Limited insights
It was hard to translate genomic data into personalised treatment plans.
Project 03 · Health tech · AI SaaS
Genesilico
An AI-powered platform that helps oncologists, genetic counselors and patients personalise breast cancer treatment using genetic and cellular profiles, making care more effective and more affordable.
Screen placeholder
Hero: Patients list (web) + gSage AI assistant (mobile)
Recommended: 1600 × 900 · PNG or JPG
01
Personalised care was out of reach
Caring for breast cancer patients is complex and overwhelming. Treatment often lacks personalisation and affordability, which leads to poor outcomes and financial strain for patients.
Oncologists had genomic data but no easy way to turn it into a personalised treatment plan, and patients often couldn't afford the testing to begin with.
Key challenges
Fragmented workflows
Collaboration between healthcare professionals was inefficient.
Complex reports
Technical NGS (Next Generation Sequencing) reports were difficult to interpret.
High costs
Expensive NGS testing and analysis was out of reach for many patients.
Lack of monitoring
There were no tools to track patient progress or how well a treatment was working.
Missed clinical trials
Patients had few ways to find affordable clinical trials.
02
One platform, three kinds of user
Genesilico brings oncologists, genetic counselors and patients together on an integrated web and mobile platform built around personalised breast cancer care.
Web · Oncologists
- AI clinical assistant for quick access to patient insights
- Secure patient information: review test results and track patient health
- Order management: place NGS test orders, check status and review results
- NGS reporting: detailed reports to inform treatment decisions
Web · Genetic counselors
- Secure patient information: review test results and track patient health
- Order management: place NGS test orders, check status and review results
- NGS reporting: detailed reports to inform treatment decisions
Mobile · Patients
- Track health progress
- View NGS reports
- Stay updated on the treatment plan and how well it's working
- Message their healthcare providers
Together, this improves collaboration between clinicians, keeps patients engaged in their own care, and makes personalised treatment faster and cheaper to deliver.
03
Research and exploration
The mission was clear: give oncologists tools that simplify their workflows and improve patient care. I started from the business goals, which were data-driven insights and streamlined tools for better treatment decisions, then looked at what already existed.
- Business goals
- Competitive landscape
- Initial wireframes
- Early designs
- Oncologist interviews
- Iterate
Competitive landscape
| Platform | Early detection & screening | Genomic profiling | Drug response | Efficacy evaluation | Clinico-genomic database | Drug discovery support |
|---|---|---|---|---|---|---|
| Genesilico | 3rd party | Multiomic | AI-powered recommendation | AI-powered biological insights in patient profile | Screens patient genetic profile | APIs for clinico-genomic DB |
| Data Cancer GeneticsCancer research & molecular oncology | In-house | Multi-omics | AI-powered recommendation | — | — | — |
| Flatiron HCOncology EHR platform | — | — | Knowledge-based recommendation | Knowledge-graph insights in patient profile | Screens patient profile | — |
| Boston GenePrecision oncology reports in EHRs | In-house | Multi-omics | Knowledge-based recommendation | — | — | — |
| TempusPrecision medicine across specialties | 3rd party | Multi-omics | AI-powered recommendation | — | — | — |
Genesilico was the only platform covering every stage, from genomic profiling through efficacy evaluation to drug-discovery APIs.
Screen placeholder
Initial wireframes and early designs
Recommended: 1600 × 1000
04
Understanding oncologists
Early designs were only the start. To build something meaningful, we went to the people who would use it. Interviews with oncologists showed their daily challenges and the features they needed most.
Pain points
- DataPatient data was fragmented, so full histories, reports and treatment plans were hard to access.
- OrderingOrdering NGS tests was cumbersome, especially aligning with curated gene lists from bioinformatics teams.
- DecisionsWithout AI-driven insights, personalised treatment decisions were slow and complex, particularly for unique cases.
- CollaborationSharing patient data with peers and getting second opinions in real time was difficult.
Expectations
- AIIntelligent insights to guide treatment decisions, faster and more accurately.
- OrderingA seamless way to order NGS tests with curated gene information, with less manual effort and fewer errors.
- ReportsDetailed, easy-to-read NGS reports for understanding genetic variations and finding suitable treatments.
- PlanningPatient-specific treatment plans tailored to genetic profiles.
- TeamworkConsult peers, share patient data securely and make joint decisions efficiently.
- MobileAccess patient data quickly on the go.
05
Meet the oncologist
I built a persona from the research and the users' pain points, needs and expectations.
Bio
An experienced oncologist specialising in breast cancer at a metropolitan hospital. He runs a busy practice, joins tumour board meetings, and spends his days consulting patients, reviewing test results and creating treatment plans.
Aims
- Improve patient outcomes with the most advanced, effective treatments
- Stay at the forefront of breast cancer research so patients can access cutting-edge therapies
- Educate patients about their condition, treatment options and support
- Foster collaboration with his team and other healthcare professionals
Behaviour
- Long hours balancing patient care, treatment planning and specialists
- Attends tumour boards for multidisciplinary decisions
- Frequently checks test results, diagnostic reports and histories
- Not tech-savvy, so prefers a user-friendly interface
- Uses his phone to check patient data on the go
- Reads journals and joins forums to discuss cases with peers
06
Information architecture
This is where I planned how the product works, organising the flows and information so they're easy to use and understand.
Main screen
AI Assistance
- Chat interface
- Voice input
- Suggested prompts
- New chat
- Chat history
- Search · Pin · Rename
Patient list
- Add new patient
- Patient registration
- Patient profile (basic)
- View profile
- Clinical overview
- Timeline → Add event
- TRF ordered list / Start new TRF
- NGS report list → View · Download
- Clinical records → View · Upload
- Start TRF / Order
- View profile
Notifications / Activity
User profile
- Personal information
- Notification preferences
- Help & support
TRF = test requisition form, used to order an NGS test.
07
gSage, the AI clinical assistant
Busy oncologists need answers about their patients without digging through records. gSage puts patient insights a question away.
Screen placeholder
gSage: welcome & suggested prompts
Mobile · 1170 × 2532
Screen placeholder
gSage: answer with feedback controls
Mobile · 1170 × 2532
- Personalised welcome. It greets the doctor by name ("Welcome, Dr. Bob Parker"), so it feels like their assistant.
- Intuitive interface. Suggested prompts, chat history and "New chat" keep the doctor moving.
- Feedback system. Thumbs up, sideways and down let users rate each answer in a tap, and that feedback improves the system.
- Voice dictation. Busy oncologists can speak instead of type, which is more accessible and faster between consultations.
08
Ordering an NGS test
Test requisitions were long, error-prone forms. Incomplete or incorrect submissions delayed results that treatment decisions depended on.
Goal 1Streamline filling out and submitting test requisitions
Goal 2Reduce errors and incomplete submissions
Goal 3Give clear guidance at every step
Key design decision: select-and-type fields
Gene panel
BRCA▾
- BRCA1
- BRCA2
- + Use "BRCA1/2 + PALB2" as custom value
Problem
Users often couldn't find the exact option they needed in a dropdown, so they submitted incomplete or incorrect requisitions.
Solution
I designed select-and-type dropdowns. If the right option isn't in the list, users type a custom value straight into the field. That value is captured as a variable and stored for future use.
Progress indicator
Shows users where they are in the requisition and how many steps remain.
Validation & feedback
Mandatory fields validate immediately, with tooltips or error messages that explain what's needed.
Screen placeholderNGS order: step-by-step form with progressWeb · 1600 × 1000
Screen placeholderNGS order: validation & reviewWeb · 1600 × 1000
09
Designing with a multidisciplinary team
Healthcare products need scientific rigour as well as usability, so I worked closely with four groups.
Bioinformatics specialists
Explained the complex data and functionality, so the design matched the scientific rigour accurate analysis needs.
Oncologists & researchers
Gave feedback on usability, which kept the design grounded in what clinicians actually need.
Software developers
Regular discussions on feasibility and technical constraints, so the UI shipped as intended.
Project managers
Kept communication flowing across the team and the project on track for its deadlines.
10
What I learned
Healthcare systems
How test requisitions, clinical trials and data-driven decisions fit together in personalised medicine and oncology workflows.
Accessibility is non-negotiable
Designing for oncologists, genetic counselors and patients means solutions that are intuitive, reliable and adaptable.
Simplicity is key
Where accuracy and speed matter most, streamlined workflows and clear navigation led to better outcomes.
Research and feedback
Stakeholder interviews and usability testing with oncologists and lab technicians let me iterate and validate designs.
The future of health tech
Users shape how these tools evolve, and the work deepened my passion for the meeting point of technology and healthcare.
Empathy and innovation
At the heart of every solution is the human experience. Empathy and innovation have to go hand in hand.