Radiation Treatment Planner

Deliverables

  • Low-fidelity wireframes

  • High-fidelity mockups

  • User testing prototypes

Roles & Responsibilities

  • UX Designer

  • Lead Business Analyst

Project Facts

  • Duration: 11 months

  • Industry: Healthcare

Skills Used

  • Agile UX

  • Requirements Gathering

  • Client Relationships

Using AI to expand access to healthcare expertise

IT ALL STARTED WITH A VISION...and not much else.
On Day 1, all I got was a word document with the client’s mission:

From: Project Sponsor
To: Luis

Preparing a radiation therapy treatment plan involves two time-consuming tasks:1.) the segmentation of targets and normal tissues on a CT image, and2.) the optimization of the treatment beams. Our goal is to automate these 2 tasks and to offer it as a service, primarily for Low- and Middle-Income countries.
— Laurence

The client, a professor with a team of researchers and medical students, developed Machine Learning algorithms that could automate treatment planning; thus cutting the time it takes to produce a treatment plan from about ~2 days to just ~40 mins. However, this futuristic system was hidden behind outdated design patterns and lived in local servers. This case study summarizes a 14-month effort to create a complex medical webapp from scratch, and make it accessible by radiation therapists all over the world.

Problem Statement

Healthcare Provider Shortage

The world is facing a shortage of doctors and healthcare professionals. By some estimates, we would need as many as 6.4 million physicians and 30+ million nurses to keep up with global demand. This shortage is no different in the discipline of Cancer treatment and unfortunately, it impacts Low & Middle-Income Countries the most.

The Radiation Treatment Planner (RTP for short) program has a noble mission to improve access to Cancer Treatment Planning to potentially save thousands of lives.

The client’s team consisted researchers developing in-house algorithms that could help automate how Cancer treatments are generated. Their system was futuristic in that it used deep learning to automate complex clinical processes. But it was hidden behind old-school design patterns and built on unacessible platforms.

The definition of success going forward would include:

  1. expanding access to their back-end systems from their Hospital Research via a webapp

  2. encrypting Protected Health Information before storing & sharing it between external institutions 

  3. designing a seamless experience while expecting errors in file transfers and algorithmic calculations

  4. exploring how to verify legitimate institutions, and then decide how to charge customers on an ability-to-pay basis (different

  5. rates for developed countries vs LMICs (Low- to Middle-Income Countries)

Outcomes & Results

From Scratch, to Go-Live

In the end, I was able to oversee and support this project through all phases of the SDLC, from ideation all the way through design, development and Go-Live. Here are a few highlights of design iterations that we worked through.

By self-teaching myself design on this project, I performed all the core functions of both a Lead Business Analyst, and a UX Designer. Including specifically, I was able to run bi-weekly Requirements Gathering and Design Review sessions with client Product Owners to:

  • identify 9+ Key Features of the to-be product, and scoped the necessary features for a MVP

  • produce rough estimates of LOE for each key feature

  • 4 different Personas & their permissions needed by different user types

  • preliminary pricing and billing tiers based on user demographics

  • built an entire Agile Requirements Backlog from scratch for back-end offshore devs (eventually growing to 85+ user stories and 150+ change requests)

  • drafted low-fidelity sketches in Figma for UI/Front-End Developers

  • designed an interactive prototype in Sketch with 50+ screens & interactions

  • iterated through 2 major re-designs due to legal & branding guidelines by building high-fidelity prototypes in Sketch

  • clarified Testing Scripts for offshore QA teams based on User Story Acceptance Criteria

  • ran User Acceptance Testing sessions as product was launched through DEV > QA > Production environments

The application is still in use by the same research team, who estimates that their service could realistically help save hundreds of thousands of lives in the next 5-10 years. Since the project ended, the team has continued to test the web-app with partner hospitals by stress testing it to ensure it can support unexpectedly high numbers of patients simultaneously. Their ultimate mission is still to get FDA approval to expand access to the web-app to more institutions in Low- and Middle-Income Countries.

Project Feedback

Previous
Previous

Orbital Oil Sentiment Analysis Dashboard