accurate and personalised healthcare.

Negatives:


There is no graph meaning no visual indicator to performance/recovery.


The doctor may not want to listen to many recordings for each patient

Content focused visuals

Examples of current questions during assessment  

Problems found

A large consensuses of users felt some questions were unclear and took some thought to understand.


As mentioned by the medical experts during research, identifying the type of pain felt during an assessment can prove to be very important.


For a lot of assessments, examples of movements which most reduce pain are very valuable.

Area on user journey

How does this concept tackle its challenges?

I developed this concept to help clarify some questions during the assessment that some people from the general public may find difficult to understand instantly. The less brain power needed to engage with the app, the less stressful the experience will be.

Animations of physical movement to determine the severity of pain may increase the user’s experience.

Negatives:

Even though knowing the position of pain can be useful info to an examining doctor, it might not be as useful for an AI evaluation.

Slider to control position of the leg

Colour coded areas with corresponding visuals can allow the user’s mind to relax

Having an animation that imitates the type of pain felt might lead to more accurate answers from the user

Body map/Avatar

Examples of current questions during assessment  

Problems found

When searching for a symptom, the list of choices can seem overwhelming and very alike. Users spent extra time going through each symptom.

Area on user journey

How does this concept tackle its challenges?

Similar to the previous concept, I created an avatar to use as a visual aid to increase the user experience. Having visuals for each description can allow the user to skip through symptoms that are irrelevant to them, meaning less reading and trying to figure out the correct selection.

Symptom selection

Rather than scrolling through descriptions, users swipe through description cards accompanied by visuals.

The benefits of this are to clarify the symptom descriptions and to help relax the user.

Original profile screen

Profile avatar

The avatar could be used as a profile area with all of the user’s information. This area also includes the symptom(s) tracker.

Tap the grey circles to enter symptom history/tracker.

Reflection

If I had more time and resources I would have loved to redo the remote user testing using tools like Userlytics or UsabilityHub. Another option I could have tried is doing a Google Hangouts video call with the user during the test. 

I would have really enjoyed testing out the devised solutions if I had the time and resources. Creating a high fidelity working prototype with animations may have been difficult, unless the symptoms the user had were know ahead of development.