VAST: Usable & Accessible Clinical App Redesign

I led the redesign of VAST, a voice ai swallow test app designed for clinicians and patients, to incorporate AI features and increase the usability and accessibility of the mobile app.

Company

Weill Cornell Medicine

Team

Product Designer

PM (Dr. Rameau, MD), 3 Dev engineers (Alex, Pantelis, Jeff), 2 ML researchers (Chen, John)

Role

Time

Apr 2025 - Present

Unclear flows and inaccessible UI blocked clinicians from finding patients, tracking visits, or viewing AI insights—leading to delays and errors.

The challenge

Specific goals

1. Enable Fast Patient Lookup
2. Ensure Complete Session Tracking
3. Surface AI Diagnostic Insights
4. Improve UI Usability & Accessibility

1. Patient‑Search Time (Target: ≤ 10 s)
2. Session‑Capture Completion (Target: ≥ 95%)
3. Accessibility Compliance (Target: 100%)
4. Clinician Satisfaction (Target: ≥ 80 (SUS) / ≥ 4 out of 5)

Key metrics

Approach

Usability & Accessibility Audit

To identify the problems within the current screens, I audited current flows, friction points, accessibility and usability of the app and reported to the team.

I conducted interviews with 3 clinicians and researchers to map workflow challenges and confusions. The main problems I identified include:

Difficulty to look up patients or return to the most recent data entry
Unable to record multiple sessions for a patient
Unclear states & system indications
Difficult navigation and hand interactions

User Interviews

Co-design workshops

I gathered PM, engineers, and clinician users together in 5 workshops to sketch and test new flows and ideas.

Based on our feedback from devs and users, I modified interactions and flows to reduce steps and incorporate identified new needs.

Redesigned Patients Hub

The enroll and data collection now can be conducted in one page without having to go back and forth. And the newly added filter features help clinicians locate a patient faster than before (within 10s).

Patient Profile

The new feature helps clinicians manage multiple entry of session data, as well as providing easy access to AI generated risk prediction, hitting our goal Session‑Capture Completion > 95%.

Redesigned Tasks

The new tasks screen allows selective display of tasks to help clinicians find tasks easily. It also allows marking the status of the session for data management. Along with the reorganized visual hierarchy, the time on locating a task decreased by 30%.

Redesigned Task Interfaces

The specific task screens clinicians perform with patients have also been redesigned with better accessibility and readability. With options to increase font sizes as well as better navigation of progress, the accessibility and readability increased by 68% and 34%.