In my Graduate Education at Carnegie Mellon University, we had been challenged with choosing a client at Pittsburgh to integrate an intelligent voice user interface system into the service.
We chose the University of Pittsburgh Medical Center (UPMC) as our client. Although we only had 6 short weeks to complete the project, we didn't shy away from the challenge of healthcare research.
Our research has provided problem space in Making Appointments, Navigating the Hospital, Access to Hospital Services, and Receiving Test Results. Our final design, Lora, is a smart assistant that uses professional and warm conversations to better connect healthcare services with hospital Patients
To find care, the user could ask Lora to find a physician or provide suggestions on the type of specialty to see.
To better help the users with scheduling, Lora also looks into their calendar to avoid appointment conflicts.
To optimize patients' experience, Lora will inform them about the types of accommodation services UPMC offer.
Because most tests are sent to the patient prior to the physician's note, Lora provides options for the patient to wait for the physician's note or see the results in advance.
Lora helps users to define medical terms and compare results to previous tests to better understand their progress of change.
By upgrading the current kiosks with Lora, patients can experience a check-in process without waiting in long lines at the front desk.
UPMC is an old yet prestigious hospital; however, many of the designs are outdated. Therefore, we did an I/A Audit on their My UPMC app to learn and find insights into the current user flow. We discovered that:
The use of text-dominant GUI 🤯 is forcing the user to go through a lot of reading and redundant steps.
The current digital ecosystem is in bits and pieces 🌀, with many actions that need to be redirected to a browser or another app, due to the lack of app functionality.
Through our on-site interview at UPMC Presbyterian of a physician, a front desk staff, and 8 hospital visitors, we not only received insights into what these professionals think their system and patient service, as well as these visitors' own frustrations and pain points during hospital experiences.
Phone calls to make appt. results in wait times and voice mail (est. 5 min) and finding the right specialties without referral requires knowledge or research.
Near 33% of first-time visitors get lost. The route to the right facilities is confusing and difficult to find the right facility. It also detains injured visitors’ mobility.
Many visitors are not aware of the various courtesy services, like escorts, and interpreters; n risking the quality of care and decreasing patient satisfaction.
A rich insight is that most test results are sent to patients before Doctor's clearance, resulting in a lot of self-misinterpretation of the results.
A streamlined experience would not only make the patient happy but also reduces the workloads of the physicians and staff in the hospital. Most patients' struggles are compensated by additional staff effort.
The features of VUI are the primary scope as it helps to make the app more intuitive for the user to access features. Additionally, the VUI enhances the human aspect, making it easier for users to find the information intuitively, without having to go through a lot of reading or redundant steps in a Graphical User Interface (GUI).
Our team brainstormed features that could address the pain points and tasks identified through our research. We started by mapping out the entire patient experience, from scheduling an appointment to receiving follow-up care. Here are a few of many refined journeys from feature brainstorming:
Journey 1)
During Care Visits, patients may need to navigate to different locations, either the doctor's office or the lab center. This can be achieved through the use of an in-app hospital map feature 🗺️ — powered by VUI.
Journey 2)
Post-Care, patients may need to understand their test results, which could be achieved through features that compare previous test results and interpret medical terms 📖🔍 — with Lora as an AI-language interpreter, to help patients better understand their results and alleviate any confusion that patients may have.
I led the creation of user flows that take into account both current and potential features. By analyzing these flows, the team identified pain points in the current user journey and worked to solve them through redesign, rearrangement, and the addition of new interactions. This process allows us to create intuitive and seamless experiences for our users that make interacting with our products effortlessly and enjoyable.
USER FLOW: Booking an Appointment
1) A patient is experiencing symptoms that they are unsure of what type of doctor they should see. They turn to Lora for help. By simply sharing their symptoms, Lora provides suggestions for doctors who specialize in treating those symptoms. If the patient is still unable to make a decision on an unhappy path, Lora directs them to call the clinic for further assistance.
2) To place importance on timely and efficient healthcare. We redesigned our time scheduling process with the help of Lora. With permission, Lora accesses the patient's calendar to check for conflicts and remember the appointment time, reducing the chance of missed appointments and ensuring a smooth schedule for the clinic.
3) Patients are reminded of the services and accommodations UPMC offers prior to confirming the appointment. It is placed here because it gives the hospital time to prepare for the patient's needs in advance. Additionally, by tracking the patient's usage of these services, if the patient has a tendency not to use them, Lora would stop offering these services in future appointments, making the process more streamlined.
We started wireframing and tested out two types of concepts with VUI. One is the heading-dominate design on the left, and the other is a Conversational UI 💬 on the right. We wanted to see which design would be more intuitive for users to navigate and complete their tasks, went to do an A/B testing among our peers.
The initial wireframe presented a straightforward design but received feedback that it lacked the ability to keep track of past interactions and the interface was less intuitive, without leveraging the full potential of VUI.
Ultimately, I discovered that the Conversational UI design 💬 had a higher success rate, as it provided a more natural human-like interaction. It is also easier for users to see their dialogue history, and past actions and complete their tasks. The end result was a more efficient and streamlined task completion process.
By aiming to make our visual system to be reliable, caring, and calming, yet professional, the team created our new palette consisting of blue and purple.
Considering the symbol as a dynamic logo, I led the design by considering how the shape would be in static and motion. The roundness and gradient of this design have successfully transformed what was an overused iconography into a warm and caring symbol for Lora.
Understanding universal design principles and patient privacy inside the hospital, we looked into prototypes of voice activation for users to achieve their goals. The user can activate Lora by saying "Hey Lora", tapping to talk, or inputting via keyboard.
Understanding universal design principles and patient privacy inside the hospital, we looked into prototypes of voice activation for users to achieve their goals.
After initiating Lora, the App moves to a conversation page. The VUI states are designed to complement GUI patterns on the screen.
Subtle Idle
Motion + Text Feedback
Conversation Window
Processing Cues
The team covered moments from Pre-visit, In-Hospital, and Post-visit. In Pre-visit, booking is utilized by the VUI assistant to optimize the user experience of finding a doctor for a symptom. The user will go through choosing a provider, scheduling, and requesting accommodations in a form of a friendly and casual conversation.
To help patients have a better check-in experience inside the hospital, we introduce using a kiosk to optimize the flow. Understanding universal design principles and patient privacy inside the hospital, we looked into prototypes of voice activation for users to achieve their goals.
To reduce the wait time to check in, patients could simply tap their phones to check-in. This avoids the steps of inputting their names, birthdays, and biometric verification (some feel uncomfortable).
To confirm if there are errors or if information is up to date.
To better guide a patient's hospital experience, Lora kindly reminds the patient of all the scheduled appointments for today, while providing direction to help them navigate to the right office.
The Kiosk could also work as a beacon support or a location pin for special services like an escort. Instead of calling the office and struggling to communicate where you are, patients could request the service directly at the kiosk and wait near the kiosk before the help arrives.
We believe Lora would help UPMC to build a great healthcare ecosystem, it will not only improve patients' experience but also better create less workload for the understaffed healthcare professionals. The current features are designed to focus on the top pain points from our research. Moving forward with more added functionality, Lora could be more comprehensive and we see potential that not just UPMC, but other hospitals could look into how smart AI and VUI could be used to bridge patient experience with healthcare services.
