In this project, I aimed to solve the problem of high cognitive burden during symptom reporting, Research revealed a recurring gap between experiencing a symptom and knowing what to do with that sensation. Leading to unreported severe side effects and delayed clinical interventions.
Core HMW : How might we reduce the communication burden between oncology patients and their care teams without replacing clinical judgment?
The Research & Key Discovery
We conducted social listening across oncology patient communities, analyzed local clinical practices, and evaluated portal usage patterns to pinpoint why communication breaks down.
Three Core Research Insights:
The Context Gap: Patients report symptoms as raw feelings ("I feel freezing"), whereas oncologists require structured metrics (temperature, duration, neutropenia status).
Input Friction: Typing detailed reports during active chemotherapy is a major physical barrier.
Communication Anxiety: Patients feel uncertain after submitting reports, wondering if their doctor actually reviewed their status or if they should call urgent care.
Key Takeaway: Patients needed a low-effort way to speak naturally, while doctors needed structured, risk-stratified data. This informed our decision to build a voice-first intake paired with a complementary AI context engine.
The Design Action & Artifacts
We used FigJam and interactive prototypes (Base44) to design a voice-first intake system paired with a complementary AI probing engine.
Core Interaction Touchpoints:
Voice-First Intake: Patients speak freely in natural language without worrying about clinical terms.
Complementary AI Probing: The AI evaluates raw audio and asks a maximum of 2 targeted follow-up questions to fill critical clinical gaps.
Triage-Ready SBAR Handoff: Generates a structured summary (Situation, Background, Assessment, Recommendation) for the oncology care team.
Usability Testing & Key Design Iterations
Based on an initial usability testing, and expert feedback (including Eric Shumake's critique), several key iterations were made to lower friction and improve safety:
Iteration 1: Smarter Input Flexibility (Auto-Fill & One-Tap Controls)
Problem: Even with voice intake, answering multi-step follow-up questions caused fatigue.
Design Pivot: Introduced Auto-fill smart options and one-tap chips for common clinical markers (e.g., fever levels, onset duration), giving patients an immediate alternative to speaking or typing.
Iteration 2: Deepening Patient-Specific Cancer Context
Problem: Early drafts treated symptom complaints generically, missing critical cancer-specific nuances.
Design Pivot: Integrated patient-related cancer context directly into the AI engine — factoring in current chemotherapy cycles, medication regimens, and pre-existing side effect baselines to ask smarter, highly relevant questions.
Iteration 3: In-Panel Report Editing & Fluid Navigation
Problem: Users felt trapped if the AI misheard a detail and had to restart the whole intake.
Design Pivot: Redesigned the navigation architecture to allow updating and editing the generated report directly from inside the symptom panel before sending.
The Human Impact: From Uncertainty to Safety
Behind every SBAR report is a patient navigating therapy at home. When a sudden symptom strikes, traditional portal forms force patients into manual typing and complex terms leading to anxiety, hesitation, and unreported side effects
With OnCure, that moment changes:
The Patient: Speaks or taps an auto-fill chip in under a minute, hears her report verified out loud by the AI reader, and sees it delivered directly into her care team chat.
The Clinician: Receives a structured, triage-ready summary instantly—eliminating chaotic text blocks.
That transition from anxious isolation between chemo cycles to feeling heard, supported, and protected is OnCure’s true metric of success.
This project reflects a continuous cycle of evidence-based research, clinical alignment, and empathetic iteration. I invite you to test the experience, inspect the research, or share feedback:
Launch Interactive Live Demo : Test the voice-first intake, auto-fill chips, and care team chat layer in real time. Live Demo https://oncure.base44.app
Check FigJam UXR Workspace : Inspect social listening notes, 5-participant usability logs, and full iteration mapping
Connect on LinkedIn : Always eager to collaborate on high-stakes healthcare AI products that reduce cognitive load. LinkedIn Profile
