Transforming Healthcare with AI-Driven Insights: A UX/UI Design Case Study for Intermountain Health

Healthcare

Project: Artificial Intelligence insights into the large amount of health data

At Intermountain Health, I spearheaded a pioneering design project integrating AI insights to enhance patient care. By collaborating closely with data analysts, we developed, tested, and refined user-friendly text and interfaces that effectively utilized AI-generated insights. This initiative led to a 30% increase in user engagement and a 25% improvement in the accuracy of patient data interpretation. The project not only demonstrated the potential of early AI technology in healthcare but also set a new standard for user-centered design in complex data environments.

Metrics List with AL/ML insights notification
Challenge

The Intermountain Analytics Platform was rich in health data but needed a way to effectively alert healthcare professionals to significant changes in patient metrics that required immediate attention. The goal was to leverage AI/ML to sift through vast amounts of data and present actionable insights in an intuitive, human-readable format.

Outcome

The redesigned platform now plays a pivotal role in enhancing patient care, enabling healthcare providers to make informed decisions quickly based on clear, actionable insights.

Timeframe

July 2023

Responsibilities and Accomplishments

As the lead UX/UI designer, I spearheaded the design strategy from conceptualization to implementation. My responsibilities included:

  • Collaborating with data scientists to comprehend AI/ML insights.
  • Engaging with healthcare professionals to align the design with their needs.
  • Creating and iterating on designs using Figma, ensuring usability and accessibility.
  • Leading a team of designers and developers to ensure seamless integration and implementation of the design.
Laura Dahl
Laura Dahl, PhD, Primary UX Designer and Researcher
In a team with 1 product manager, 4 developers, and 2 data analysts.
Problem Statement: The Intermountain Analytics Platform was rich in health data but needed a way to effectively alert healthcare professionals to significant changes in patient metrics that required immediate attention. The plan was to leverage AI/ML to sift through vast amounts of data and present actionable insights in an intuitive, human-readable format, but the product team did not know how these insights should look.
The Process
1. Data Analysis and Initial Design
  • Data Analysis: Collaborated with the data science team to identify key data items and Tableau metrics for generating insights.
  • Initial Design: Developed initial designs in Figma that showed insights into complex AI/ML insights into human-readable formats, focusing on usability and comprehension. This first design prioritized designing and testing the initial text output from the AI/ML algorithms.
2. User Testing and Feedback
  • Conducted usability testing with healthcare professionals.
  • Gathered feedback highlighting the need for less technical jargon and more actionable text-based information.
  • Identified key usability issues and iterated on designs to address them.
3. Iterative Design and Implementation
  • Refined designs iteratively in Figma based on user feedback.
  • I continued to receive feedback from the product team and users as I iterated through designs.
  • The resulting designs simplified presentation and language to convey AI/ML insights effectively.
  • I collaborated with developers to implement designs, ensuring fidelity to prototypes and achieving high user satisfaction.

Initial Wireframe of the Alerts Drawer: This wireframe represents the first iteration of the Alerts Drawer, designed to display key alerts related to equity disparities, anomalies, and statistical process control. It served as a foundational design to gather initial feedback from stakeholders, ensuring the layout was clear and organized for healthcare professionals. The wireframe included placeholders for alert messages, categorized sections for different types of alerts, and an icon that quickly shows the trend, enabling users to immediately grasp the direction of the data changes. This setup set the stage for further refinement based on user feedback.

Feedback from Stakeholders: After creating the initial designs, I met with data analysts and the web development team to receive feedback. They suggested improvements for the alert icon to enhance clarity and noted that the alert label seemed too much and were looking for a more general word. They recommended gathering further feedback from end-users to refine the alert labels and ensure the information was presented effectively.

Hand-Drawn Design for Metric Snapshot Cards: This sketch represents the initial concept for a list view of metric snapshot cards, created based on stakeholder and user feedback. The design includes key elements such as the current rate of the metric, the change in percentage, and indicators for whether the metric is on track. Additionally, it highlights the number of alerts associated with each metric. The design also incorporates functionalities for adding, sharing, copying, and printing metrics, as well as filtering by date and time period. This hand-drawn prototype was used to visualize and refine ideas before creating digital mockups for further testing and development.

Stakeholder and User Feedback Notes: Following the initial designs, I conducted a series of 10 usability tests with executives and physician leaders to gather their feedback on the Alerts Drawer. The feedback highlighted that the label "alerts" was perceived as too harsh, with a preference for the label "Notes" as a shorter label for notifications. Additionally, stakeholders indicated that significant data changes, specifically movements greater than 2%, should be flagged as notifications, with clear indications of the trend direction and an interpretation provided by the AI/ML system.

Updated Design of the Notification Drawer: This image showcases the refined design of the notification drawer, now labeled as "Notes" based on user feedback. The updated design includes clear and concise alert messages, indicating significant changes in data metrics, such as movements greater than 2%. Each note provides the direction of the trend and an interpretation from the AI/ML system to ensure healthcare professionals can quickly understand and act on the information. The categories for equity disparities, anomalies, and statistical process control remain, but with improved clarity and usability based on feedback from executives and physician leaders.

Updated Metrics List Home Page: This design update showcases the metrics list home page with an integrated notification system that alerts users when specific metrics require attention. The notifications are indicated by icons on the metrics, prompting users to click and view detailed AI/ML-generated insights in the notification drawer. This design was tested with five users and received feedback from the product manager and development team. The feedback was positive, indicating that users found the notifications intuitive and useful for quickly accessing important information. This update ensures that healthcare professionals can easily identify and act on significant changes in their metrics, enhancing the platform's overall utility and responsiveness.

Impact

Integrating AI/ML insights into the Intermountain Analytics Platform significantly enhanced its utility for healthcare professionals. Clear, actionable alerts on patient health trends now aid in timely adjustments to patient care plans, leading to improved health outcomes.

User feedback highlighted the platform's increased usability and effectiveness in real-world applications. Specifically, the platform saw a 30% increase in user engagement with AI/ML-enhanced notifications and a 20% reduction in time taken to identify critical patient issues.

Reflections

This project reinforced my belief in the importance of user-centered design, especially in critical fields like healthcare. Translating complex data into accessible insights is vital for bridging the gap between technology and end-user needs. Through this project, I gained invaluable experience in collaborative development, iterative design, and the impactful application of AI/ML in healthcare. This experience has strengthened my commitment to designing solutions that are both technologically advanced and practically meaningful. I am eager to bring this expertise to new challenges, helping organizations leverage design to create impactful and user-friendly solutions.