Objective
remotely assessing fall risk without requiring expensive hardware
Outcome
an at-home functional mobility assessment performed via a mobile app (iOS)
Launch on the App Store and pilot study with 50 patient at Stanford Hospital
remotely assessing fall risk without requiring expensive hardware
Outcome
an at-home functional mobility assessment performed via a mobile app (iOS)
Launch on the App Store and pilot study with 50 patient at Stanford Hospital
During winter 2021/22 I participated in the Stanford Byers Center for Biodesign course CS342 / MED253 Building for Digital Health. Over the course of the quarter I worked with physicians from the Stanford Health Care Department for Emergency Medicine to build a user-friendly app using Stanford’s Cardinal Kit to automate at-home safe functional mobility assessments and ultimately build predictive models using at-home data. Apple's open source frameworks ResearchKit and CareKit were used to realize the iOS research app called GaitMate.
Project Roadmap
Several features were implemented over the course of 10 weeks. After having included an informed consent process using ResearchKit, we started to track day-to-day adherence to a schedule including weekly and monthly tasks for the study participant using CareKit. In addition to tasks showing up in the schedule, the user is reminded via push notifications to complete them. Next the Report a Fall feature was included which allows the study participant to report any falls in the past week. After the midterm presentation the app's most important task, the assessment of the gait was realized using CoreMotion to access accelerometer data collected by the phone's sensor. The gait task records the study participant's gait data for a short walking task during which the iPhone is placed in a pouch around the participant's waist. The final feature implemented enabled fall risk prediction using a pre-trained gait identification PyTorch machine learning model.
Project Architecture
The frontend was realized using Apple SwiftUI and the Apple frameworks CareKit and ResearchKit, as well as CoreMotion for accelerometer data collection. To be able to store not only the gait, but also the survey data collected via the frontend an integration with Google Firebase was realized.
Walkthrough of the finished app
Pilot Study
The app was launched on the App Store in November 2022. Dr. Brian Suffoletto (Associate Professor - Stanford Emergency Medicine) and Dr. David Kim (Assistant Professor - Stanford Emergency Medicine) enrolled 50 patients for a pilot study. The participants are patients who were discharged from the Emergency Department at Stanford and they will use the app over the course of 3 months to gather data on their fall risk.
User Flow Diagram - Sketching
UI/UX Design - SwiftUI
App Development - Xcode, Swift
Product Management - Github
