Asset A

Design Competition Winner

B2B Saas

Healthcare

AI Service Design

Human-in-the-loop AI for ethical SLP treatment

B2B SaaS designed to facilitate AI transparency in patient-clinician communication and AI-powered diagnostic procedure.

My Role

Product Designer

Timeline

6 weeks (Spring 2024)

team

1 Clinician 3 Designers

Tools

Figma AfterEffects Qualitative Research

Design Competition Winner

B2B Saas

Healthcare

AI Service Design

Human-in-the-loop AI for ethical SLP treatment

B2B SaaS designed to facilitate AI transparency in patient-clinician communication and AI-powered diagnostic procedure.

My Role

Product Designer

Timeline

6 weeks (Spring 2024)

team

1 Clinician 3 Designers

Tools

Figma AfterEffects Qualitative Research

Overview
Background

As a product designer on the UW HCDE team, we survived a 6-month (2025) competition held by a global Service Design Council with IBM. We built an AI system that is feasible, safe, and human agency focused for Speech-language pathologists. We focused on facilitating communication between clinicians and patients and increasing AI transparency for clinicians’ diagnosis and treatment.

Impact

Impact

2nd Prize 🏆

We won second place globally in a prestigious service design competition

AI with Human Agency

This project was a huge learning for me, building a product that is ethical and safe.

solution

Echo, where every Voice is shared to empower all

We built an AI empowered Saas software called Echo. Echo is a digital platform that facilitates transparent, ethical communication between speech-language pathologists and their clients around the use of AI.

View our final project

01/ Main Features
Feature 1- Echo View

Context-rich intake process to mitigate bias

Conventional intake process omits information information for patients with multilingual background. With Echo, intake process becomes more seamless, transparent, and clear for both clinicians and patients.

FEATURE 2- Research Echo Board

Bridging clinical bias and potential participants

Beyond diagnostics, it helps patients discover relevant research opportunities, empowering both patients and AI developers to co-create more inclusive AI systems.

FEATURE 3-Clinicians Insight Network

Healthcare AI needs guardrails, shaped by clinicians

Echo supports clinicians with a peer-driven network. Designed to address the biggest challenge, this community empowers clinicians to take greater agency in how they use AI, while fostering transparent, real-world conversations.

02/ Problem Space
Mission

1

Empower society to overcome bias

2

Create a people-centric, technology enabled, and business viable in the team's local context

Context setting

Finding where the healthcare bias is coming from

Based on our mission, we spent 4 months in total for researching a specific problem domain, interviewing diverse healthcare and AI experts.

Problem Scope

AI usage in Speech Language Pathology

Through interviews with a Speech-Language Pathology (SLP) clinician at the UW Speech & Hearing Clinic, we discovered that while AI tools are already being used in clinical practice, there’s a lack of guidance around their ethical and responsible use.

Why it matters

SLP is a field where bias can significantly affect especially when treating patients with diverse languages, dialects, and cultural background. Yet current AI systems often fall short.

Why It’s Impactful

Large language models (LLMs) are increasingly entering the SLP space. A 2025 study in Frontiers in Digital Health found that most audiologists and SLPs see AI as beneficial for diagnosis and treatment, highlighting the urgent need for clinician-informed guardrails.

Identifying Key Issues

AI usage in Speech Language Pathology

Through interviews with a Speech-Language Pathology (SLP) clinician at the UW Speech & Hearing Clinic, we discovered that while AI tools are already being used in clinical practice, there’s a lack of guidance around their ethical and responsible use.

“Even we don’t fully trust standardized tests.

When we’re relying on an AI-based report, clinicians need to know its limits.”

-Clinician 1

Part of the In-depth Interview with UW SLP Clinicians (n=2)

“[I want to know] how it’s being used in my care- is it diagnosis or treatment etc. There should be full transparency so that clinicians can build trust with us.”

-Patient 2

Perception Survey on AI Usage in Healthcare, 2025 (n=25)

“When we compare [clinical scores] of bilingual kids to monolingual kids using [an AI transcription service], of course they score lower—but that’s not because they have a disorder.”

-Clinician 2

Part of the In-depth Interview with UW SLP Clinicians (n=2)

“If a doctor is using AI, I should be able to request a full report of the AI suggested diagnoses.”

-Patient 3

Perception Survey on AI Usage in Healthcare, 2025 (n=25)

clinicians' painpoints

1

1

1

Lack of depth on intake forms

Artists need risk management

2

1

1

Time constraints

Artists need risk management

3

1

1

Ethical uncertainty around AI usage

Artists need risk management

4

1

1

Uncertainty regarding limitations of AI

Artists need risk management

Patients' painpoints

1

1

1

Unclear integration of AI in their care

Artists need risk management

2

1

1

Low AI literacy & trust

Artists need risk management

3

1

1

No validated channels to resolve concerns

Artists need risk management

How might we
Create a single source of transparency
that helps patients communicate their concerns clearly,
and supports clinicians in using AI ethically and efficiently

How might we
better connect people who use drugs with local harm reduction service(HRS)

Problem Statement
03/ Wireframes & Feedback
Key Touchpoints

Critical: Before & during the initial visit

We chose before visiting & initial clinic visit as main touch points. These initial moments are critical in patient and clinician relationship because this is where all the contextual data is heavily collected, which affects the first diagnosis, having a lasting ripple effect.

User testing

Learning from lived experiences

With our wireframe, we conducted UI evaluation session with SLP clinicians to identify missing points.

Takeaways

1

Real experience matters.

Healthcare is complex, but nothing exists beyond real-world insights. Talking directly with clinicians and researchers helped us design a meaningful solution for SLP, even as we entered the domain as newcomers.

1

Real experience matters.

2

Zooming in and out

In the beginning, our focus was improving clinical research recruitment. But as we zoomed in on the SLP, we realized that this clinical research gap was relevant. This insight led to the creation of the Research Echo Board, zooming out again, to connect underrepresented voices with research opportunities.

2

Zooming in and out

3

Balancing Revenue and Safety

Building an AI tool for healthcare meant balancing safety with financial value. We carefully gathered insights from diverse stakeholders to ensure the feasibility of our solution.

3

Balancing Revenue and Safety