ECIVI: AI-Powered (Electrocardiogram) EKG Interpretation
Empowering healthcare professionals with instant cardiac analysis
Introduction
ECIVI transforms how healthcare professionals interpret EKGs—delivering instant AI-powered analysis with the educational depth of a cardiology consultation, right from a smartphone. In clinical settings where every second counts, healthcare providers need immediate access to accurate cardiac assessments. ECIVI bridges the gap between machine interpretation and specialist expertise by combining computer vision, artificial intelligence, and collaborative features into a mobile-first experience that builds clinical confidence while saving time.
My Role
I led the design, user testing and development of this project from end to end with a 4 member team—from surveying 31 healthcare professionals across multiple specialties to delivering tested, high-fidelity prototypes ready for development.
Problem
Healthcare professionals frequently interpret EKGs (61% do so regularly) but lack confidence (58% express uncertainty) and rely on fragmented resources. This creates delays in critical decision-making and limits learning opportunities at the point of care.
Timeline
3 months
Tools
Figma, Google forms, Figjam, Photoshop
Platform
Mobile Application
Responsibilities
UX/UI Design, UX Research, User Testing
Project Goals
The objective of this project was to design a mobile AI-powered solution that empowers healthcare professionals to interpret EKGs with greater confidence, speed, and accuracy while maintaining patient privacy and enabling collaborative decision-making.
Design Process
My task was to design the end-to-end mobile experience for capturing, interpreting, and managing EKG assessments at the point of care. I began with the assumption that healthcare professionals frequently encounter EKGs but lack immediate access to expert interpretation, leading to delayed decisions and reliance on multiple disconnected resources. The design framework I followed validated this assumption through mixed-methods research—surveying 31 healthcare professionals revealed that 58% lacked confidence despite 87% having formal training, and users relied on 5+ different tools for a single interpretation. I then shaped the solution through defining key user workflows (capture, interpret, collaborate), creating personas based on real user data, and testing iterative prototypes with 8 healthcare professionals to refine features like AI confidence indicators, privacy masking, and visual highlighting of concerning areas.
Research
Understanding Real-World EKG Interpretation Challenges
I distributed an online survey via Google Forms to healthcare professionals across multiple locations and specialties. The survey collected both demographic data and behavioral insights about EKG interpretation practices.
During this stage, I surveyed 31 healthcare professionals to deeply understand their EKG interpretation challenges and workflow frustrations. By collecting data from users ranging in experience from "just 2 days of basic training" to "years of residency and continuing medical education", I was able to validate assumptions about the confidence gap and uncover what healthcare providers truly need when making critical cardiac assessments. The diverse participant pool—spanning RNs, physicians, and specialists across the United States, Haiti, and Ghana—revealed that despite different specialties and settings(like hospitals, clinics, emergency departments, rural facilities, etc.), all shared common pain points: fragmented resources, distrust of current auto-interpretation tools, and the need for faster, more transparent solutions.
Key User Research Insights
Accuracy & Confidence: The value of EKG interpretation tools is for healthcare professionals to make accurate cardiac assessments and gain confidence in critical decisions.
Transparency Builds Trust: There is peace in transparent technology - users tend to distrust “black box” systems and seek AI that explains its reasoning with confidence indicators and visual highlighting.
Speed & Education: Users value speed and educational depth in interpretation tools, from capturing the image to understanding the clinical rationale behind findings.
Fragmented Workflow: Fragmented resources and lack of explainability are common pain points - causes users to consult multiple sources (machines, colleagues, textbooks, online references) and still feel uncertain.
Target Users
Who is the Target User?
Based on patterns in user behavior and common traits from the survey responses, I created personas to help focus upcoming design efforts. Keeping in mind ECIVI's goal of serving front-line healthcare professionals who frequently encounter EKGs but lack immediate specialist access, these personas represent the needs and behaviors of users from the most prominent demographics in my research: registered nurses (45% of respondents), internal medicine physicians (19%), and early-career providers with limited training. Each persona addresses distinct use cases while sharing core needs for transparency, education, and efficient collaboration.
Defining Strategy
How might healthcare professionals capture, interpret, and manage EKG assessments in their workflow?
Given that ECIVI aims to consolidate fragmented resources into one comprehensive platform, it was crucial that users could seamlessly move through core workflows—from capturing an EKG to receiving interpretation, understanding the rationale, and collaborating with colleagues—without cognitive overload or workflow disruption.
In order to organize the information architecture and prioritize features, I conducted user flow mapping and feature prioritization exercises. I analyzed the survey data to identify which tasks were most frequent (interpretation frequency), which caused the most friction (confidence gaps, fragmented resources), and which were most critical (time-sensitive cardiac decisions).
Surprising Research Findings
There was user consensus only on the need for instant interpretation speed and HIPAA compliance—all other priorities varied significantly by specialty and experience level
There was confusion over whether AI confidence indicators should be expressed as percentages, color codes, or plain language descriptors like “High/Medium/Low confidence”
Participants would debate over whether certain EKG findings required immediate action or could wait for routine follow-up, revealing varying risk tolerance and practice patterns
Defining “educational rationale” was a challenge—some users wanted brief explanations while others wanted comprehensive guideline references and differential diagnoses
Design & Prototype
Design ideation
Based on the organization and features defined during strategizing, I began to construct the user experience flow in sketches and wireframes. During design, I focused on fleshing out the following:
- Capture experience for photographing EKG strips with automatic detection and one-tap privacy masking
- Interpretation flow for both quick scanning (visual highlights and summary) and detailed analysis (clinical rationale and guidelines)
- Ability to save and retrieve previous EKG interpretations with notes and tags
- Collaboration workflow for sharing cases with colleagues and receiving second opinions
Testing & Iteration
Testing usability for EKG interpretation workflow
A prototype was tested with 8 healthcare professionals to identify critical usability issues. The main goals were to validate core workflows, identify bottlenecks, and ensure the app worked for users across experience levels.
- Highlighted areas weren't prominent enough—users missed circled regions during quick scans
- Urgency levels lacked visual hierarchy—couldn't quickly distinguish immediate from non-urgent actions
- Confidence indicators were overlooked—users went straight to text without noticing the trust-building features
Priority revisions made:
- Enhanced highlighting with bolder markers, color coding, and direct labels on the image
- Redesigned urgency system with prominent badges and color-coded backgrounds
- Elevated confidence indicators to primary position with plain language
- Added voice-to-text notes and push notifications based on tester requests
These changes resulted in 100% of users immediately recognizing critical findings and 35% faster task completion in follow-up testing.
Next Steps & Future Roadmap
Following UI revisions, the next steps in the project involve updating the high-fidelity prototype and conducting additional usability tests with a larger sample of healthcare professionals to verify these improvements. Given additional time, more features would be designed and tested, such as offline mode for resource-limited settings, integration with EHR systems (Epic, Cerner), and advanced teaching modules with quizzes for continuing medical education credits.
To move toward implementation, designs and documentation would be handed off to the development team along with AI model specifications, HIPAA compliance requirements, and a phased rollout plan starting with pilot programs at 2-3 healthcare institutions for real-world validation.