MSc. Design for Interaction, TU Delft
Bremo
A Personalised AI Assistant for Individuals with Breast Cancer Post-Diagnosis
Technical University of Delft, Netherlands
.png)
Project Type
Role
Skills
Tools
Year
UX Research, Product Design
Project Lead
Metro Mapping, Design Thinking, Teamwork, Time Management
Figma, Figjam, atlas.ti, Miro, Adobe Photoshop, Adobe Illustrator
2024
Project Overview:
Breast Cancer is a frequent and commonly diagnosed cancer in females aged 45-75. Post-diagnosis, individuals receive a lot of information about their cancer which can be complex and result in information overload. Information overload is a multi-dimensional construct and a complex phenomenon with various factors such as volume of information, complexity, available time, fragmented sources, etc. The challenge was to understand their needs and support individuals in dealing with information overload.
​
Project Outcome:
A personal AI companion concept that is proposed to be accessible via chat or voice interaction to support individuals with breast cancer. It serves as a centralised gateway connecting various existing credible sources of information in the Netherlands.
​
Learnings from this Project:
Developed research analysis skills, Metro Mapping, how to effectively manage project timelines, embraced the principle of "kill your darlings," design process, and the importance of taking healthy breaks.
​
​
Interactive Visions​
​
A digital intervention for museum visitors to discover depths and stories of intricate paintings
​
​
​
​​
​
​
Read Case Study
The Background
Breast Cancer is a frequent and commonly diagnosed cancer in females aged 45-75. According to the National Institute of Public Health and the Environment of the Netherlands, approximately one in eight females will develop breast cancer at some point in their lives. The growing number of cases demands more holistic support during this period. ​An individual with cancer has numerous tasks. They consult the healthcare providers, visit the hospital, go through examinations, learn about the illness, get medications and make decisions. They engage with various people which can appear as a complex web of interactions. ​
I found myself asking—
"How do they navigate this complexity? With so many interactions and decisions to make? How do they make decisions in moments of vulnerability, when emotions run high? How do they understand and process information and arrive at decisions?"
"What do they need and how can I help as a designer?"
So, I started my research and framed an overarching research question
Research Question
"How can an AI-based intervention support individuals with breast cancer in dealing or preventing information overload post-diagnosis?"
But wait, Why AI?
Since the 2019 COVID pandemic, healthcare has undergone a digital transformation. AI's integration into the healthcare system provides valuable support by streamlining repetitive and time-consuming tasks of healthcare providers. Apart from its application in medical imaging, diagnostics, and virtual patient care, it can enhance patient engagement through interactive platforms such as websites, mobile health applications, chatbots, etc. AI technologies can provide individuals with relevant, accurate and understandable information by tailoring it to their specific needs and preferences.
.png)
The Process
To find the answer to the research question
I conducted
25+ Qualitative Interviews
4 Healthcare Providers and Nurse Practitioners
7 Individuals with Breast Cancer
2 Companions
7 Representatives of Research Institutes and Cancer Organisations
​2 AI Experts
​
and a few more conversations
and gathered
Insights from the Interviews
Overwhelming Nature of Cancer Diagnosis
Nature of Information: Complex and Abundant
Influence of Seeking Information from Various Sources
Variability in Seeking and Understanding Information
Short Duration of the Consultation Sessions
Avoid Joining the Patient Organisation
Variations in Information-Seeking Behaviours
​Varied Emotional Support and Coping Strategies
Image 1: Metro Mapping: Information Map in Breast Cancer Care; Image 2: Thematic Analysis of Interviews
Glimpses from the interview
- Healthcare Provider
"Sometimes the information is too hard to understand. And that’s not easy."
"She said, “now I have a new job because the job is now breast cancer."
- Representative from Patient Organisation
"I think the world, this hole opened up and I just fell like several levels through the floor."
- Individual with Breast Cancer
"I’m not educated as a physician. So I think some of the abbreviations were new to me [...] so I think it’s a lot of professional language that you are suddenly confronted with."
- Individual with Breast Cancer
and then translated the findings using thematic analysis to find
The Unmet Needs
Despite the support available from various sources, the overwhelming nature of cancer diagnosis makes individuals feel isolated and highlights the need for more support that can address their information and emotional needs. These needs are interconnected and interact continuously with each other. To reduce uncertainty, they go look for information and due to the nature of illness and the information it presents, they need emotional support. The findings also complement the literature study. Although there is an overlap between these needs, they vary from person to person and hence require more personalised support.
The Design
Needs > Design Principles > Design Goal > Design Explorations > Design Characteristics > Desired Interaction
Design Outcome: Bremo
Creative Facilitation Workshops
With fellow design students
With healthcare designers at Panton
The Interactions
-
One-dimensional
-
Communication Barrier
-
Lack of Clarity
-
The information presented is Fragmented, complex and irrelevant to the particular case
Current Interaction​
-
A centralised source of information that facilitates conversations with the healthcare providers and helps understand information through simple Interactive conversations
Desired Interaction​
Introducing Bremo
Bremo breaks the traditional, one-dimensional approach of static documents by offering interactive, conversational access to information, from multiple reliable sources. It also serves as a sounding board where individuals can share their feelings and concerns. It is envisioned as a secondary source of information to be accessible via smartphones.
Login
Chat with Bremo
Creating a Question Bank
Bremo: Personalized AI companion to support individuals in understanding information about their breast cancer post-diagnosis
Evaluation
What did participants think of Bremo?
- Individual with Breast Cancer
"I like your idea of not reinventing the content, but using the really good content that's out there and maybe these sites all do talk a lot about what is it. I think you found something that there's a gap and then it's valuable. I like that it's very integrative with various established data sources."
“I would maybe say very useful because if you can use, use it as a recording app for your conversations and have it summarize it, and also have the only information that you need in one app then is each, it could make our complete patient journey a little easier and even and we reduce some of the stress.”
- Healthcare Designer from Panton
Desirability, Feasibility and Viability
During the discussion with healthcare designers at Panton, a mapping activity to determine the concept's desirability and feasibility was done. Some concepts were found to be desirable and added value to the individual's life. For the concept's viability, it was suggested that it could be incorporated into the existing tools like Luscii applications which integrate with other healthcare applications in the Netherlands.
Trust, Privacy and Bias
With opportunities, comes challenges related to trust, privacy, and confidentiality of patient data with AI. AI systems are versatile but also complex and unpredictable. Tools like these might influence the further development and enhancement of ethical guidelines for AI in healthcare.
There were of course some limitations, design recommendations for the future and some proposed evaluation methods
​
To read about it in detail, read the entire report here.