— TEAM SIZE
3 Designers
1 Product Manager
3 Developers
1ML Engineer
CTO/Co-Founder
— MY ROLE
Product Design
User Research
Branding & Visual Design
Design System
— DATE
2020-2021
I helped design an AI-driven software application for breast imaging.
The wearable scanner automatically captures the entire breast volume allowing for repeatable breast imaging at point of care.
Product goal:
Help radiologists improve breast cancer detection and risk assessment
Target users:
● Radiologist
● Trained operators
- ● Medical assistants
Software application and the wearable hardware
Design &
Research
Research Goals:
- ● Understand clinician's pain points using current imaging softwares and how that affects their workflow
- ● Learn most desirable features and functionalities
- ● To learn about how users leverage image viewer
We assessed existing software solutions by comparing their features, functionalities, and value to end users.
User Interviews
- We started our user research by interviewing five radiologists from multiple services such as critical care and breast cancer.
We drafted several questions focused on:
- ● Image viewing applications in use and who uses them
- ● Workflow
- ● Pain points
- ● Features; what’s used most and what they’re used for
- ● Image viewers and patient outcomes
Kano Survey
- Our initial focus was to understand what features are most important for testing and planning. The purpose of this survey was to evaluate X features being considered for inclusion in the MVP of the iSono software to pinpoint the most desired and needed capabilities.
How does Kano work?
- Users are asked a pair of questions for each potential feature/capability:
Functional state: If you can [describe feature here] – how do you feel?
Dysfunctional state: If you cannot [describe feature here] – how do you feel?
Research Synthesis
- These are the following methods we used to synthesize the raw information we had and arrive at our hypothesis.
- ● Discovery in Dovetail
- ● Affinity map in Miro
Dovetail
It helped us formulate our observation and outline what we learned. Also helped synthesize the questions we asked users and determining if there are repeatable tags that we can create in Dovetail and see where there are commonalities and where the differences are.
We learned a lot from what we knew before and this helped us to see if there are further knowledge gaps and drive home some of the findings with the observations.
Affinity Mapping
- To understand the needs, behaviors, and motivations of users we collaborated on creating affinity map in Miro where we were able to record all notes, survey results and observations on individual cards and look for common patterns.
We synthesized our learnings to share out with the team and stakeholders.
Some of the key learnings:
Core functionality was very important to the users
- Annotations, cuts/measurements, print screen, timeline, and presets were mentioned specifically. Everything else was “bells and whistles” that were nice, but unnecessary
Generally medical experts are approaching AI with caution
The hope is that AI becomes a tool to help radiologists and their patients. Generally
Loading times and software crashes are affecting patient care, troubling clinicians
Speed is a big thing. one of the greatest frustrations physicians have is delays of the EMR or any, any version of an electronic system
- Another pain point was the lack of tool support, education and training
- Clinicians needed additional preparation and training with every software update
- Cloud integration was desirable
- It would allow clinicians to access the software and their data from anywhere, and enable them to access to images for remote tracking and remote care
Intuitive user experience that would support their workflow was important
- Users became frustrated when software was difficult to use and make it hard to achieve their goals
LOW-FI DESIGN
Design Process
- Define the Problem: We began by clearly understanding the problem we aimed to solve and the objectives we wanted to achieve.
- Generate Ideas: With a solid understanding of the problem, we brainstormed potential solutions to address the needs and goals.
- Evaluate and Refine Ideas: We assessed each idea to identify the most promising solutions and refined them based on feasibility and impact.
- Test and Iterate: We tested our design and iterated based on user feedback, ensuring the solution aligned with user needs.
I collaborated closely with another UX designer to establish the framework and create initial wireframes. Given the complexity of the tool, which involved multiple workflows for different user types, I worked alongside the product manager to prioritize design tasks. We used insights from user studies to inform and write user stories.
User Flow
Low-Fidelity Wireframes
HI-FI DEDISGN
Device Setup
Account Creation / Sign In
Patient Management Portal
Pre-Scan Flow
Scan Flow
Post Scan / Labeling And Annotation
Report Creation
Design System
Onboarding was slow due to a lack of consistency in the design process. To address this, I created a style guide and collaborated with the team to develop a reusable UI component library. This streamlined the design process and improved implementation with engineering.
The design system saved time, unified the product, enabled faster iterations, and improved usability by providing a consistent, collaborative framework for design, development, and QA teams.
Color Palette / Typography
- I gathered the learnings from Market/competitor analysis and prior user research data, to initiate color palette and typography explorations. The goal was to establish an authentic brand look and feel that's familiar to the users, but at the same time looks clean, modern and represent a cutting edge technology. I created several explorations and worked with the product manager to get user feedback (ultrasound technicians) and see how they react to the overall look and feel.
