chartER

task
‍‍Design a solution that utilizes AI to support ER doctors in providing quality care to more patients.

problem
High patient volumes in emergency rooms force doctors to rush assessments, increasing the risk of misdiagnosis and compromising the quality of care.

solution
Utilize AI to summarize important patient info and create an interface that allows doctors to access all patient information in a simple and intuitive way.

Context

The founders of ChartER tasked me with designing an AI-powered desktop platform that assists physicians in providing better care to more emergency room patients. This case study explores the problem space and the steps taken to create a tool that improves patient outcomes and optimizes care delivery. Let's dive in ⬇️

The Problem

Emergency rooms often deal with a high volume of patients, leading to significant time spent on logistics and reviewing medical files. This dilemma leaves doctors with less time to interact directly with patients, resulting in rushed assessments and less personalized care.

User Research

In efforts to understand the target users (physicians), I conducted 3 rounds of interviews with various emergency room physicians. This is what I learned:

Emergency room doctors must attend to numerous patients within a short timeframe, leaving them with little time to thoroughly review each patient's file.
Many doctors find it frustrating to navigate multiple screens to access the information they need, consuming valuable time that could be better spent on patient care.
When doctors rush to review patient files, they are more likely to overlook critical medical details, increasing the risk of misdiagnosis.

Actionable Items

After synthesizing my research insights, I came up with these action items to implement as features in my design solution.

Immediately display relevant patient history information to prevent doctors from having to search through multiple pages.
Generate a summary of the ER patient's condition based from their medical history and current triaged symptoms, utilizing artificial intelligence.
Display a follow-up plan that includes recommended workups, such as tests and imaging, to further investigate and address the patient's condition before offering a suggested treatment plan.

Wireframes

Emergency Room Overview

I developed three versions of the 'Patient Dashboard' for an emergency room setting. The dashboard provides an overview of checked-in patients, highlighting their visit reasons and the status of ongoing tests.

First Version: Featured rectangular cards with patient details, but became cluttered and lacked visual symmetry due to varying text lengths.
Second Version: Proposed a visual map of the emergency room, but the scattered layout proved inefficient for quick reference.
Final Version: Incorporated horizontal, color-coded cards for each patient, optimizing space and enhancing both informativeness and visual appeal.

This final design effectively balances usability and aesthetics, improving the user experience for healthcare professionals.

Patient Focus Screen

The patient history screen offers a comprehensive view of a patient's medical history, chief complaint, and suggested treatment plans. Leveraging artificial intelligence, it analyzes patient data to provide actionable recommendations for physicians, eliminating the need to navigate multiple screens. A key feature is the instant access to source details, streamlining the review process and allowing doctors to quickly verify information without disrupting their workflow.

Branding

The vision for this platform was to inspire innovation and usability in a medical context. Stakeholders preferred a clean, trustworthy aesthetic, which we achieved by drawing inspiration from other internal medical AI platforms, as shown in the mood board. The style guide incorporates a color palette of blue and cool-toned neutrals, paired with Merriweather and Lato fonts, evoking modernity and simplicity.

Moodboard

Style Guide

High-Fidelity Solution

These are the finished high-fidelity screens after implementing all feedback from user testing and building a design system.

Reflection

As the sole designer, I faced many challenges and had to resolve them quickly to keep production on track.

The primary challenge was displaying all patient information on a single screen without overwhelming the user, as physicians often struggle to quickly access critical data amid vast amounts of patient information. This required rethinking the information architecture and user interactions to create a more intuitive flow. I tackled this by designing low-fidelity wireframes and conducting multivariate testing with target users, leading to a streamlined solution.

If I had more time, I would prioritize developing a feature to rank patients by the severity of their condition, allowing doctors to quickly identify those needing immediate attention. I would also enhance the live alerts system to display multiple lab results simultaneously, offering a more comprehensive view of each patient's status.

If my work on this project resonates with you and you have an exciting opportunity, feel free to contact me here.

emi sandoval