ULTRASOUND WORKFLOW FOR HEART
Company: Echonous is a medical device startup for miniature ultrasound devices.
Context: Heart is one the hardest anatomy to scan and study as it is constantly in motion. A lot about the human body can be understood by the health of the heart and EF (Ejection fraction: % blood pumped out of left ventricle ). However, since it is not an easy task to ultrasound for non trained clinicians, most primary care physician (PCP)'s either refer to sonographers or to cardiologists only when the patient complains of certain symptoms, which is usually when the disease has reached advanced stages. Point of care ultrasound (POCUS ) is an emerging $2.5 bn industry.
​
Our product aims to enable *all clinicians*, not just cardiologists, to be able to perform initial heart diagnosis that was only possible in labs before. We deliver this capability by combining advancements in AI's pattern recognition and machine learning algorithms, with ease of use in a small form factor.
​
Project Goal: Create a cardiac workflow for a medical practitioner for point of care, who may not be trained to do an ultrasound of the heart.
​
Title: UX/ Usability Manager
​
Time: August 2016 to present
People don’t simply buy products or services, they ‘hire’ them to make progress in specific circumstances. #JobsToBeDone #GuidingPrinciple
Doctors look at EF (Ejection Fraction: the % blood pumped out of left ventricle) to figure out the next steps in diagnosis and treatment of a patient. To get to an EF value traditionally can mean multiple referrals and appointments. I was part of this ambitious goal to accomplish this in less than 3 minutes.
My ROLE & PROCESS
I was hired as the 9th person in the start-up to be the UX lead. My goals were -
-
Understand the ultrasound devices and their unique knobology
-
Read relevant clinical journals to educate myself in the field of heart ECHOs, enrich my vocabulary and understand how EF is calculated traditionally
-
Talk to various doctors and sonographers to understand their pain points in their process of not just calculating EF but examining a patient
-
Design for the various clinical cohorts from nurses, EMTs, MDs, and specialists who may or may not be trained to do ECHO
-
Test initial prototypes, iterate on the design, collaborate with the software development team to get it built
-
Perform formal usability studies as described by IEC 62366 (Usability of Medical Devices)
STEPS IN CALCULATING HEART'S EF
American Society of Echocardiography's recommended method uses Simpson's biplane method.
I broke the steps down to challenges and opportunities.
GET A GOOD VIEW OF THE HEART
Challenge - Finding the right view of the heart can be the hardest step for the novice user.
Opportunity - Show user a tutorial. Use AI to guide the user in probe placement. Give feedback to the user based on the view.
MANUALLY IDENTIFY ED AND ES FRAMES
Challenge - If the image quality is poor, identifying ED (end-diastole) and ES (end-systole) frames can be very hard.
Opportunity - Use ECG signals to determine ED frame.
DRAW CONTOURS ON THE SELECTED FRAMES
Challenge - This is the most time-consuming. In traditional devices, the sonographer has to adjust the contour based on the underlying heart anatomy.
Opportunity - Use AI algorithms to draw the contours.
​
CALCULATE EF
Challenge - What might an EF value mean? Is it in normal range?
Opportunity - Show the user the averages of normal population.
PERSONAS
I created 4 personas differing on their experiences and environments. Listing out their needs, frustrations and different use environments helped our team stay empathetic and think about them when making key decisions.
Validating our vision
I created a short video of our vision and showed it to sonographers, doctors, and nurses to get their feedback.
​
Major takeaway was that showing tutorials in the same imaging window is a bad idea because it makes the patient lose trust in the medical provider​. I adapted, I replaced the tutorial with reference images which are very helpful for the novice user.
​
Voice credit - Simon Kim
PAPER PROTOTYPES
 |  |  |
|---|---|---|
 |  |
HIGH FIDELITY PROTOTYPES
EXHAUSTIVE STATES FOR ALL SCREEN
The engineering team was very grateful for the exhaustive versions of the screens based on the state of the device as it increased their efficiency and time to implement.
PRIORITIZING AND ADAPTING TO CHANGING REQUIREMENTS
VALIDATION TOOL FOR EF VALUES
A medical device needs to be submitted to the FDA and must go through clinical validation testing to verify the calculated EF values. Our original plan was to use the existing tools to measure EF values but the ones that we found were neither compatible with the RAW format of our images, nor user-friendly. So we had to quickly come up with designs for a validation tool for clinicians.
Time: 4 weeks.
CAN THE CLINICIAN MAKE CHANGES TO THE LV CONTOUR?
As a risk mitigation item, the device must allow the user to override the segmentation of the left ventricle. Our device may not have knowledge of other anatomical specifics of the patient so a seasoned clinician is allowed to make necessary edits. I collaborated with my team to create UIs for editing of the AI segmentation.
GUIDING USER ON PROBE PLACEMENT WILL INCREASE THE DEVELOPMENT TIME
We prioritized the left-ventricle edit feature to ensure that we earn our users trust and let them apply their contextual knowledge for the best patient outcomes. As much as we were excited about other AI-related features that were already planned, we knew the above benefits justified deprioritizing them for the current phase in favor of this one.
USABILITY STUDIES
Formative and Summative Study
​
Goals:
-
Evaluation of the UI
-
Evaluation of the screen with gloves and gel on
-
Long fingernails, 60+ age groups were also tested
​
Results:
The results were very promising. The device was usable even with gel and gloves on. Long fingernails with gloves on did not work very well as anticipated since the air was trapped between the glove the nail and therefore capacitive touch was not registered. This would be addressed by adding a section in instructions for use. All UI related-use errors discovered were addressed.
Product Launched!
​​
-
Over 5000 devices sold since launch.
-
The device will be featured in Dr. Richard Hoppman's book. He is a pioneer in ultrasound education and teaches at the University of South Carolina.
-
Android app for Samsung Tablet is also launched. IOS app is in the works.