Interactive Report System

12 week intern at iRhythm | Summer 2019 | San Francisco

*After the summer, I was luckily offered an extension to work as a part-time intern for the Fall 2019 semester. Thus, this project is still evolving at the moment!

Product DESIGN, PROTOTYPING

Type

A "northstar" project for the web application

Foundation

Preliminary UX research insights

Deliverable

Ideal UX flow to be incrementally shipped in the next few years

Tools

Figma, Sketch, Zeplin, Principle

My ROle

As the only Product Design Intern within the young UX team in a big healthcare tech company, I worked closely with:
    - 3 Product Designers,
    - 3 User Researchers,
    - 2 Product Managers,
    - 1 System Engineer,
and scheduled multiple inner research sessions with developers, cardiologists, clinical techs, and sales roles.

During the scope of this project, I'm mainly re-designing the report presenting experience for our web portal -- ZioSuite.
‍
I conducted inner research sessions, understood technical structure and limitations, created 10+ iterations of prototype, and proposed a polished ideal key user experience of digesting clinical data.

Context

A bite of background about iRhythm

iRhythm is a digital healthcare company focused on identifying cardiac arrhythmias through biosensors, ML, and data analytics.

Our digital tools are used by health providers, and over 1 million patients have worn our device, ECG (Electrocardiogram) patches (Zio AT and XT).

‍

Opportunity

The new Zio AT devices can have numerous transmission reports. That’s where the previous report platform starts to break down.

For Zio XT (traditional heart rate monitor), each device only generates one final report. Previous platforms like ZioReport can handle well.

However, since Zio AT devices are normally prescribed for patients who need special care, there can be numerous transmissions reports during the wear period.

PROBLEM

How might we provide users a more digestible and efficient way of consuming Zio AT “real-time” clinical data?

We spent most efforts on understanding the problem space and actually defining the problem. So far, since customers' behavior pattern for Zio AT devices has not been fully established, it remains the biggest challenge.

However, previous UX research insights suggests that some users feel a little spammed by the overwhelming transmissions. They are also eager for a tool to distinguish artifacts from real arrhythmias.

Solution

Before

After

A list view of PDF reports
Previous ZioReport Portal

Interactive visualization of clinical data
Current ZioSuite Portal

* According to HIPAA (Health Insurance Portability and Accountability Act), all patient information on this page is fake.
‍

Jump to prototype ↓

process Overview

Started with preliminary user research insights, I absorbed inputs from various inner stakeholders in each phase of this "Northstar" project, and proposed an ideal UX flow at the end (of the summer intern period).

Target Users

Medical Assistants are dealing with report logistics everyday. Therefore, they are the major users. Meanwhile, doctors are interpreting the clinical data we collected. Since guiding them to do online interpretations is on the roadmap, we defined doctors as potential users.

Research

"Report-centric" (current mindset of the company)

When we started, we approached the problem from more of a "report-centric" mindset, focusing on
‍
How do we improve the existing report history timeline view?
‍

I dig deep into understand our current report system, how they are generated, grouped and tagged. Then, I proposed a new report grouping rule, from a user-centered perspective with some initial lo-fi mock-ups as instances.
‍

Subject Matter Expert (SME) Interviews‍

With the initial proposal in mind, we conducted interviews intensively with inner stakeholders to hear from different perspectives to better understand the problem and set constraints (technical and business).

Major take-aways from more than 13 interviews:
‍

We were also actively detecting information about user's unmet need to guide the design direction.

For example, roles like Sales Manager and UX Researchers are working with end users everyday. They have accumulated rich secondary insights about users' behavior. Meanwhile, Inner Clinical Experts' work content of interpreting clinical data is similar to doctors'. Thus, their input may reflect the need from a physician's perspective,
‍

* Due to the short time limit and the nature of the project that it's conceptual and not going to be shipped within the next few months, we didn't get the research resources with end users at this stage. If given more time, we would conduct more qualitative and quantitive research to learn user behaviors.

‍

Ideation | Turning point

Actual value lies in the clinical data, instead of PDF reports.

“If I had asked people what they wanted, they would have said faster horses.”

-- Henry Ford

As we gained a better understanding of the problem space, we realized that for target users, the actual value lies in the clinical data. Thus, we determined that the ideal solution might actually look quite different from what we have today.

‍

So we pivoted, focusing on redesigning ZioSuite report service with innovative clinical data aggregation and visualization.

Mock-up, iteration & testing

With new solution direction settled, we started to hone it in by conducting inner feedback session about clinical information, interaction, technical limits, etc.
‍

E.g. Symbol System

Taking one of the most important design decision, the Symbol System, as an example, with a low-fi wireframe as the tool, we started SME interviews again to gather information about how clinical information (various types of arrhythmias) should be presented on the chart. These conversations helped us to understand the goal and challenges of designing symbols for an interactive feature.

Based on the goal, I studied symbol patterns in current reports and mocked-up two directional options accordingly.

Constraints and challenges

Keep in consistent with current symbol system of existing Zio Reports to reduce the learning curve.
Be cautious not to set over reliable expectations about the data during wearing period.

We chose to continue with Episode-based symbol system because

Reducing cognitive burden should be prioritized considering the goal of offering a digestible way of consuming real-time data.
As for data confidence, since final reports are one of iRhythm's competitive differentiators, our approach should be: presenting real-time data and offer statement that "this will get much better when you get our AI-powered comprehensive final reports".

Final choice: Episode-based symbol system

From there on, we worked towards determining symbol hierarchies and retouching visual details.

After information hierarchy rearrangement

After Visual polishing

With similar mindsets and methods, I also mocked-up and iterated other features through inner testing sessions such as time scale interaction, pop-up window information, labels, report navigation, etc.
‍

Current Solution

"Concept Car": a presentation tool and a tool for further researches

This is a long-term project, which means it's unrealistic to address every uncertainty and justify every design decision in 12 weeks. We are still, and will continue to stay in the research phase for quite a long time.

‍

However, at the end of the internship, I still polished the wireframe into a high-fi, interactive prototype with a user-centered best guess for each question since it can help to ground the conversation with stakeholders and end users.
‍

Surfacing up relevant ECG (Electrocardiogram) data

Extracting ECG data from PDF reports to the page surface with arrhythmia labels, allowing users to confirm arrhythmia types, measure durations, and retouch labels easily.

Preliminary Information in a hover-over window

Input from inner clinical experts helped us to extract preliminary information for each type of arrhythmia and put them in hover-over windows. In this way, users can make a quick judgement about the severity of an event and decided how detailed they need to know.

Adjustable time scale

By default, the chart shows a wear period summary for users. However, it also allows multiple zooming level based on users' workflow feedback.

* Clicking on date labels would be a shortcut to zoom in for a daily view.

* Clicking on clustered events would automatically expand them to individual events. According to our usability tests, this is the most common used shortcut.

Consistent visual system for report history timeline

The report history timeline section adopts consistent visual symbol system as the interactive chart. This helps to reduce cognitive burden and emphasized information hierarchy.

Validation with a true end user

“I have this fantasy and that fantasy is the perfect reporting system.”

During the last two days of the internship, I got the chance to validate the design with a true end user- a doctor on our scientific advisory board.
‍
Overall, the feedback was quite positive. The user successfully picked up the visual hierarchy system and said explicitly he would love to use this all the time since it's way more efficient.

We also got suggestions about the interaction with ECG strips, clinical information arrangement, etc.

Next Step

Luckily, I was offered the opportunity to extend the internship and continue working on this project for Fall 2019. With a longer time frame, we plan to focus on:

further understanding the relationship between the new interactive section and existing reports
conducting more user studies about AT habits
performing usability tests about the detailed interactions
taking technical feasibility (short-term and long-term) into consideration and incrementally launch the new feature