Ez Software, an iOS MedTech application

A quick redesign, realize at testing that users won't use the product, strategy review, and new solution to discover

Ez Software

Ez Software is a startup that has a mission to ease the work of clinicians and help home care agencies earn more by adopting a speech recognition (dictation) function.

I was invited to the project as a UX Designer to improve the usability of the OASIS scrubber (patient data fill-out form) using the IOS 15 dictation feature. At the time, the MVP was in the active development stage.

What is OASIS

The OASIS is the Outcome and Assessment Information Set. It is a comprehensive assessment designed by CMS (Centers for Medicare & Medicaid Services) to collect information on nearly 100 items related to a home care recipient's demographics, clinical status, functional status, and service needs.

Home Health OASIS is used for three main purposes: It serves as the basis for insurance reimbursements, a tool that measures the quality of home health care, and a foundation for home health care planning. It must be updated every 60 days.

Quick redesign

At the initial meeting, I was asked to do a “quick redesign of the toolbar to make it look nice” without getting too much into details.

Testing showed that users won't use the product

The testing of a redesigned toolbar showed that the problem was not in UI but with the functionality of the voice recognition.

Most of the time, the users would write down codes in scrubbers or make a selection from several options in existing software, and voice input would take longer to achieve the same result.

Strategy review and improvements

The testing learnings helped me to review the product development strategy. I proposed to the team to learn more about the typical user’s day, analyze the existing products and tools, and identify the parts of the OASIS scrubber where voice recognition would actually help clinicians.

This approach helped us implement voice recognition for sections of the scrubber where all information had to be entered manually.

Designing in a hurry

Fast pace

The team needed the solution as quickly as possible, so I had to redesign the VUI swiftly.

Limited knowledge about users and their work specifics

All we knew was that clinicians are very busy, and the VUI was supposed to help them to save time on routine tasks.

Technical constraints

The project was already mostly hardcoded, and I had to make minimum changes to avoid extension of the project delivery timeline.

Implement the voice user interface to save clinicians' precious time so that they can help more patients

Save time on filling out OASIS scrubber using VUI

The solution could save up to 50% of clinicians’ time on filling OASIS scrubber

Home care agencies revenue increase

Using Ez Software, homecare agencies could increase their revenue by 15%, at least, as their clinicians’ productivity grows.

Beyond the OASIS scrubber

The solution can potentially become an internal API for adoption in other parts of the application.

Learnings from the usability testing allowed us to find a valuable solution

Reviewing the product strategy

I proposed to review the strategy of the product. I urged the stakeholders that we needed research to discover clinicians’ work specifics and how the dictation feature can improve their work.

Research findings: Users

Clinicians are doing an important job helping patients
The more patients they help, the more they feel valued
Their income grows proportionally to the number of patients
Clinicians usually work a stressful 10-hour day on average
They visit 3-5 and sometimes more patients during the day, and not all of them are "easy to deal with."
Their job involves driving between patients' houses
They "barely have time to go to the bathroom."

To better understand the users of Ez Software, I interviewed active clinicians and a nursing department director. The main goal was to learn how they gather the data during their working day and what frustrates and motivates them the most.

Research findings: OASIS

There are different data entry types in the OASIS scrubber. And as the testing showed, the dictation feature wouldn’t be the best tool for all of them.

I analyzed the “OASIS-D Guidance Manual Effective January 1, 2019” and proposed to provide the dictation option to the demographics section, which mainly consisted of fields that required manual input.

Additional findings: Voice input is 20% faster than manual

I decided to check if making entries with voice is significantly faster than manual input. We ran a quick test using a stopwatch to track the speed and an iOS notes app for dictation.

The testing showed that the voice input feature is almost 20% faster than the manual input. Implementing the VUI could help clinicians save precious time with many fields to fill. Also, there would be no need to glance back and forth between a keyboard and textbox.

Research findings slides (press an image to enlarge)

Brainstorming on HMW design the voice input feature so that users will save time on manual input and keep access to the keyboard

Based on the research insights, there were several things to consider before looking for possible solutions:

Clinicians are super busy, so saved seconds can be a game changer
Clinicians have to have quick access to their tools
Tools and their location on the screen should be intuitive to users
Main tools VUI and keyboard should not compete with each other. Instead, users should have quick access to either of them
VUI for the input fields should provide an error recovery (erase and repeat option or manual editing using a keyboard, for example)

I sketched out different layouts to find a working solution for the voice input toolbar. I learned about the best practices of VUI applied in mobile apps like WhatsApp, Apple messaging, LinkedIn, Otter, etc. All this software locates the main record button on the right side of the screen within a thumb's reach.

Adopting this pattern was the first step in sketching solutions. Since an on-screen keyboard had to be accessible to the users, I decided to keep the open/hide keyboard in the toolbar along with supportive navigation arrow buttons. To provide users with more control over the VUI, I added the erase button that could clear the entry made in the input field. In my sketches, I experimented with various toolbar layouts: a gamer joystick-like, sliding-out toolbar strip, a separate record button for each input field, etc.

At the team meeting, we concluded that making a toolbar as an expanding element would be the best option as it wouldn't distract users and could be hidden/opened anytime for better usability and flexibility of use. So we agreed that I prepare a ready-to-test prototype for this idea.

Brainstorming slides(press an image to enlarge)

Mid-fidelity prototype

Mid-fidelity mockup (collapsed toolbar), active textbox

Mid-fidelity mockup (collapsed toolbar), voice recognition in progress

Mid-fidelity mockup (expanded toolbar), voice recognition in progress

Mid-fidelity mockup (expanded toolbar), voice entry made, erase button appeared

Usability Testing

We moderated testing with 4 participants in person. Given 6 fields to fill, the users had to use voice recognition at first. Then they had to enter the information again manually on the iPad notes. To track the speed of making entries, we used a stopwatch.

Testing showed that on average, users are making entries with speech 33.3% faster than typing. However, there were some improvements to work on: some users thought that the "Erase" button function was to collapse the toolbar. Second insight was that the prototype didn't allow users to switch manually between fields by pressing directly at them.

Mid-fidelity prototype testing results

Mid-fidelity prototype testing model

High-fidelity prototype

While making the prototype, I wanted to improve the flaws I had in the testing prototype. I changed the "Erase" button that some users confused with the "Collapse toolbar." I decided to make it a text button that said "Erase" to avoid confusion. Also, I made available fields to become active by tapping so that the users wouldn't have to use arrow buttons only to navigate the form.

High-fidelity prototype prototype (Figma screenshot)

A technology isn’t a silver bullet, it’s a tool that enhances the achievement of results if used wisely.

While working on this project, our team learned how ignoring the research and definition phase can negatively impact the project. The lesson I learned: is how misleading work could be to the whole project when you agree to jump directly into the solution search. As a designer, my job is to validate the main problem of the project and discover all the support materials that would help me to find a solution that works for businesses and users.