Lab Insights AI · Product Design

Every patient is different. Now the app knows that.

Cancer treatment is overwhelming. Blood tests come back full of confusing numbers. Patients go home not knowing what any of it means, what to eat, how much to drink, or why they feel the way they do. We built a feature that changes that — turning lab reports into clear, personal guidance that helps patients prepare for every chemotherapy session.

My Role
Lead Product UX Designer
Platform
Everhope iOS & Android App
Timeline
14 Weeks · 2024
Team
Oncology · AI/ML · Product · Dev
78%
Patients understood their blood report — up from just 17%
−41%
Fewer chemo sessions postponed due to low blood counts
3.2×
More patients following diet and hydration guidance daily
91%
Patients and families said the feature was genuinely helpful
The Bigger Picture

Cancer care in India leaves patients alone between appointments.

India sees over 1.4 million new cancer cases every year. Most patients are diagnosed late — at Stage III or IV, when treatment is at its hardest. They travel long distances to reach a specialist. They sit through chemotherapy. And then they go home with a printed lab report full of medical numbers — and no one to help them make sense of it.

The gap between appointments is where patients struggle most. No guidance. No monitoring. No one to tell them what's low in their blood, what to eat to fix it, or when a symptom is worth calling about. This is the gap we were asked to design for.

1.4M+
New cancer cases per year in India, growing every year
71%
Diagnosed at a late stage, when treatment is most intensive and side effects are hardest
1 in 4
Chemo sessions get delayed because blood counts are too low — many of these are preventable
83%
Patients couldn't explain a single value from their lab report, even after a consultation that same day
The Problem

The three weeks between sessions are a complete black hole.

Patients go home after a chemo session with a blood report, a lot of anxiety, and almost no guidance on what to do. Here's what we found was actually happening.

📋

Reports are full of words nobody understands

CBC, ANC, ALT, creatinine — the report is a wall of medical terms and numbers. Doctors try to explain, but consultations are short. Most patients forget what was said within a day. Nothing is written down in plain language they can actually use.

🧬

No two patients react to treatment the same way

One patient's haemoglobin crashes in Cycle 3. Another's neutrophils drop in Cycle 2. A third develops fatigue from low iron despite normal haemoglobin. The symptoms are different. The causes are different. Generic advice like "eat well and rest" helps nobody.

🥗

"Eat nutritiously" is not good enough

If your haemoglobin is low, eating spinach with lemon at meals makes a measurable difference. If your neutrophils are low, there are specific foods to eat, hygiene steps to take, and things to avoid. But patients are sent home with no food-specific, number-specific guidance. They're left to guess.

⏱️

Cancer treatment leaves no energy left for anything else

By the third cycle, patients are exhausted. Managing nausea, fatigue, medications, and appointments is already too much. Remembering to drink enough water or track what they eat gets dropped entirely. Not because they don't care — because there's simply nothing left.

What we kept hearing in interviews: "The doctor tells me something is low and says eat better. I go home and I don't know if low means dangerous or just low. I cook the same things and hope the next test is better."

The Question We Designed To Answer
How do we help a cancer patient understand their specific blood report in plain language, know exactly which values need attention and why, and get daily, practical guidance on what to eat, drink, and do — so they show up to every session as prepared as possible?
Research

We spent three weeks listening before we designed anything.

18 interviews with patients, family members, and oncology nurses at Everhope's centres in Gurugram. We sat in consultations, read 200+ real lab reports, and ran three rounds of testing with actual patients. Here's what we found.

83%
Patients could not explain any value from their lab report — even right after a consultation
67%
Family members wanted to help but had no idea what to do with the report or between appointments
1 in 4
Chemo sessions in our report audit were delayed — low neutrophils and low platelets the most common reasons
91%
Patients said they'd use an app feature that told them exactly what to eat or drink to improve a specific low value
01

Symptoms happen but patients don't know which ones to report

Fatigue, mouth sores, unusual bruising — patients write these off as "normal side effects." But each symptom is often a sign of a specific low value in their blood. Connecting the symptom to the number, in plain language, was the missing link. Once patients understood why they felt what they felt, they were far more likely to act on guidance.

02

In Indian families, it's the family member who does everything

In almost every home we visited, the patient didn't cook their own meals or manage their own medications — a spouse, son, or daughter did. If the guidance was only useful to the patient, it would mostly go unused. The family member needed to understand it too, get it on their phone, and know what to cook for dinner.

03

WhatsApp is currently the best support system patients have — and it's not enough

Patients photograph their report and send it to anyone they know with a medical background. They get answers that are incomplete, alarming, or just wrong. The anxiety of not knowing is often worse than a bad result. People needed a source they could actually trust — not a panicked group chat at midnight.

04

Simpler than simple — because users are often unwell when they open this

Most of our users are 45–70 years old, have moderate smartphone experience, and are opening this app while feeling sick or scared. A feature with more than 3 taps to reach an answer, or that used medical jargon without explaining it, would simply not be used. Radical simplicity wasn't a design preference — it was the only viable approach.

Who We Were Designing For
👩
Meera, 52 · Gurgaon
Breast cancer patient · 3rd of 6 chemo sessions
  • Gets her blood report the night before chemo and has no idea if her numbers are safe
  • Feels very tired in this cycle — doesn't know it's directly linked to her low haemoglobin
  • Has been told to "eat well" but doesn't know that spinach with lemon could specifically help her
  • Has sent her report to 4 different relatives on WhatsApp looking for a clear answer
  • Biggest fear: arriving at the chemo centre and being told her session is postponed again
👨
Rohan, 34 · Gurugram
Son and caregiver — managing his father's Stage III lung cancer
  • Manages all doctor visits, medicines, and home meals alongside a full-time job
  • His father's neutrophil count dropped dangerously in Cycle 2 — nobody explained what that meant
  • Googles lab values at midnight, finds alarming results, and ends up more anxious than before
  • Wants to feel useful and in control; right now feels completely lost between appointments
  • Would trust guidance from Everhope's platform far more than anything he finds on Google
The Core Insight

Each patient's body responds to treatment differently. The guidance has to match.

The single biggest flaw in generic health apps is that they treat all cancer patients the same. They don't. Different blood values drop at different times for different people on different treatments. Our AI reads each patient's specific numbers and gives guidance that's relevant to them — not to some average patient.

🩸

Low Haemoglobin → Fatigue, breathlessness, dizziness

Common during chemo. Patients often blame tiredness on "the treatment" without knowing it's tied to a specific number they can actually improve. Eating iron-rich foods with vitamin C, avoiding tea at meals, and resting properly makes a measurable difference in the days before the next session.

Diet + Rest guidance
🛡️

Low Neutrophils → Infections, mouth sores, fever risk

This is the most common reason chemo sessions get delayed in India. Low neutrophils mean the body can't fight infections well. Most patients don't know this — they continue eating raw food and going to crowded places. A specific checklist of hygiene steps, food to eat, and things to avoid changes the outcome.

Hygiene + Protection guidance
💊

Low Platelets → Unusual bruising, slow healing, bleeding gums

Patients often don't feel low platelets until something unexpected happens. They need to know to avoid certain medications, watch for bruising, and be careful with physical activity. This guidance is specific, time-sensitive, and almost never communicated at discharge.

Activity + Medication awareness
💧

Dehydration → Makes nausea worse, adds to fatigue, stresses kidneys

Chemo makes drinking feel difficult. Nausea, fatigue, and the general exhaustion of treatment mean most patients drink far less than they should. Dehydration then amplifies every other side effect. A simple daily water tracker with push reminders is low-tech and genuinely changes behaviour.

Hydration tracking + Daily reminders
Design Principles

Before any screens — the rules we agreed to follow.

The best design decisions get made before opening a design tool. These six principles were our shared agreement throughout the project. Every screen was measured against them.

01

Calm, not clinical

A patient opening their lab report is already anxious. Every screen should lower that anxiety, not raise it. No medical term appears without an immediate plain-language explanation. Colour signals are used to inform, not to frighten. Red is reserved only for values that truly need urgent action.

02

Explain first, then show the numbers

Never lead with data before the patient understands what it means. The AI summary always appears before the list of values. When people read context first, they understand the numbers better and panic less. This one change was the biggest improvement we made in testing.

03

Specific enough to act on today

"Eat nutritiously" helps nobody. "Have spinach with a squeeze of lemon at lunch — vitamin C helps your body absorb iron" is something a person can actually do today. Every piece of guidance must be concrete, Indian-household-friendly, and appropriate for someone on active cancer treatment.

04

Designed for the family, not just the patient

In most Indian homes, it's a family member who cooks, manages medicines, and keeps track of everything. If this person can't access and understand the guidance, nothing gets implemented. Every screen must be readable on a shared phone and shareable via WhatsApp in one tap.

05

The AI supports. The doctor decides.

This feature explains and guides — it never diagnoses. If blood values are dangerously low, the screen shows nothing but a clear message to call the Everhope care team immediately. No tips, no remedies, just: call now. This boundary was agreed with our clinical team from day one.

06

Progress across cycles keeps people engaged

Showing a patient their haemoglobin is 8.9 is a data point. Showing them it improved from 8.2 in Cycle 1 to 9.8 in Cycle 3 — because of the changes they made — is motivation. Tracking improvement over the full course of treatment is what turns this from a tool into a companion.

Design Decisions

Six choices that defined how this feature works.

Good design is a series of deliberate decisions. Here are the six that mattered most — what we chose, why, and what it changed.

Decision 01

The AI summary always comes before the numbers

Early versions showed the list of blood values first. In testing, patients saw one red value and panicked before reading anything else. We flipped it: the summary comes first, always. When patients read context before data, they understood better and were much less anxious. This one reorder was the most impactful single change in the whole project.

↓ 34% drop in reported anxiety during report review
Decision 02

Three states: Normal, Watch, and Low — not just green and red

We nearly launched with just two colours. A clinical advisor pointed out that values getting close to dangerous — but not yet there — needed their own signal. Adding an amber "Watch" state stopped patients from treating every session as a crisis, and made "Red" genuinely alarming when it appeared. Caregivers in testing said this was the most helpful change we made.

↑ User trust improved by 28% after adding Watch state
Decision 03

Every recommendation is built for someone on active cancer treatment

Generic health apps say "eat more iron" for low haemoglobin. But a patient undergoing chemotherapy might have mouth sores that make chewing painful, or nausea that makes eating nearly impossible. Every recommendation in this feature was co-developed with Everhope's oncology nutritionists and validated against the most common treatment types and side-effect profiles.

91% of families rated the guidance as "specific enough to act on"
Decision 04

The Share button is always visible — because the family is the real user

Research showed that in most Indian homes, the family member does the cooking, manages medicines, and takes notes at appointments. If the guidance only reached the patient, very little of it would get implemented. We put a persistent Share button on every report and guidance screen. Post-launch: 62% of users shared a screen with a family member within 24 hours.

62% of reports shared with family within 24 hours
Decision 05

A simple water tracker instead of a full wellness module

We considered building a detailed daily health log. We built a water tracker instead. Dehydration was the single most neglected and most impactful behaviour we found in our research. The drop visualisation and push reminders required no explanation, worked on any phone, and took one second to use. 71% of users set a hydration reminder in their first two weeks.

71% of users set a daily hydration reminder within 2 weeks
Decision 06

For dangerously low values: no tips, just call the care team

If neutrophils or platelets drop to a dangerous level, the entire Lab Summary screen is replaced with one clear message: call Everhope immediately. No AI tips. No remedies. Just a single action. This was non-negotiable from our clinical team — and it was the decision that earned their trust to build this feature in the first place.

Zero adverse incidents from missed escalation post-launch
Before vs After

What actually changed for patients between sessions.

Before the feature
After the feature
Understanding the report
Got the report. Couldn't explain any value. Sent it to relatives on WhatsApp hoping for answers.
Plain-language AI summary within seconds of uploading. 78% of users understood their report fully for the first time.
Connecting symptoms to values
Fatigue, mouth sores, and bruising written off as "normal side effects." No connection made to specific blood values.
Each symptom shown next to the value causing it. Patients understand why they feel what they feel.
Diet guidance
Told to "eat well." Cooked the same meals. No idea which specific foods could help which specific values.
Specific, treatment-appropriate food guidance for each low value. Shareable with the family member doing the cooking.
Drinking enough water
No tracking. Most patients drinking under 1 litre a day during fatigue periods — making side effects much worse.
Daily water tracker with reminders. Average hydration among active users up 58%.
Arriving prepared for chemo
No preparation guidance between sessions. 1 in 4 sessions delayed due to low blood counts.
Daily checklist 5–7 days before each session. 41% fewer postponements in the active user group.
Results — 90 Days After Launch

The numbers that made the medical team pay attention.

Data from in-app analytics, post-session patient surveys (312 responses), a comparison of chemo postponement rates between users and non-users at Everhope's Gurugram centres, and a follow-up usability study with 14 patients across multiple treatment cycles.

was 17% before launch
78%
Patients who fully understood their lab report after using the feature
was 26% before launch
−41%
Fewer chemo sessions postponed due to avoidable low blood counts
baseline 1.0×
3.2×
More patients following daily diet and hydration guidance
post-launch survey
91%
Patients and family members who said the feature genuinely helped them
62%
Shared a report or tip with a family member within 24 hours of viewing
71%
Set a daily water reminder within their first two weeks on the feature
+19 pt
Improvement in app NPS score among Lab Insights users vs non-users
4.1×
Higher frequency of app sessions among users who activated this feature

"My husband's haemoglobin kept dropping before every session and we didn't know what to do. The app showed me exactly what to cook. His third-cycle count was the best it had been in months. The doctor actually asked us what had changed."

— Caregiver, wife of Stage III colorectal cancer patient · Post-launch interview · Gurugram, 2024

"I used to dread looking at the report. Now I open the app first, read the summary, and I actually understand what's going on. Just knowing — without having to Google it or call someone at midnight — makes everything less frightening."

— Meera, 52 · Breast cancer patient · Cycle 5 feedback · Gurugram, 2024
What I Learned

The things this project taught me about designing for people in difficult situations.

🧠

How a person feels when they open the app is a design constraint

Most apps are designed for a calm, curious user. But someone opening a cancer blood report is likely scared, tired, and possibly in physical discomfort. That changes everything — the words you use, the order you show information, the colours on screen. I now think about a user's emotional state at the moment they open the app as seriously as I think about their device or screen size.

👩‍⚕️

Doctors and nutritionists should be in the room from week one — not at the end

We brought Everhope's clinical team in during the research phase. This saved us from building things that looked right but weren't — like recommending foods that are actually problematic during certain treatments, or using exercise guidance that ignores how extreme fatigue can get. Clinical input early isn't a bottleneck. It's what makes the product worth trusting.

🤝

In India, the family is the patient — design for both from the start

In most Indian households, it's a family member — not the patient — who manages the meals, tracks medications, and comes to appointments. We nearly missed this entirely. Once we understood it, everything changed: every screen became something a family member needed to understand, every recommendation became something the person doing the cooking needed to act on. Designing for the household, not the individual, was the biggest shift in our thinking.

🏥

AI in a medical context earns trust by staying close to the doctor — not by replacing them

We spent months making the AI summaries accurate and clear. But in testing, patients would read a perfect summary and immediately ask: "has a real doctor seen this?" The answer was to keep the doctor's note visible on every screen, not buried in a separate section. When the AI said something and the patient could see their own doctor saying something similar two cards below, trust followed immediately. The AI doesn't replace the clinician. It extends them.

📈

Don't design for the single session — design for the whole journey

Cancer treatment is months of connected cycles, not isolated appointments. The most meaningful thing this feature can do is show a patient their haemoglobin improving from 8.2 in Cycle 1 to 9.8 in Cycle 3 — and connect that progress to the changes they made. I'd build that longitudinal view much earlier in v2. It's not a feature to add later. It's the whole story.

What Comes Next

This feature filled the gap between sessions.
The next version should close the full loop.

We proved that patients act on guidance when it's clear, specific, and trusted. The next phase takes that foundation and makes the experience proactive — anticipating problems before they become delays.

Predict problems before the next blood test Use trend data across cycles to flag values that look like they're heading toward dangerous territory — before the test even happens. Move from reactive to anticipatory.
A dedicated view for the family caregiver The family member needs a different view than the patient. Same data, different priorities — focused on what to cook today, what medications are due, and when to call the care team.
Guidance that changes as treatment progresses What works in Cycle 2 may not work in Cycle 5. Recommendations should adapt based on the current cycle, what side effects have appeared, and what has actually worked before.
Send a pre-session summary directly to the oncologist One tap: labs, symptoms, water intake, and checklist progress sent to the treating doctor the day before each session. So the doctor knows how the patient has been — before they even walk in.