Project Overview

MY ROLE

I worked on this cross-functional project involving the solutions management, clinical, engineering, and design teams. This was a high priority project that was a part of a larger effort to create an improved version of our existing referral management product.

Over a 4 week period, the design team was tasked with conducting a UX review of existing AI tools, attending meetings with our PM to understand constraints and acceptance criteria, and lastly developing high-fidelity prototype.

PROBLEM SPACE

Today, users will read through the electronic referral packet to find pertinent information, wasting time they could spend replying to other referrals.

Due to the manual nature of this task, users waste valuable time looking for information on the clinical appropriateness of the patient and often times find they cannot take the patient. This laborious process results in the users’ facility not being able to review other referrals that come in, resulting in them losing potential patients.

Hypothesis

By providing the most relevant information through the clinical summary...

1. Healthcare facilities will reply faster and more accurately to referrals
2. Less communication will be required between healthcare facilities and Discharge Planners
3. There will be a decrease in the length of hospital stays due to healthcare facilities being able to respond to referrals quicker

Design Exploration

Next, I moved onto design exploration to see how we could add the clinical summary to the our patient referral dashboard. The screen below is our V1 MLP (Minimum Lovable Product) ❤️.

CLINICAL SUMMARY WIDGET

I explored how the “Referral Summary” (previously called “Key Questions”) widget could be added to the design layout of patient referral dashboard. At this point of the design process the acceptance criteria was not yet finalized. After showing this mockup to the clinical team we iterated the design based on the feedback we collected.

Solution

After the design exploration, I continued to work with my team to build out a high-fidelity prototype for our clients to navigate through during our discovery calls. Charles Mastin, the teams’ Principal UX Designer, referenced my Figma mockups to develop this html prototype.

Our solution provides an intuitive way for users to choose between one-time add and subscription by highlighting one payment option at a time in the modal.

Solution Components

1. Complete a “checklist” of what is included in the packet. Intake coordinators must determine if the packet contains certain information such as an H&P in the last 12 months, or a negative Hepatitis test in the last 60 days. Based on a set list of clinical documentation, we will utilize AI to indicate if these are included in the packet. This lets the Intake coordinator know what additional information to request up front (in a future version will automatically alert the DP of this).
2. Summarize/answer key questions from what is in the packet. We will utilize AI to answer pre-defined clinical questions based on information contained in the referral packet with a link back to the section of the packet that contains the actual answer
3. Create a table of contents for referral packets to help Intake Coordinators find the key clinical information. Utilize AI to create table of contents and organize those sections by date so the discharge planner can easily navigate to the areas of the packet most useful in their decision making process.

Outcomes

The engineering team has completed 39 out of 105 JIRA tasks for the CRI project. The work for Clinical Summary is currently on hold as there are still acceptance criteria being worked out between product and the engineering team. The release date for WCRI is not set as of 11/03/2024, but the projected date is tentatively during the summer of 2025.

Takeaways

CRI is a large project that has proven to be more complex than we initially predicted. The clinical summary AI tool was added as way to incite potential and existing clients to move over to our new and improved referral management solution. The design process was complex as it challenged me to interpret user feedback into design iterations & enhancements. I had to be aware of what feedback could be bucketed into our next design phase and what enhancements could be added to our MLP (Minimum Lovable Product). As of 11/03/2024, I am continuing to support engineers and product managers as they build out this MLP solution.