Mobile app for smarter greenhouse training

SkyAcres App · Agritech · Education · Mobile App

Team

Individual

My Role

Product Designer

Duration

1 month

My contribution

Research, UI/UX design, Prototype, Usability test

Overview

Controlled Environment Agriculture (CEA) is growing rapidly, but onboarding new workers in vertical farms and greenhouses remains a challenge. Training is often manual, inconsistent, and difficult to scale across large facilities. SkyAcres company set out to design a mobile-first training app that provides clarity, structure, and confidence for new trainees helping them adapt quickly and perform complex tasks with ease. My role was to design the core user experience for trainees, ensuring the app works seamlessly in real-world farming environments.

The Solution

Dual-interface web platform: Navigator module for patient care, Admin module for oversight and analytics.

Patient Navigator Module

Core Features:

Appointments Dashboard with real-time delay tracking and priority indicators

→ Designed for frontline care coordination

Administrator Module

Core Features:

→ Built for strategic oversight and analytics

Official Launch

Roche Africa and the Embrace Society announce Ghana's first patient navigation app

LinkedIn Article

Official Announcement

First Patient Navigation App Launches in Ghana, Connecting Patients to Care Pathways

By Embrace Society Ghana & Roche Africa

Roche Africa is proud to announce that the Embrace Patient Navigation Programme (PNP) has launched Ghana's first breast cancer patient navigation app. This pioneering digital platform supports the patient navigation model that has already helped over 1000 breast cancer patients navigate their journey through 15 public institutions across Ghana.

Read full article

1000+

Patients Navigated

20

Trained Navigators

15

Public Institutions

A Beginning No One Chooses

Let me introduce you to

Sarah

.

She's sitting in a clinic room hearing the words:

"You need treatment."

And in that moment, the world around her blurs. She nods, she signs, she listens… but in her heart, she's terrified.

"You need treatment."

The moment that changes everything

Then, A System That Breaks Hearts

The Human Reality

Women face screening with fear, confusion, and often without support. Many don't know where to start, what to expect, or who to trust.

The Navigator's Burden

Patient navigators juggle screening camps, track dozens of patients, coordinate referrals, and provide emotional support -all with paper records and scattered data.

85%

of navigators feel overwhelmed by manual tracking

60%

of referrals get lost in paper systems

40%

of patients miss follow-up appointments

The tool navigator uses…

does not see Sarah.

PDFs

Emails

Missing docs

Conflicting forms

Sarah becomes a case number. A file. A tab in a spreadsheet.

The real pain isn't only the disease.

It's the

waiting.

It's the

not knowing.

It's the

silence.

Where I Entered

This is where I entered the project.

Not as a designer who builds screens, but as someone who asked:

"How do we design for

fear

?"

"How do we design for

uncertainty

?"

"How do we design for people who

cannot wait

?"

Research & Discovery

2 months of research before design - understanding workflows, pain points, and impact on cancer survival rates.

Remote Video Interviews

Why I chose this method:

I was working remote so, I conducted remote interviews with patient navigators and some recordings were given to me of patients. This actually proved beneficial as participants were more comfortable in their own environment and I could observe their actual workspace setup.

Key Insights Discovered:

Navigators kept paper notebooks with patient notes - extremely personal and detailed

Phone was the primary communication tool, not official hospital systems

Many navigators created their own tracking systems in Excel spreadsheets

Trust and personal relationships were the foundation of successful navigation

Field Observations at Clinics

Why I chose this method:

While interviews provided insights into stated behavior, understanding real workflows required in-context observation. We had a manager visit three clinics across Ghana to shadow patient navigators during screening camps, follow-up calls, and patient encounters, capturing on-ground workflows firsthand. This helped uncover the gap between what people say they do and what actually happens in practice.

Key Insights Discovered:

Navigators constantly interrupted during patient calls - environment matters

Mobile phones used while walking between departments - desktop isn't realistic

Limited internet connectivity - offline mode is crucial, not optional

Patients often showed up unannounced - rigid scheduling doesn't work

Journey Mapping Workshops

Why I chose this method:

I needed to map the complete patient journey from screening to treatment. In collaboration workshops with navigators and administrators, we created detailed journey maps identifying every touchpoint, handoff, and potential failure point. This revealed where patients were getting lost in the system.

Key Insights Discovered:

Critical drop-off after referral issued - no system to track if patient followed through

Transportation and cost were biggest barriers, not awareness

Navigators spent 2 days/month manually compiling Excel reports for funders

Administrators couldn't identify problems until quarterly reviews - too late to intervene

User Personas: Two Distinct Needs

Through research, I identified two primary user types with completely different workflows and needs. Designing for both without compromising either was a core challenge.

Persona 1: The Patient Navigator

Age 28-45, Nurses or Social Workers, Managing 40-60 active patients

Daily Tasks:

Follow-up calls, appointment scheduling, barrier identification, emotional support, care coordination

Pain Points:

Overwhelmed by admin work, can't remember all patient details, losing track of who needs follow-up when

Design Implication:

Need automated reminders, quick patient lookup, mobile-friendly interface for calls on-the-go

Persona 2: The Program Administrator

Age 35-55, Program Managers, Overseeing 15-20 navigators across institutions

Daily Tasks:

Performance monitoring, reporting to funders, identifying systemic bottlenecks, resource allocation

Pain Points:

No visibility into real-time data, manual report compilation takes weeks, can't identify problems until too late

Design Implication:

Need dashboard with real-time metrics, exportable reports, delay identification by institution/navigator

The Solution

Dual-interface web platform: Navigator module for patient care, Admin module for oversight and analytics.

Patient Navigator Module

Core Features:

Appointments Dashboard with real-time delay tracking and priority indicators

→ Designed for frontline care coordination

Administrator Module

Core Features:

→ Built for strategic oversight and analytics

Design Process & Iterations

Evolved through real user feedback — watching navigators use prototypes taught more than any design theory.

1

Wireframes

Paper sketches mapping information architecture & task flows

2

Prototypes

Clickable Figma prototypes tested via WhatsApp across Ghana

3

High-Fidelity

Refined with real data, offline mode, accessibility optimizations

What I Learned from Version 1 (Sharing few)

These are the some of the problems discovered during testing that shaped the entire design direction. Each problem led to fundamental changes in how the system works.

Navigators had to click each patient to see delay status

❌ Version 1

User feedback: "I can't quickly see who needs urgent attention"

✓ Final Version

Solution: Added color-coded delay indicators in table

✓ Result: 70% faster triage of urgent cases

Overwhelming for users, high error rate on required fields.

❌ Version 1

⚠️ 32% of forms had missing required data

✓ Final Version

Solution: Split into logical 5-step workflow with progress indicator

✓ Reduced errors by 32%, completion time down 40%

Inconsistent camp screening data, hard to aggregate

❌ Version 1

⚠️ Unable to generate accurate screening reports

✓ Final Version

Solution: Structured digital form with validation and auto-aggregation

✓ 100% data completeness, instant reporting capability

Accessibility & Privacy

Colors That Work in Sunlight - couldn't read anything in harsh clinic sunlight

❌ Version 1

⚠️ Problem: Washed out in sunlight, hard to read gray text, color-only signals fail for colorblind users

✓ Final Version

Solution: High contrast (7:1), icon + color + text

✓ WCAG 2.1 AAA Compliance

Text:

7.8–9.1:1 contrast (AAA: ≥7:1)

Icons:

7.8–9.1:1 contrast (AAA: ≥3:1)

Touch:

48×48px minimum target (Level AAA)

Redundant encoding:

Color + Icon + Text

Sunlight-tested:

Readable at 100,000 lux

The Glove Problem while using buttons

❌ Version 1

⚠️ Problem: Buttons too small for gloved hands, text unreadable while moving, targets too close together

✓ Final Version

Solution: 48px touch-friendly buttons, Glove-compatible design, 16px+ text, generous spacing

Shared Tablets - No indication who's logged in, stays unlocked indefinitely

❌ Version 1

⚠️ Problem: No indication who's logged in, stays unlocked indefinitely

✓ Final Version

Solution: Always-visible session, gentle logout warning, clear logout button

Final Designs

Designed for frontline healthcare workers managing day-to-day patient care

Comprehensive Appointment Management

The main dashboard displays all appointments with color-coded delay indicators (High/Med/Low), patient details, and quick actions. Stats cards show at-a-glance metrics for urgent cases and follow-ups needed.

Real-time delay tracking with priority indicators

Searchable patient database with filters

One-click appointment scheduling

Statistics cards for quick insights

5-Step Patient Onboarding

A comprehensive multi-step form guides navigators through collecting complete patient information - from personal details to medical history and screening results.

Step 1: Personal Information

Step 2: Medical History

Step 3: Insurance

Step 4: Appointment details

Step 5: Consent form

Outreach & Camp Management

A dedicated outreach module empowers navigators to run on-site breast-cancer screening camps, track attendees, and collect data even offline , syncing automatically later.

Quick patient entry (minimal fields, auto-save)

Auto-sync after connectivity

View participants grouped by camp name

Generate post-camp reports and attendance summaries

Track and Manage Patient Referral

A streamlined referral workflow helps patient navigators seamlessly connect patients to the right facilities, improving follow-up rates and ensuring timely diagnostics.

Create and manage referrals in one place

Monitor referral status (Pending, Accepted, Completed)

Track priority levels for urgent cases

Export referral summaries for reporting

Statistics & Reports for healthcare leaders

Comprehensive analytics dashboard with cards, charts, and detailed tables showing program performance and patient outcomes.

Treatment stage distribution visualizations

Exportable reports for stakeholders

Regional and facility-level breakdowns

Key performance indicators with trends

Impact & Results

Measured remotely using standard UX frameworks with on-ground teams in Ghana

78%

Fewer Missed Follow-ups

Patients completed their care journey with automated reminders

How We Measured

Behavioral Analytics

Appointment records tracked (N=847 patients)

Calculation

Before: 342/847 missed (40.4%) → After: 76/847 missed (9.0%)

Remote Coordination

Clinic directors shared data via Google Sheets weekly

+68

Patient NPS Score

Strong patient satisfaction and likelihood to recommend

How We Measured

Net Promoter Score (NPS)

Standard 0-10 scale survey (N=214 patients)

Calculation

75% Promoters - 7% Detractors = +68 (Excellent)

Remote Coordination

Field researchers conducted phone interviews, sent results via WhatsApp

93%

Rural Referral Tracking

CHAMs' rural screening referrals tracked successfully

How We Measured

Task Success Rate

CHAM referral logs tracked (N=156 referrals)

Calculation

Before: 58% tracked → After: 93% tracked successfully

Remote Coordination

8 CHAMs submitted weekly logs via SMS and WhatsApp

84.2

Navigator SUS Score

High system usability - Grade A "Excellent"

How We Measured

System Usability Scale (SUS)

Standard 10-item survey (N=12 navigators)

Calculation

Score: 84.2/100 = Grade A (Above 80 is excellent)

Remote Coordination

Navigators filled Google Forms, reviewed via Zoom calls

62%

Faster Task Completion

Key workflows completed significantly faster

How We Measured

Time on Task Analysis

Timed observations of core tasks (N=12 navigators)

Calculation

Before: 11.2 min/task → After: 4.3 min/task (62% faster)

Remote Coordination

Navigators self-reported times in daily tracking sheet

-71%

Fewer Data Entry Errors

Mistakes in patient records reduced significantly

How We Measured

Error Rate Tracking

Record audits by clinic staff (N=428 patient records)

Calculation

Before: 14 errors/100 records → After: 4 errors/100 records

Remote Coordination

Clinic directors reported error counts in weekly check-ins

8 weeks

Mar-Apr 2024

3 clinics

Across Ghana

12 navigators

+ 8 CHAMs

847 patients

Active journeys

Remote

Coordinated via Zoom

A New Era for Healthcare in Ghana

A New Era for Healthcare in Ghana

RedPath represents a transformative leap in Ghana's healthcare system - from manual, paper-based patient tracking to a comprehensive digital platform that empowers healthcare workers and saves lives. As the first patient navigation app in the country, RedPath sets a new standard for cancer care delivery and demonstrates the power of technology to address critical healthcare challenges.

Problem

Problem

New trainees struggle with complex greenhouse workflows that involve sensors, equipment, and repetitive tasks.

  • Training is often time-consuming, relying on supervisors for guidance.

  • Workers need to manage tasks under challenging conditions: wearing gloves, working under bright lights, and sometimes in areas with poor Wi-Fi.

Research

70%

Small agencies found billboard advertising confusing and hard to manage

60%

Agencies reported losing time in back-and-forth calls and manual spreadsheets

#1 frustration

Lack of performance tracking was identified as the biggest frustration

City-Owned

Billboards are owned by cities and leased to large distributors. Agencies take these billboards from them and run campaigns of their clients.

High Concentraton

Large players like Pattison Outdoor dominate the market with costly bulk deals

Locked Out

Smaller agencies and businesses often can’t access affordable inventory

New trainees struggle with complex greenhouse workflows that involve sensors, equipment, and repetitive tasks.

  • Training is often time-consuming, relying on supervisors for guidance.

  • Workers need to manage tasks under challenging conditions: wearing gloves, working under bright lights, and sometimes in areas with poor Wi-Fi.

Challenges identified:

Manual speaker identification

Split and unorganized notes

Time-consuming post-meeting summaries

Time-consuming meeting summaries

Goal: Reduce user effort and increase efficiency for a more intuitive, automated post-meeting experience.

Research Insights

Research Insights

Interviews with few realtors revealed:

80%

Record meetings but rarely rewatch

60%

Lose track of follow-ups

70%

70%

Want automatic summaries

Opportunity: Use AI to organize transcripts and action items in real time.

User Journey — Realtors

User Journey - Realtors

Phase

Key Actions

Pain Points

Pre-meeting

Schedule meeting, send invites

Manual contact setup

During meeting

Record, identify speakers

Speaker mismatch

Post-meeting

Write summary, assign tasks

Fragmented notes

Phase

Key Actions

Pain Points

Pre-meeting

Schedule meeting, send invites

Manual contact setup

During meeting

Record, identify speakers

Speaker mismatch

Post-meeting

Write summary, assign tasks

Fragmented notes

User Friction

User Friction

Cognitive Overload

"When people make errors, change the system so that type of error will be reduced or eliminated."

"When people make errors, change the system so that type of error will be reduced or eliminated."

— Don Norman

Previously, users had to manually match speakers with participants, increasing mental effort and chances of error. The new flow introduces AI-based speaker detection, reducing manual work and cognitive load.

Poor Data Visualization

"Recognition over recall: Users shouldn't have to remember what was in another tab."

"Recognition over recall: Users shouldn't have to remember what was in another tab."

— UX Principle

Earlier, notes and summaries were split across separate tabs, forcing users to recall information between views. The redesigned layout uses context-aware tabs that maintain visibility of key details, minimizing friction and improving clarity.

Solution - Recording Feature Reimagined

Solution - Recording Feature Reimagined

Pre-Meeting Setup

Select meeting type and participants.

Transcript Tagging

Automatic speaker labeling with confidence scores.

Transcript Summary

AI generates key insights and action items.

Pre-Meeting Setup

Select meeting type and participants.

Transcript Tagging

Automatic speaker labeling with confidence scores.

Transcript Summary

AI generates key insights and action items.

Before the Meeting

Before the Meeting

Add or search participants → select meeting type → start recording

During the Meeting

AI captures live transcript → identifies speakers → allows re-analysis

During the Meeting

During the Meeting

AI captures live transcript → identifies speakers → allows re-analysis

AI captures live transcript → identifies speakers → allows re-analysis

After the Meeting

After the Meeting

Summaries + Action Items auto-generated → Chat with AI for deeper context

Summaries + Action Items auto-generated → Chat with AI for deeper context

Anticipated Impact & Design Outcomes

Anticipated Impact & Design Outcomes

Designed to eliminate post-meeting friction, UltraSense is projected to transform how realtors capture and act on client conversations.

Designed to eliminate post-meeting friction, UltraSense is projected to transform how realtors capture and act on client conversations.

Estimated reduction in setup

Pre-filled participant lists and meeting-type detection simplify preparation by reducing manual input and repetitive steps.

Projected faster summaries

AI-generated key points and action items appear instantly after recording stops, replacing the need for manual note-taking

Instant AI recap access

A single-tap summary brings together transcript, speakers, and sentiment allowing realtor to revisit meetings without rewatching

Key Benefits for Realtors

Never Miss Follow-ups

Automatic action items ensure every client request is tracked

Save 2-3 Hours Weekly

Eliminate manual note-taking and post-meeting admin work

Better Client Relationships

Focus on the conversation, not on taking notes

Searchable Meeting History

Quickly find past discussions with transcript search

"It's like having an assistant who never forgets. I can be fully present with clients knowing everything is being captured."

Representative Realtor insights (based on research findings)

Reflection

Reflection

UltraSense showed how AI + design thinking can reduce user friction. By automating repetitive work, it transforms meetings into actionable insights.

"Design isn't about adding more — it's about removing friction."

"Design isn't about adding more - it's about removing friction."

Research

Research

80%

People had experienced minor injuries like neck strain, shoulder stiffness, or upper back pain

40%

Using YouTube or other apps weren’t confident if they are doing exercises the right way

60%

Skipped exercises due to cost, long wait times, or lack of motivation

1-6+ months

Public physiotherapy/Kinseology wait times in Canada

$75 - $150

Private sessions are too costly

Travel times

Rural areas: long travel distances, fewer options available

Research

Unstable Wi-Fi

Connectivity can be unreliable, so the app must work offline and sync later when a network is available.

Gloves & Bright Lights

Workers often use touchscreen gloves and operate under intense greenhouse lighting, requiring large tap targets and high-contrast colors.

Urgent Tasks

Critical issues like water leaks or pH imbalance need clear visual language red icons and always shown at the top.

Task Flow

Most tasks are auto-generated by sensors or routines, with guidance delivered in a step-by-step checklist format.

Scalability

Large farms can have hundreds of racks/zones, making a “My Systems” view and clear navigation essential.

Proof & Accountable

Supervisors need photo/video uploads, notes, and timestamps for task verification and traceability.

Solution

Solution

I created a mobile-first training app for new trainees in Controlled Environment Agriculture (CEA), designed to simplify onboarding in vertical farms and greenhouses. The app delivers sensor-driven tasks, step-by-step checklists, and short training videos, while supporting offline use and glove-friendly interactions. My focus was on making training clear, accessible, and confidence-building from day one.

Key feature // 1

Daily Streak & Progress Tracking

The app motivates trainees with a daily streak system and clear task summaries. Workers can see weather conditions, planned tasks with estimated hours, and track start-to-finish times. On completion, streaks and stats reinforce consistency and productivity in the field.

Key feature // 2

Task Management

The dashboard helps greenhouse and farm workers stay organized with a clear view of tasks, estimated time, and progress. Urgent tasks are flagged in red and shown on top, while each card includes location, duration, and a one-tap start button. Workers can easily browse remaining tasks like leak detection or nutrient checks, with a progress bar keeping them on track throughout the day.

Key feature // 3

Guided task with visual systems map

The SkyAcres app guides workers step by step with clear instructions, videos, and checklists, ensuring accuracy even for new trainees. Tasks are pinned on a visual system map, showing racks and sections with color-coded markers (red for urgent, green for pending). Workers can tap a marker to view details like product type and container count, making it easy to locate, understand, and complete tasks in real time.

Results

Results

Accessibility Test and Impact

Designed for real greenhouse conditions, I ensured tasks stay clear, quick, and easy to complete. Workers can navigate confidently even with gloves, bright light, and urgent demands.

Big Taps

Buttons sized for easy use with gloves.

High Contrast

Ensured visibility under strong greenhouse lights using hight contrast green and orange

Text to Speech

Enabled quick input without typing.

Impact

Faster task completion when urgent tasks were clearly flagged and prioritized. Fewer errors with step-by-step checklists and visual guidance Stronger adoption by new trainees through text-to-speech notes and large tap targets


Takeaways

Takeaways

Designing with the realities of greenhouse environments gloves, bright light, and urgent conditions highlighted the importance of clarity and accessibility. By focusing on prioritization, visual language, and easy input methods, the app not only improved efficiency but also increased trainee engagement and learning outcomes.

SkyAcres App · Agritech · Education · Mobile App

Mobile app for smarter greenhouse training

Team

Individual

My Role

My Role

Product Designer

Duration

Duration

1 month

My contribution

Research, UI/UX design, Prototype, Usability test

Official Launch

Roche Africa and the Embrace Society announce Ghana's first patient navigation app

LinkedIn Article

Official Announcement

First Patient Navigation App Launches in Ghana, Connecting Patients to Care Pathways

By Embrace Society Ghana & Roche Africa

Roche Africa is proud to announce that the Embrace Patient Navigation Programme (PNP) has launched Ghana's first breast cancer patient navigation app. This pioneering digital platform supports the patient navigation model that has already helped over 1000 breast cancer patients navigate their journey through 15 public institutions across Ghana.

Read full article

1000+

Patients Navigated

20

Trained Navigators

15

Public Institutions

A Beginning No One Chooses

Let me introduce you to

Sarah

She's sitting in a clinic room hearing the words:

"You need treatment."

And in that moment, the world around her blurs. She nods, she signs, she listens… but in her heart, she's terrified.

"You need treatment."

The moment that changes everything

Then, A System That Breaks Hearts

The Human Reality

Women face screening with fear, confusion, and often without support. Many don't know where to start, what to expect, or who to trust.

The Navigator's Burden

Patient navigators juggle screening camps, track dozens of patients, coordinate referrals, and provide emotional support -all with paper records and scattered data.

85%

of navigators feel overwhelmed by manual tracking

60%

of referrals get lost in paper systems

40%

of patients miss follow-up appointments

The tool navigator uses…

does not see Sarah.

Sarah becomes a case number. A file. A tab in a spreadsheet.

PDFs

Emails

Missing docs

Conflicting forms

Research & Discovery

2 months of research before design - understanding workflows, pain points, and impact on cancer survival rates.

Remote Video Interviews

Why I chose this method:

I was working remote so, I conducted remote interviews with patient navigators and some recordings were given to me of patients. This actually proved beneficial as participants were more comfortable in their own environment and I could observe their actual workspace setup.

Key Insights Discovered:

Navigators kept paper notebooks with patient notes - extremely personal and detailed

Phone was the primary communication tool, not official hospital systems

Many navigators created their own tracking systems in Excel spreadsheets

Trust and personal relationships were the foundation of successful navigation

Field Observations at Clinics

Why I chose this method:

While interviews gave me stated behavior, I needed to observe actual workflows in context. I visited three clinics across Ghana, shadowing navigators during screening camps, follow-up calls, and patient encounters. This revealed the gap between what people say they do and what they actually do.

Key Insights Discovered:

Navigators constantly interrupted during patient calls - environment matters

Mobile phones used while walking between departments - desktop isn't realistic

Limited internet connectivity - offline mode is crucial, not optional

Patients often showed up unannounced - rigid scheduling doesn't work

Journey Mapping Workshops

Why I chose this method:

I needed to map the complete patient journey from screening to treatment. In collaboration workshops with navigators and administrators, we created detailed journey maps identifying every touchpoint, handoff, and potential failure point. This revealed where patients were getting lost in the system.

Key Insights Discovered:

Critical drop-off after referral issued - no system to track if patient followed through

Transportation and cost were biggest barriers, not awareness

Navigators spent 2 days/month manually compiling Excel reports for funders

Administrators couldn't identify problems until quarterly reviews - too late to intervene

User Personas: Two Distinct Needs

Through research, I identified two primary user types with completely different workflows and needs. Designing for both without compromising either was a core challenge.

Persona 1: The Patient Navigator

Age 28-45, Nurses or Social Workers, Managing 40-60 active patients

Daily Tasks:

Follow-up calls, appointment scheduling, barrier identification, emotional support, care coordination

Pain Points:

Overwhelmed by admin work, can't remember all patient details, losing track of who needs follow-up when

Design Implication:

Need automated reminders, quick patient lookup, mobile-friendly interface for calls on-the-go

Persona 2: The Program Administrator

Age 35-55, Program Managers, Overseeing 15-20 navigators across institutions

Daily Tasks:

Performance monitoring, reporting to funders, identifying systemic bottlenecks, resource allocation

Pain Points:

No visibility into real-time data, manual report compilation takes weeks, can't identify problems until too late

Design Implication:

Need dashboard with real-time metrics, exportable reports, delay identification by institution/navigator

SkyAcres App · Agritech · Education · Mobile App

Mobile app for smarter greenhouse training

Team

Individual

My Role

Product Designer

Duration

1 month

My contribution

Research, UI/UX design, Prototype, Usability test

Overview

Controlled Environment Agriculture (CEA) is growing rapidly, but onboarding new workers in vertical farms and greenhouses remains a challenge. Training is often manual, inconsistent, and difficult to scale across large facilities. SkyAcres company set out to design a mobile-first training app that provides clarity, structure, and confidence for new trainees helping them adapt quickly and perform complex tasks with ease. My role was to design the core user experience for trainees, ensuring the app works seamlessly in real-world farming environments.

Core Mission

"Faster Pathways Save Lives" - Reducing treatment delays through digital patient navigation

Official Launch

Roche Africa and the Embrace Society announce Ghana's first patient navigation app

LinkedIn Article

Official Announcement

First Patient Navigation App Launches in Ghana, Connecting Patients to Care Pathways

By Embrace Society Ghana & Roche Africa

Roche Africa is proud to announce that the Embrace Patient Navigation Programme (PNP) has launched Ghana's first breast cancer patient navigation app. This pioneering digital platform supports the patient navigation model that has already helped over 1000 breast cancer patients navigate their journey through 15 public institutions across Ghana.

Read full article

1000+

Patients Navigated

20

Trained Navigators

15

Public Institutions

Research & Discovery

2 months of research before design - understanding workflows, pain points, and impact on cancer survival rates.

Remote Video Interviews

Why I chose this method:

I was working remote so, I conducted remote interviews with patient navigators and some recordings were given to me of patients. This actually proved beneficial as participants were more comfortable in their own environment and I could observe their actual workspace setup.

Key Insights Discovered:

Navigators kept paper notebooks with patient notes - extremely personal and detailed

Phone was the primary communication tool, not official hospital systems

Many navigators created their own tracking systems in Excel spreadsheets

Trust and personal relationships were the foundation of successful navigation

Field Observations at Clinics

Why I chose this method:

While interviews gave me stated behavior, I needed to observe actual workflows in context. I visited three clinics across Ghana, shadowing navigators during screening camps, follow-up calls, and patient encounters. This revealed the gap between what people say they do and what they actually do.

Key Insights Discovered:

Navigators constantly interrupted during patient calls - environment matters

Mobile phones used while walking between departments - desktop isn't realistic

Limited internet connectivity - offline mode is crucial, not optional

Patients often showed up unannounced - rigid scheduling doesn't work

Journey Mapping Workshops

Why I chose this method:

I needed to map the complete patient journey from screening to treatment. In collaboration workshops with navigators and administrators, we created detailed journey maps identifying every touchpoint, handoff, and potential failure point. This revealed where patients were getting lost in the system.

Key Insights Discovered:

Critical drop-off after referral issued - no system to track if patient followed through

Transportation and cost were biggest barriers, not awareness

Navigators spent 2 days/month manually compiling Excel reports for funders

Administrators couldn't identify problems until quarterly reviews - too late to intervene

User Personas: Two Distinct Needs

Through research, I identified two primary user types with completely different workflows and needs. Designing for both without compromising either was a core challenge.

Persona 1: The Patient Navigator

Age 28-45, Nurses or Social Workers, Managing 40-60 active patients

Daily Tasks:

Follow-up calls, appointment scheduling, barrier identification, emotional support, care coordination

Pain Points:

Overwhelmed by admin work, can't remember all patient details, losing track of who needs follow-up when

Design Implication:

Need automated reminders, quick patient lookup, mobile-friendly interface for calls on-the-go

Persona 2: The Program Administrator

Age 35-55, Program Managers, Overseeing 15-20 navigators across institutions

Daily Tasks:

Performance monitoring, reporting to funders, identifying systemic bottlenecks, resource allocation

Pain Points:

No visibility into real-time data, manual report compilation takes weeks, can't identify problems until too late

Design Implication:

Need dashboard with real-time metrics, exportable reports, delay identification by institution/navigator

Final Designs

Designed for frontline healthcare workers managing day-to-day patient care

Comprehensive Appointment Management

The main dashboard displays all appointments with color-coded delay indicators (High/Med/Low), patient details, and quick actions. Stats cards show at-a-glance metrics for urgent cases and follow-ups needed.

Real-time delay tracking with priority indicators

Searchable patient database with filters

One-click appointment scheduling

Statistics cards for quick insights

5-Step Patient Onboarding

A comprehensive multi-step form guides navigators through collecting complete patient information - from personal details to medical history and screening results.

Step 1: Personal Information

Step 2: Medical History

Step 3: Insurance

Step 4: Appointment details

Step 5: Consent form

Outreach & Camp Management

A dedicated outreach module empowers navigators to run on-site breast-cancer screening camps, track attendees, and collect data even offline , syncing automatically later.

Quick patient entry (minimal fields, auto-save)

Auto-sync after connectivity

View participants grouped by camp name

Generate post-camp reports and attendance summaries

Track and Manage Patient Referral

A streamlined referral workflow helps patient navigators seamlessly connect patients to the right facilities, improving follow-up rates and ensuring timely diagnostics.

Create and manage referrals in one place

Monitor referral status (Pending, Accepted, Completed)

Track priority levels for urgent cases

Export referral summaries for reporting

Statistics & Reports for healthcare leaders

Comprehensive analytics dashboard with cards, charts, and detailed tables showing program performance and patient outcomes.

Treatment stage distribution visualizations

Exportable reports for stakeholders

Regional and facility-level breakdowns

Key performance indicators with trends

The real pain isn't only the disease.

It's the

waiting.

It's the

not knowing.

It's the

silence.

The Solution

Dual-interface web platform: Navigator module for patient care, Admin module for oversight and analytics.

Patient Navigator Module

Core Features:

Appointments Dashboard with real-time delay tracking and priority indicators

→ Designed for frontline care coordination

Administrator Module

Core Features:

→ Built for strategic oversight and analytics

A New Era for Healthcare in Ghana

RedPath represents a transformative leap in Ghana's healthcare system - from manual, paper-based patient tracking to a comprehensive digital platform that empowers healthcare workers and saves lives. As the first patient navigation app in the country, RedPath sets a new standard for cancer care delivery and demonstrates the power of technology to address critical healthcare challenges.

Impact & Results

Measured remotely using standard UX frameworks with on-ground teams in Ghana

78%

Fewer Missed Follow-ups

Patients completed their care journey with automated reminders

How We Measured

Behavioral Analytics

Appointment records tracked (N=847 patients)

Calculation

Before: 342/847 missed (40.4%) → After: 76/847 missed (9.0%)

Remote Coordination

Clinic directors shared data via Google Sheets weekly

+68

Patient NPS Score

Strong patient satisfaction and likelihood to recommend

How We Measured

Net Promoter Score (NPS)

Standard 0-10 scale survey (N=214 patients)

Calculation

75% Promoters - 7% Detractors = +68 (Excellent)

Remote Coordination

Field researchers conducted phone interviews, sent results via WhatsApp

93%

Rural Referral Tracking

CHAMs' rural screening referrals tracked successfully

How We Measured

Task Success Rate

CHAM referral logs tracked (N=156 referrals)

Calculation

Before: 58% tracked → After: 93% tracked successfully

Remote Coordination

8 CHAMs submitted weekly logs via SMS and WhatsApp

84.2

Navigator SUS Score

High system usability - Grade A "Excellent"

How We Measured

System Usability Scale (SUS)

Standard 10-item survey (N=12 navigators)

Calculation

Score: 84.2/100 = Grade A (Above 80 is excellent)

Remote Coordination

Navigators filled Google Forms, reviewed via Zoom calls

62%

Faster Task Completion

Key workflows completed significantly faster

How We Measured

Time on Task Analysis

Timed observations of core tasks (N=12 navigators)

Calculation

Before: 11.2 min/task → After: 4.3 min/task (62% faster)

Remote Coordination

Navigators self-reported times in daily tracking sheet

-71%

Fewer Data Entry Errors

Mistakes in patient records reduced significantly

How We Measured

Error Rate Tracking

Record audits by clinic staff (N=428 patient records)

Calculation

Before: 14 errors/100 records → After: 4 errors/100 records

Remote Coordination

Clinic directors reported error counts in weekly check-ins

12 navigators

+ 8 CHAMs

8 weeks

Mar-Apr 2024

3 clinics

Across Ghana

Remote

Coordinated via Zoom

847 patients

Active journeys

New trainees struggle with complex greenhouse workflows that involve sensors, equipment, and repetitive tasks.

  • Training is often time-consuming, relying on supervisors for guidance.

  • Workers need to manage tasks under challenging conditions: wearing gloves, working under bright lights, and sometimes in areas with poor Wi-Fi.

Challenges identified:

Manual speaker identification

Split and unorganized notes

Time-consuming post-meeting summaries

Goal: Reduce user effort and increase efficiency for a more intuitive, automated post-meeting experience.

Research Insights

Interviews with few realtors revealed:

80%

Record meetings but rarely rewatch

60%

Lose track of follow-ups

70%

Want automatic summaries

Opportunity: Use AI to organize transcripts and action items in real time.

User Journey - Realtors

Phase

Key Actions

Pain Points

Pre-meeting

Schedule meeting,

send invites

Manual contact setup

During meeting

Record, identify speakers

Speaker mismatch

Post-meeting

Write summary, assign tasks

Fragmented notes

User Friction

Cognitive Overload

"When people make errors, change the system so that type of error will be reduced or eliminated."

— Don Norman

Previously, users had to manually match speakers with participants, increasing mental effort and chances of error. The new flow introduces AI-based speaker detection, reducing manual work and cognitive load.

Poor Data Visualization

"Recognition over recall: Users shouldn't have to remember what was in another tab."

— UX Principle

Earlier, notes and summaries were split across separate tabs, forcing users to recall information between views. The redesigned layout uses context-aware tabs that maintain visibility of key details, minimizing friction and improving clarity.

Solution - Recording Feature Reimagined

Pre-Meeting Setup

Select meeting type and participants.

Transcript Tagging

Automatic speaker labeling with confidence scores.

Transcript Summary

AI generates key insights and action items.

Before the Meeting

Add or search participants → select meeting type → start recording

During the Meeting

AI captures live transcript → identifies speakers → allows re-analysis

During the Meeting

AI captures live transcript → identifies speakers → allows re-analysis

After the Meeting

Summaries + Action Items auto-generated → Chat with AI for deeper context

Anticipated Impact & Design Outcomes

Designed to eliminate post-meeting friction, UltraSense is projected to transform how realtors capture and act on client conversations.

Estimated reduction in setup

Pre-filled participant lists and meeting-type detection simplify preparation by reducing manual input and repetitive steps.

Projected faster summaries

AI-generated key points and action items appear instantly after recording stops, replacing the need for manual note-taking

Instant AI recap access

A single-tap summary brings together transcript, speakers, and sentiment allowing realtor to revisit meetings without rewatching

Key Benefits for Realtors

Never Miss Follow-ups

Automatic action items ensure every client request is tracked

Save 2-3 Hours Weekly

Eliminate manual note-taking and post-meeting admin work

Better Client Relationships

Focus on the conversation, not on taking notes

Searchable Meeting History

Quickly find past discussions with transcript search

"It's like having an assistant who never forgets. I can be fully present with clients knowing everything is being captured."

Representative Realtor insights (based on research findings)

Reflection

UltraSense showed how AI + design thinking can reduce user friction. By automating repetitive work, it transforms meetings into actionable insights.

"Design isn't about adding more — it's about removing friction."

Where I Entered

This is where I entered the project.

Not as a designer who builds screens, but as someone who asked:

"How do we design for

fear

?"

"How do we design for

uncertainty

?"

"How do we design for people who

cannot wait

?"

Impact & Results

Measured remotely using standard UX frameworks with on-ground teams in Ghana

78%

Fewer Missed Follow-ups

Patients completed their care journey with automated reminders

How We Measured

Behavioral Analytics

Appointment records tracked (N=847 patients)

Calculation

Before: 342/847 missed (40.4%) → After: 76/847 missed (9.0%)

Remote Coordination

Clinic directors shared data via Google Sheets weekly

+68

Patient NPS Score

Strong patient satisfaction and likelihood to recommend

How We Measured

Net Promoter Score (NPS)

Standard 0-10 scale survey (N=214 patients)

Calculation

75% Promoters - 7% Detractors = +68 (Excellent)

Remote Coordination

Field researchers conducted phone interviews, sent results via WhatsApp

93%

Rural Referral Tracking

CHAMs' rural screening referrals tracked successfully

How We Measured

Task Success Rate

CHAM referral logs tracked (N=156 referrals)

Calculation

Before: 58% tracked → After: 93% tracked successfully

Remote Coordination

8 CHAMs submitted weekly logs via SMS and WhatsApp

84.2

Navigator SUS Score

High system usability - Grade A "Excellent"

How We Measured

System Usability Scale (SUS)

Standard 10-item survey (N=12 navigators)

Calculation

Score: 84.2/100 = Grade A (Above 80 is excellent)

Remote Coordination

Navigators filled Google Forms, reviewed via Zoom calls

62%

Faster Task Completion

Key workflows completed significantly faster

How We Measured

Time on Task Analysis

Timed observations of core tasks (N=12 navigators)

Calculation

Before: 11.2 min/task → After: 4.3 min/task (62% faster)

Remote Coordination

Navigators self-reported times in daily tracking sheet

-71%

Fewer Data Entry Errors

Mistakes in patient records reduced significantly

How We Measured

Error Rate Tracking

Record audits by clinic staff (N=428 patient records)

Calculation

Before: 14 errors/100 records → After: 4 errors/100 records

Remote Coordination

Clinic directors reported error counts in weekly check-ins

8 weeks

Mar-Apr 2024

3 clinics

Across Ghana

12 navigators

+ 8 CHAMs

847 patients

Active journeys

Remote

Coordinated via Zoom

Impact & Results

Measured remotely using standard UX frameworks with on-ground teams in Ghana

78%

Fewer Missed Follow-ups

Patients completed their care journey with automated reminders

How We Measured

Behavioral Analytics

Appointment records tracked (N=847 patients)

Calculation

Before: 342/847 missed (40.4%) → After: 76/847 missed (9.0%)

Remote Coordination

Clinic directors shared data via Google Sheets weekly

+68

Patient NPS Score

Strong patient satisfaction and likelihood to recommend

How We Measured

Net Promoter Score (NPS)

Standard 0-10 scale survey (N=214 patients)

Calculation

75% Promoters - 7% Detractors = +68 (Excellent)

Remote Coordination

Field researchers conducted phone interviews, sent results via WhatsApp

93%

Rural Referral Tracking

CHAMs' rural screening referrals tracked successfully

How We Measured

Task Success Rate

CHAM referral logs tracked (N=156 referrals)

Calculation

Before: 58% tracked → After: 93% tracked successfully

Remote Coordination

8 CHAMs submitted weekly logs via SMS and WhatsApp

84.2

Navigator SUS Score

High system usability - Grade A "Excellent"

How We Measured

System Usability Scale (SUS)

Standard 10-item survey (N=12 navigators)

Calculation

Score: 84.2/100 = Grade A (Above 80 is excellent)

Remote Coordination

Navigators filled Google Forms, reviewed via Zoom calls

62%

Faster Task Completion

Key workflows completed significantly faster

How We Measured

Time on Task Analysis

Timed observations of core tasks (N=12 navigators)

Calculation

Before: 11.2 min/task → After: 4.3 min/task (62% faster)

Remote Coordination

Navigators self-reported times in daily tracking sheet

-71%

Fewer Data Entry Errors

Mistakes in patient records reduced significantly

How We Measured

Error Rate Tracking

Record audits by clinic staff (N=428 patient records)

Calculation

Before: 14 errors/100 records → After: 4 errors/100 records

Remote Coordination

Clinic directors reported error counts in weekly check-ins

8 weeks

Mar-Apr 2024

3 clinics

Across Ghana

12 navigators

+ 8 CHAMs

847 patients

Active journeys

Remote

Coordinated via Zoom

Design Process & Iterations

Evolved through real user feedback — watching navigators use prototypes taught more than any design theory.

1

Wireframes

Paper sketches mapping information architecture & task flows

2

Prototypes

Clickable Figma prototypes tested via WhatsApp across Ghana

3

High-Fidelity

Refined with real data, offline mode, accessibility optimizations

What I Learned from Version 1 (Sharing few)

These are the some of the problems discovered during testing that shaped the entire design direction. Each problem led to fundamental changes in how the system works.

Navigators had to click each patient to see delay status

❌ Version 1

User feedback: "I can't quickly see who needs urgent attention"

✓ Final Version

Solution: Added color-coded delay indicators in table

✓ Result: 70% faster triage of urgent cases

Overwhelming for users, high error rate on required fields.

❌ Version 1

⚠️ 32% of forms had missing required data

✓ Final Version

Solution: Split into logical 5-step workflow with progress indicator

✓ Reduced errors by 32%, completion time down 40%

Inconsistent camp screening data, hard to aggregate

❌ Version 1

⚠️ Unable to generate accurate screening reports

✓ Final Version

Solution: Structured digital form with validation and auto-aggregation

✓ 100% data completeness, instant reporting capability

Accessibility & Privacy

Colors That Work in Sunlight - couldn't read anything in harsh clinic sunlight

❌ Version 1

⚠️ Problem: Washed out in sunlight, hard to read gray text, color-only signals fail for colorblind users

✓ Final Version

Solution: High contrast (7:1), icon + color + text

✓ WCAG 2.1 AAA Compliance

Text:

7.8–9.1:1 contrast (AAA: ≥7:1)

Icons:

7.8–9.1:1 contrast (AAA: ≥3:1)

Touch:

48×48px minimum target (Level AAA)

Redundant encoding:

Color + Icon + Text

Sunlight-tested:

Readable at 100,000 lux

The Glove Problem while using buttons

❌ Version 1

⚠️ Problem: Buttons too small for gloved hands, text unreadable while moving, targets too close together

✓ Final Version

Solution: 48px touch-friendly buttons, Glove-compatible design, 16px+ text, generous spacing

Shared Tablets - No indication who's logged in, stays unlocked indefinitely

❌ Version 1

⚠️ Problem: No indication who's logged in, stays unlocked indefinitely

✓ Final Version

Solution: Always-visible session, gentle logout warning, clear logout button

Design Process & Iterations

Evolved through real user feedback — watching navigators use prototypes taught more than any design theory.

1

Wireframes

Paper sketches mapping information architecture & task flows

2

Prototypes

Clickable Figma prototypes tested via WhatsApp across Ghana

3

High-Fidelity

Refined with real data, offline mode, accessibility optimizations

What I Learned from Version 1 (Sharing few)

These are the some of the problems discovered during testing that shaped the entire design direction. Each problem led to fundamental changes in how the system works.

Navigators had to click each patient to see delay status

❌ Version 1

User feedback: "I can't quickly see who needs urgent attention"

✓ Final Version

Solution: Added color-coded delay indicators in table

✓ Result: 70% faster triage of urgent cases

Overwhelming for users, high error rate on required fields.

❌ Version 1

⚠️ 32% of forms had missing required data

✓ Final Version

Solution: Split into logical 5-step workflow with progress indicator

✓ Reduced errors by 32%, completion time down 40%

Inconsistent camp screening data, hard to aggregate

❌ Version 1

⚠️ Unable to generate accurate screening reports

✓ Final Version

Solution: Structured digital form with validation and auto-aggregation

✓ 100% data completeness, instant reporting capability

Accessibility & Privacy

Colors That Work in Sunlight - couldn't read anything in harsh clinic sunlight

❌ Version 1

⚠️ Problem: Washed out in sunlight, hard to read gray text, color-only signals fail for colorblind users

✓ Final Version

Solution: High contrast (7:1), icon + color + text

✓ WCAG 2.1 AAA Compliance

Text:

7.8–9.1:1 contrast (AAA: ≥7:1)

Icons:

7.8–9.1:1 contrast (AAA: ≥3:1)

Touch:

48×48px minimum target (Level AAA)

Redundant encoding:

Color + Icon + Text

Sunlight-tested:

Readable at 100,000 lux

The Glove Problem while using buttons

❌ Version 1

⚠️ Problem: Buttons too small for gloved hands, text unreadable while moving, targets too close together

✓ Final Version

Solution: 48px touch-friendly buttons, Glove-compatible design, 16px+ text, generous spacing

Shared Tablets - No indication who's logged in, stays unlocked indefinitely

❌ Version 1

⚠️ Problem: No indication who's logged in, stays unlocked indefinitely

✓ Final Version

Solution: Always-visible session, gentle logout warning, clear logout button

Design Process & Iterations

Evolved through real user feedback — watching navigators use prototypes taught more than any design theory.

1

Wireframes

Paper sketches mapping information architecture & task flows

2

Prototypes

Clickable Figma prototypes tested via WhatsApp across Ghana

3

High-Fidelity

Refined with real data, offline mode, accessibility optimizations

What I Learned from Version 1 (Sharing few)

These are the some of the problems discovered during testing that shaped the entire design direction. Each problem led to fundamental changes in how the system works.

Navigators had to click each patient to see delay status

❌ Version 1

User feedback: "I can't quickly see who needs urgent attention"

✓ Final Version

Solution: Added color-coded delay indicators in table

✓ Result: 70% faster triage of urgent cases

Overwhelming for users, high error rate on required fields.

❌ Version 1

⚠️ 32% of forms had missing required data

✓ Final Version

Solution: Split into logical 5-step workflow with progress indicator

✓ Reduced errors by 32%, completion time down 40%

Inconsistent camp screening data, hard to aggregate

❌ Version 1

⚠️ Unable to generate accurate screening reports

✓ Final Version

Solution: Structured digital form with validation and auto-aggregation

✓ 100% data completeness, instant reporting capability

Accessibility & Privacy

Colors That Work in Sunlight - couldn't read anything in harsh clinic sunlight

❌ Version 1

⚠️ Problem: Washed out in sunlight, hard to read gray text, color-only signals fail for colorblind users

✓ Final Version

Solution: High contrast (7:1), icon + color + text

✓ WCAG 2.1 AAA Compliance

Text:

7.8–9.1:1 contrast (AAA: ≥7:1)

Icons:

7.8–9.1:1 contrast (AAA: ≥3:1)

Touch:

48×48px minimum target (Level AAA)

Redundant encoding:

Color + Icon + Text

Sunlight-tested:

Readable at 100,000 lux

The Glove Problem while using buttons

❌ Version 1

⚠️ Problem: Buttons too small for gloved hands, text unreadable while moving, targets too close together

✓ Final Version

Solution: 48px touch-friendly buttons, Glove-compatible design, 16px+ text, generous spacing

Shared Tablets - No indication who's logged in, stays unlocked indefinitely

❌ Version 1

⚠️ Problem: No indication who's logged in, stays unlocked indefinitely

✓ Final Version

Solution: Always-visible session, gentle logout warning, clear logout button

Design Process & Iterations

Evolved through real user feedback — watching navigators use prototypes taught more than any design theory.

1

Wireframes

Paper sketches mapping information architecture & task flows

2

Prototypes

Clickable Figma prototypes tested via WhatsApp across Ghana

3

High-Fidelity

Refined with real data, offline mode, accessibility optimizations

What I Learned from Version 1 (Sharing few)

These are the some of the problems discovered during testing that shaped the entire design direction. Each problem led to fundamental changes in how the system works.

Navigators had to click each patient to see delay status

❌ Version 1

User feedback: "I can't quickly see who needs urgent attention"

✓ Final Version

Solution: Added color-coded delay indicators in table

✓ Result: 70% faster triage of urgent cases

Overwhelming for users, high error rate on required fields.

❌ Version 1

⚠️ 32% of forms had missing required data

✓ Final Version

Solution: Split into logical 5-step workflow with progress indicator

✓ Reduced errors by 32%, completion time down 40%

Inconsistent camp screening data, hard to aggregate

❌ Version 1

⚠️ Unable to generate accurate screening reports

✓ Final Version

Solution: Structured digital form with validation and auto-aggregation

✓ 100% data completeness, instant reporting capability

Accessibility & Privacy

Colors That Work in Sunlight - couldn't read anything in harsh clinic sunlight

❌ Version 1

⚠️ Problem: Washed out in sunlight, hard to read gray text, color-only signals fail for colorblind users

✓ Final Version

Solution: High contrast (7:1), icon + color + text

✓ WCAG 2.1 AAA Compliance

Text:

7.8–9.1:1 contrast (AAA: ≥7:1)

Icons:

7.8–9.1:1 contrast (AAA: ≥3:1)

Touch:

48×48px minimum target (Level AAA)

Redundant encoding:

Color + Icon + Text

Sunlight-tested:

Readable at 100,000 lux

The Glove Problem while using buttons

❌ Version 1

⚠️ Problem: Buttons too small for gloved hands, text unreadable while moving, targets too close together

✓ Final Version

Solution: 48px touch-friendly buttons, Glove-compatible design, 16px+ text, generous spacing

Shared Tablets - No indication who's logged in, stays unlocked indefinitely

❌ Version 1

⚠️ Problem: No indication who's logged in, stays unlocked indefinitely

✓ Final Version

Solution: Always-visible session, gentle logout warning, clear logout button