Digital Emergency Service
CHATBOT & TELEMEDICINE SOLUTION
Digital emergency service with a chatbot that performed a medical triage, a videocall tool and an administration module.
ROLE
UX, UI, Writing, Research, Product Design
DATE
2023
TOOLS
figma, miro, maze
DESIGN CHALLENGE
Research, design and test a product seeking to improve the experience for both patients, doctors and administrators.
DELIVERABLES
hi-fi prototype, design system, UX Research, Usability Testing
UX DESIGN / UI DESIGN / RESEARCH / UX WRITING
context & topic
The first and most important information we started with was that the majority of health situations were categorized as low risk according to the Manchester Triage System.
Another important information is that, during the pandemic, visits to the emergency centers were down 77%. Doctors pointed out the patients' fear of contamination as the main cause.
80%
of the people showing up at the emergencies are categorized as low risk
How can we safely help prevent the emergency centers from being overcrowded after the pandemic?
PAD is a whitelabel digital emergency service for low risk cases
The service consisted of 3 main areas:
​
1. Chatbot - a virtual assistant that would perform a medical screening and define if the patient's case is high risk or low risk.
2. Doctor's area - where the doctor could see the line of patients waiting for an online meeting, visualize their information and answer the calls.
3. Administrative area - reports and other features to manage healthcare companies, patients and doctors.
low risk
patient
chatbot triage
high risk
data
When I joined the project, I collected the information on the development of the product up to that moment:
-
The chatbot had already reached a safety percentage for the triages.
-
The chatbot could improve the quality of the triages
95%
Accuracy to human triage
1557
patients tested
104
flows
in total
We started with a CSD matrix that stated our certainties, suppositions and doubts.
qualitative research
After our chatbot triage system started reaching 95% accuracy compared with human triages, we decided to collect qualitative data to dig deeper in the experience.
We had 4 interviews conducted by our UX researcher. I've reviewed the recordings to get insights and later discussed them with the team.
personas
As an emergency service for health insurance companies, PAD had a large range of potential users that vary a lot in demographic information. Based on the data collected, 5 different personas were identified in order to better understand our users and their needs.
Alana, 34
Mother of a 1 year-old
Unimax plan
Global/Business
Lives in Porto Alegre
Marcela, 32
Unimax plan
Global/Business
Lives in Porto Alegre
Breno, 18
Gen Z
Unimax plan
Global/Business
Lives in Canoas
Fábio, 35
Gen Z
Unifácil plan
Outpatient/Business
Lives in Gravataí
Jussara, 72
Unipart plan
Global/Family
Lives in Porto Alegre
the chatbot
The styleguide was based on the first healthcare company that PAD was being tested.
After defining the styleguide, we've designed and developed basic components for the chatbot and screens that would display the behavior of the chat.
Chatbot desktop and mobile versions
Doctor's area
A small library was created for basic components needed for the doctor's area, which included the Patients waiting line, patients information screen and the video area. Here are some of the screens:
doctor's view of the patients waiting line, desktop and mobile version
doctor's mobile view and flow
takeaways
Working within the healthcare area made me aware of the complexity of making this kind of technology accessible for widely different people. The variety of personas (patients, doctors, administrators) and stakeholders was a great challenge, but also strenghtened my research and communication skills.
​
The agile startup environment was an amazing learning and made possible the quick development and release of the product. This also required a faster-paced ux process, so we could keep improving the product as we tirelessly tested with our user and collected feedback.
​
Our release was successful and the product is online for 2 healthcare companies and being used by dozens of doctors and thousands of patients to this day.