Anna Kim Porfolio

DURATION

16 weeks

UI/UX Design for Web/App

Product Management

The World’s Leading Causes of Death

in 2022 in million

CARDIOS was a client project in collaboration with The M.E.N.D. BioSimulator, a medical diagnostics platform founded by Professor Lee from Yonsei University and his lab researchers with the mission of reducing patient medical expenses. This project was also invited and presented at CES 2024 Las Vegas.

As global attention to cardiovascular diseases increases, the related global market size is also experiencing significant growth. The global market for cardiovascular disease-related software is projected to be around 900 million dollars by 2029, and the market related to FFR measurement is anticipated to reach 1.4 billion dollars, with the expected annual growth rates of 7.3% and 12.0%, respectively.

The traditional method of diagnosing cardiovascular disease is significantly more expensive and time-consuming for patients. After an initial examination and CT scan, patients often need to return to the hospital for an invasive FFR procedure. This process typically takes around six hours and requires multiple hospital visits. In contrast, CARDIOS streamlines this entire workflow - using a CT image obtained during a routine health check-up, the software performs automatic image segmentation in under three minutes. Once additional patient information is entered, an AI-driven FFR prediction is completed in less than one minute.

This non-invasive, rapid approach enables efficient patient screening and helps prevent unnecessary procedures and overtreatment.

Segmentation system and method of an ascending aorta and a coronary artery from CCTA using a hybrid approach
(KR: 10-2022-0027282; US: 17/820,514; CN: 202210811389.2)

Cardiovascular disease risk analysis system and method considering sleep apnea factors
(KR: 10-2451624; US: 17/532,167)

Optimization system and method of AI algorithm for prediction coronary artery lesions based on FFR
(KR: 10-2022-0030019; US: 17/820,819)

“Impact of coronary lesion geometry on fractional flow reserve data from interventional cardiology research in-cooperation society—fractional flow reserve and intravascular ultrasound registry”,
Circulation-Cardiovascular Imaging, 2018.06

"Coronary artery decision algorithm trained by two-step machine learning algorithm",
RSC ADVANCES, 2020.01

“Optimization of FFR Prediction Algorithm for Gray Zone by Hemodynamic Features with Synthetic Model and Biometric Data",
Computer Methods and Programs in Biomedicine, 2022.05

When users input cardiac CT images and patient biometric data, CARDIOS generates both a detailed 3D vascular structure and an FFR value to assess cardiovascular disease risk. This allows doctors to obtain accurate diagnostic predictions using familiar types of clinical input, streamlining their workflow while enhancing decision-making.

CARDIOS has been recognized not only in the medical community but also on global stages. The project was invited to present at CES 2024 in Las Vegas, where it drew attention as an innovative AI-powered diagnostic tool for cardiovascular health.

Its technological excellence has also been validated through multiple accolades, including the Gold Medal and the Seoul Mayor’s Award at the Seoul International Invention Fair, as well as invitations to showcase at major healthcare and technology exhibitions such as the Dong-A Health Industry Fair and the Korea Science & Technology Exhibition.

Contributions: I led the product management for the project, directing the end-to-end development of the user journey and designing the UI/UX for both the app and website. I collaborated with a software engineer on the program development, ensuring the technical functionality aligned with the project goals. Additionally, I sought guidance and consulted with doctors during the planning phase to ensure the content and features of the application were medically accurate and relevant to the needs of healthcare professionals.

Fully functional Mobile App

Responsive Website

Software Engineer

ROLES

DELIVERABLES

TEAM

OVERVIEW

AI-based automated diagnostic software to predict cardiovascular disease risk designed for heart surgeon doctors and hospitals.

DESIGN PROCESS

RESEARCH

Target Market Size

According to the WHO, cardiovascular diseases are the number one cause of death globally. Approximately 8.9M people have died due to this disease.

Diagnostic Scenario

Input & Output

Patents and Papers

Invited & Presented at

CES 2024 Las Vegas

Research

Ideation

Design Decisions

Wireframes

User Testing

Final Prototype

Neonatal conditions

2.0

Stroke

6.2

Trachea, bronchus,

lung cancers

1.8

Chronic obstructive

pulmonary disease

SOURCE: WORLD HEALTH ORGANIZATION

3.2

Alzheimer’s disease

and other dementias

1.6

Lower respiratory

infections

2.6

Diarrhoeal diseases

1.5

8.9

Cardiovascular

disease

Gobal Cardiovascular

Software Market

Gobal FFR

Measurement Market

Decision making

for Invasive FFR

Automated Segmentation

AI-based FFR Prediction

Decision making

for Invasive FFR

Traditional

Invasive FFR

Measurement

Decision making

for PCI

PCI Treatment

2029

2021

(million USD)

Hospital revisits

required for

further diagnosis

~3 min

~1 min

KRW 800,000

for coronary angiography

KRW 1,500,000

for FFR Measurement

(Percutaneous Coronary

Intervention, PCI)

KRW 2,000,000

Examination

CT Scanning

KRW 200,000

Complex & Expensive

Rapid & Cost-efficient

Cardiac CT Imaging

DICOM format

Patient Biometric Featuress

Sex (M/W), Age (yr),

Height (cm), Weight (kg),

Systolic/Diastolic BP (mmHg),

Hemoglobin (mg/dL)

3D Vascular Structure

STL format

Ascending aorta & Coronary artery

Risk Prediction: FFR Value

AI-based FFR prediction

on specific lesion

Decision making

threshold (FFR 0.8)

CARDIOS

AI-based automated diagnostic device for

predicting cardiovascular disease risk

CARDIOVASCULAR DISEASE DIAGNOSIS APP / INDUSTRY PROJECT

The structure and interaction flow of the CARDIOS app were shaped through multiple design feedback sessions with the stakeholder Professor Junsang Lee—both a Yonsei University mechanical engineering professor and founder of The M.E.N.D. BioSimulator—whose close collaboration with hospital cardiologists gave him unique insight into their needs.

These meetings highlighted the urgency doctors face during diagnosis, their preference for minimal manual input, and the need for system feedback they can trust. Based on these insights, I designed a vertically aligned, step-by-step flow—starting with Data Load, Auto Segmentation, Biometric Input, and ending with FFR Prediction—to mirror the streamlined clinical workflow.

After login, the Home tab guides doctors through the core diagnostic flow—Data Load → Auto Segmentation → Patient Biometric Input → FFR Prediction—with each step presented sequentially to minimize cognitive load. The My tab houses Patient Data History and Account Information, keeping administrative tasks separate from clinical workflows. This clear division ensures fast, focused access to both diagnostic and patient-management functions.

Based on the information architecture, I created a series of low-fi sketches to map out the user journey. I also drafted the core Home screen with a top-down flow of Data Load, Auto Segmentation, and FFR Prediction buttons.

After sketching the initial low-fidelity prototype of the Home screen, I carried out in-depth usability interviews with five practicing cardiologists by testing out my low-fidelity prototype. The goal was to validate whether each interactive element aligned with their real-world workflow—and it quickly became clear that the Auto Segmentation button was a major pain point.

To refine the final Home screen for CARDIOS, I conducted a targeted competitive analysis across both medical and general-purpose apps that handle complex, multi-step processes. Key references included OsiriX HD and HeartFlow for clinical workflows, as well as Apple’s Shortcuts app for its elegant, state-driven button interactions.

From HeartFlow’s stage-driven analytics and Apple Shortcuts’ action-card interactions into tappable cards (Data Load → Auto Segmentation → FFR Prediction) that animate through Idle → Loading → Complete (gradient sweep, spinner, checkmark) and an in-app severity slider with “Save to History.” The result brings the analysis into a single iOS workflow that clearly shows progress, reduces context switches, and speeds clinical decision-making.

Working on CARDIOS taught me how to balance the complexity of AI-driven healthcare technology with the clarity doctors need in high-pressure environments. As both Product Manager and UX Designer, I realized that I could not assume clinical needs—I had to conduct interviews and usability tests with doctors to uncover pain points and validate design decisions.

Through this process, I translated advanced algorithms into step-based interactions that aligned with real diagnostic workflows, while adding feedback mechanisms like loading indicators to build trust in AI outputs. This experience reinforced that in healthcare, good design is not only about usability but also about responsibility, since every interaction can directly impact confidence and patient outcomes.

After receiving feedback, I prioritized adding real-time progress indicators and a clear “Segmentation Complete” confirmation in the next prototype iteration—ensuring doctors always know when the complex backend process is running, how far along it is, and when it’s finished.

In CARDIOS, Auto Segmentation refers to the process of automatically extracting and visualizing the 3D structure of the coronary artery and aorta from a patient’s CT scan. Although this involves complex backend computation, the goal was to design a seamless button-triggered action that allows doctors to initiate segmentation with ease. By providing immediate visual feedback and minimizing interaction steps, the design ensures that users can trust the system’s performance without needing to understand the underlying technical complexity.

To address this question, I established three key goals to guide the entire interface design. These principles were prioritized throughout the design process to support doctors in making fast, informed decisions under time pressure. Each design decision was evaluated against these goals to ensure an intuitive experience across both the app and web platforms.

DESIGN DECISIONS

APPROACH

ITERATED DESIGN DECISION

DESIGN CONCEPT

FINAL DESIGN

REFLECTION

UX RESEARCH

WIREFRAMES

DESIGN GOAL

Expert Insights &

Clinical Workflow Alignment

Challenge: How can we simplify a complex backend process into a single, intuitive action for doctors to initiate CT image segmentation?

Auto Segmentation button Redesign

Visual Identity

Hi-fi Prototype

Key Takeaways

Actual Implementation Screen (Android)

User Testing

Competitive Analysis

User Flow Patterns

Information Architecture

Lo-fi Prototype

How might we design a diagnostic interface that helps doctors make decisions

in high-pressure environments?

Reduce cognitive and interaction load to help users complete tasks quickly.

Efficienct

Design an interface that enables quick comprehension and efficient use by doctors, who often work under time pressure.

Clear

Build trust in AI-generated results by making the diagnostic process and output transparent and interpretable.

Trustworthy

Create Account / Login

Home

Data Load

Auto Segmentation

Patient Biometric Info Input

FFR Prediction

Patient Data History

Account Information

User flow

GIF

“When I tap ‘Auto Segmentation,’ nothing on the screen changes—I have no idea if the system even received my command.”

“It’s frustrating to just stare at a blank screen; I need some kind of progress indicator so I know how long the segmentation will take.”

“After waiting, there’s still no confirmation that segmentation is done—I find myself repeatedly tapping the button or switching screens to check if it worked.”

OsiriX

HeartFlow

Shortcuts

Instead of OsiriX’s static tool icons, CARDIOS introduces dynamic “Data Load” and “Auto Segmentation” cards that visibly transition through Idle, Loading, and Complete states. This clear, animated feedback eliminates uncertainty and reduces unnecessary retries.

OsiriX

Apple’s Shortcuts app surfaces task progress with action cards that move from Idle → In Progress → Complete using subtle animations and a checkmark.
I applied this pattern to CARDIOS—each Data Load/Segmentation/FFR card uses a lightweight animated sweep, spinner, and final checkmark to show clear progress without clutter.

Shortcuts

HeartFlow’s labeled steps inspired CARDIOS’s tappable cards

(Data Load → Auto Segmentation → FFR Prediction) that visually transition Idle → Loading → Complete so doctors get instant status at a glance.

HeartFlow requires exporting a PDF report; CARDIOS instead offers an on-screen severity slider with the FFR value and a built-in “Save to History” button, also including the segmented 3D model all surfaced on the home screen, streamlining documentation without switching contexts.

HeartFlow