NaviGAIT
Navigating the Future of Preventative Care for Runners
Matt Benik, Jack Berkowitz, Trey Blystone, Thomas Eckrich, Benjamin Esquieres, Natasha Mody, Claire Olmsted
Motivation
Each year, 65% of runners in the United States, or 40 million runners, experience a running-related injury. An examination known as gait analysis, which is currently used to study the runners' biomechanics, diagnose injuries, and increase the efficiency of running, could also preemptively prevent many of these running-related injuries. Just as routine dental checkups help prevent cavities and gum disease, regular gait analysis could help prevent injury. However, gait analysis is too inefficient to be conducted often enough as a means of preventative care.
65% of runners in the United States get injured each year
NaviGAIT in Action!
Problem Discovery
We have had the opportunity to work closely with Dr. Angelini, a physical therapist at UPMC Sports Medicine who performs gait analysis. Currently, her process, along with that of several other clinicians we’ve spoken with, is plagued with technical limitations, a tedious setup, and inefficient analysis procedures. This leads to appointments that can take upwards of an hour, ending with clinicians having insufficient time to explain results to patients.
Thus, as it stands gait analysis...
- takes too long
- uses too many devices
- is not used for preventative care
Objectives
With these problems in mind, we asked ourselves how might we decrease the time of gait analysis so that it can be conducted frequently and proactively with the potential to ultimately improve preventative care for runners?
1 Device
Accurate
Efficient
Human-Centered Design
Consulted 7 Experts
9+ Iterations
Tested Design with 4 Clinicians
Tested at 1 Clinic (UMPC Sports Medicine)
Our Solution
For this project, we focused on the two most time consuming portions of gait analysis, the setup and quantitative measurements made clinicians. The prototypes and process we developed to target these portions of the analysis reduce the setup time from 5 minutes to 2 minutes, and the time to make measurements from 15 minutes to 5 minutes.
The Process
1. Clinicians start the setup process by attaching bands with codes on them to a patient’s legs and hips. They can then set up their tablet on our custom stand where everything a clinician needs for their analysis is conveniently stored.
2. As the patient runs on the treadmill, the clinician uses our mobile stand to seamlessly film the patient running from multiple angles.
3. Clinicians upload the video of their patients running on the treadmill to our website.
4. Our software will detect the codes on the patient and make automatic measurements that are crucial to a clinician’s analysis. This software eliminates the need for clinicians to spend valuable time taking these tedious measurements manually.
5. Finally, clinicians use the data output from our software to provide their patients with a diagnosis and treatment.
How the Software Works
Our software uses computer vision to detect the codes on the bands on the runner and make automatic measurements of various angles using the codes' position in each frame of the video. The codes on the bands are known as ArUco codes and are generated using the OpenCV library. Likewise, the computer vision and angle measurements are implemented in Python using OpenCV. Uploading raw footage and playing processed videos is handled by a Nodejs endpoint. As shown in the diagram, each of these processes is deployed in a Docker container on an AWS EC2 instance. The web app frontend was written using React and has its own separate backend, again running on Nodejs, used to communicate with a MongoDB database. The endpoint that handles uploading and playing videos also uses the web app backend to communicate with the database. The web app code was deployed on Heroku.
The Mobile Stand
UPMC's past stand set-up. It was inconvenient, not easily movable, and had no storage.
We built a custom stand for easier and more convenient storage. The clinician is able to record patients from multiple angles easily. The stand reduces set up time.
Our new stand design.
Outreach and Testing
We visited Dr. Angelini's clinic throughout the development of the project to learn more about the problem at hand as well as gain feedback from our various iterations. She also brought in 3 other clinicians as well as Dr. Onishi to provide additional feedback. From our testing, we completely pivoted from focusing on the patient-understanding aspect to the clinician process aspect because the feedback was that it is too hard to control.
After the pivot, we tested multiple iterations of clinician-end solutions and refined our designs as Dr. Angelini and the other clinicians suggested.
Impact and Results
Looking Ahead
We hope that once the gait analysis process becomes shorter, it can be conducted more often by clinician. As the process becomes easier to conduct, it could become a standard baseline test conducted on individuals and help catch injuries before they occur. With this in mind, clinicians can take steps to help runners improve form and prevent injuries.
We want to make our stand more robust, reduce the size of the codes, and increase the accuracy of our web interface.