Adam Rotmil
AstraZeneca · Digital health oncology

CARE

From a patient’s words to clinical context.

Discovery & design lead · Mobile & web · 3 monthsThe symptom cycle

Reflect

01Reflect

A moment for yourself.

A weekly reflection begins alongside the updates you’ve already shared.

Recognized
  • Headache
  • Acne
  • Nausea

Reflect, a patient check-in prototype. The recording moves from the reflection history to a guided video recording with the prompt “How are you feeling physically?”, while recognized symptom terms appear: Headache, Acne and Nausea.

Remote monitoring

Between doctor visits, patients in cancer research kept track of their symptoms and treatment side effects every day. I designed the platform for that remote monitoring, from a patient’s own words to the care team’s review.

My role

Product discovery and design lead in AstraZeneca’s digital health oncology division. I interviewed patients, doctors, nurses and research staff, and directed biweekly design sprints, each testing specific hypotheses with users by prototype.

Outcome

CARE shipped as part of AstraZeneca’s clinical trial technology stack alongside Medidata Rave (ePRO/EDC data capture) and Control Tower (internal trial analytics). Built for early-phase oncology dose-escalation studies, CARE enables study teams to capture and analyze higher-granularity patient ePRO data between clinic visits to better evaluate tolerability and dosing decisions.

Why it matters

Patients report symptoms earlier than clinicians record them.

That’s a finding of Dr. Ethan Basch, CARE’s clinical advisor, who led development of the National Cancer Institute’s patient-reported measure of side effects.

Voice symptom tracking

Patients can track their symptoms just by speaking in plain language.

Reflect

Start with the patient’s own words.

We explored LLM extraction of symptom keywords from a short spoken reflection, alongside PPG sensing that reads blood flow through the skin while the patient speaks.

I designed the interaction around it: permissions, recording prompts, time-limit cues, and the choice to continue, save or delete a reflection.

I specified every state around the patient experience of recording symptoms and side effects.

Permission, interruption, clarification, draft recovery and the recipient’s review.

Adam’s annotated Reflect screens: starting a reflection, recording with prompts and recognized terms, the time-limit bar, and the save, draft or delete choices

The symptom journey

From a reflection to the people who follow it.

My user flow for Reflect: record, clarify, share and review, with a way out of every interruption.

Record
Clarify
Share
Review
Reflect home: Good morning, with Start New Reflection and past reflectionsHome
Recording: How are you feeling physically? with Acne, Headache and Nausea recognizedRecording
As the patient speaksHealth words like “nausea” and “tired” are highlighted, and heart rate, oxygen saturation and breathing are read from the video.
Save it?When a recording is short: save to drafts, cancel and delete, or save and continue.
Rate the symptomReflect heard “nausea” and asks for a rating on a validated ePRO scale.
Save and shareTo the feed, with a notification to followers.
A follower reviewsPlays the reflection and its trends, each one explained.
The next appointmentThe patient shows the doctor a highlight reel of recent weeks, to avoid recall bias.
Start reflectionStopToo shortContinueSaveNotifyHighlights

Steps from my original user flow. Screens are frames of the Reflect prototype recording.

Concept testing

Ideas, tested with patients.

Weekly insights, symptom stories and pain-management content. Each concept was tested to learn which presentations felt most useful and least overwhelming.

Home: weekly insights and common symptoms this week
Weekly insights and common symptoms, from others in similar treatment.
Symptom stories: a week of General Pain, told a frame at a time
Symptom stories: a week in the cycle, a frame at a time.
Four iterations of the General Pain screen
Four iterations of the same screen.

Patient-reported outcomes

Make a symptom history understandable.

The design question was which information helps someone understand their experience without overwhelming them.

  1. 1

    The cycle

    Symptoms read against the patient’s own treatment cycle.

  2. 2

    The pattern

    Each report on a five-step severity scale, against the last cycle.

  3. 3

    The next step

    A plain language insight, and a direct line to the care team.

Rebuilt in code from my original design. Names, dates and readings are prototype data.

General Pain

Current Cycle 7 (Mar 17-31)

Pain has been increasing

Your pain is the most severe it has
been in this cycle. Typically, your pain
is mild towards the end of your cycle.

If you haven’t already, we recommend
calling your health care provider to
help manage your pain.

Contact My Care Team

Your pain over the current cycle

Outcome

A specified workflow across patient and clinician.

The flows and annotated states gave the team one account of how a patient update moves from recording to clinical review, including when someone needs to pause or change course.

Credits

Role
Product discovery, design leadership, interaction design, user research, design sprints
Clinical advisor
Dr. Ethan Basch
Worked with
Clinical research team, patients and oncologists, engineering, product management
Research
Interviews with patients, doctors, nurses and research staff
Organization
AstraZeneca, digital health oncology R&D
Platform
Mobile and web, over three months
Tools
Figma, FigJam, Miro, Sketch

Screens and recordings are my original prototypes. Names, readings and figures inside them are prototype content, not patient data or results. The evidence figures are published research on ePRO symptom monitoring, not outcomes of CARE.

Research that informed our approach

Basch et al., 2009
Adverse symptom event reporting by patients vs clinicians: relationships with clinical outcomes. JNCI 101:1624–1632.
Basch et al., 2014
Development of the NCI’s patient-reported outcomes version of the CTCAE (PRO-CTCAE). JNCI 106:dju244.
Basch et al., 2016
Symptom monitoring with patient-reported outcomes during routine cancer treatment: a randomized controlled trial. JCO 34:557–565.
Basch et al., 2017
Overall survival results of a trial assessing patient-reported outcomes for symptom monitoring during routine cancer treatment. JAMA 318:197–198.
NextAI Model TrainingRating model responses for frontier AI labs