SmartRate: A Machine Learning Approach to Predicting Cardiac Arrest
CSEF · 2018 Computational Systems & Analysis Honorable_mention Award
Overview
Objectives/Goals The goal of this project is to prevent people afflicted by cardiac arrest from dying, individuals whose lives are significantly shortened every year simply because they did not arrive at the hospital in time. This is an issue that has been plaguing society since the advent of modern medicine, and yet is has not yet been solved- until now, with an inexpensive but functional wearable. The prototype features three distinct and effectual algorithms: one for the wristband itself- to calculate heart rate accurately regardless of movement, one for the iOS application- to recognize cardiac arrhythmias and autonomously notify EMS, and one for the remote server- developed with machine learning to predict cardiac events before they occur. Methods/Materials To develop the prevention apparatus, an Adafruit based Atmel microcontroller was utilized, coded with the C/C++ compiler in Arduino. The sensor features a photoplethysmographic module, converting reflected light into signal values to measure heart rate. The iOS application was written in Xcode using the Swift language, and the server was coded in Python and hosted remotely via Django. It was tested with a dynamic and local database containing 300 electrocardiogram samples from MIT, deployed on the server. Results This project has proved incredibly accurate- the wristband has 98.5% accuracy compared to a 12-lead electrocardiogram, the application recognizes the abnormal heart rate and notifies EMS 100% of the time, and the innovative machine learning predictive algorithm has an accuracy of 90-95%. Conclusions/Discussion The cumulative device, which is a combination of three distinct and effective algorithms integrated within an inexpensive wearable that communicates via Bluetooth with an iPhone, creates the possibility of reducing excess funding within hospitals and diminishing diagnosis time for arrhythmias from days to seconds, thereby drastically decreasing Emergency Medical Service time from hospital to victim.
Summary statement
We developed a wristband, an iOS application, and a remote server- utilizing machine learning to save the life of the user by predicting cardiac abnormalities and therefore preempting cardiac arrest.
Help received
In addition to utilizing the MIT arrhythmia database, Dr. Nan Wang at CSU Fresno provided guidance in understanding machine learning classification functions.
Awards (1)
- Honorable Mention
Competition history
- CSEF 2018
Resources
Related projects
CSEF · 2017
SmartRate: An Innovation in Early Detection, Warning, and Prevention of Cardiac Death
ISEF · 2018
SmartRate: A Machine Learning Approach to Predicting Cardiac Arrest
ISEF · 2017
SmartRate: An Innovation in Automated Early Warning, Detection, and Prevention of Cardiac Death
CSEF · 2017
A Cloud Based Machine Learning Wearable Device to Monitor, Predict, and Prevent Cardiac Arrest
CSEF · 2016
The Use of Photoplethysmography (PPG) to Develop a Prototype Heart Rate Wristband to Monitor Life Threatening Arrhythmia
ISEF · 2019
Engineering a Novel Wearable Biosensing Mechanism through the Implementation of Microelectromechanical Systems and Machine Learning to Realize Anomalies Hinting towards Future Cardiac Episodes
CSEF · 2017
Combatting Cardiac Arrest through a Compact and Portable Heart Rate Monitor and Analyzer
CSEF · 2018
iCordisX: SmartPhone-Based Personalized Cardiac Monitoring Using Computer Vision and Bluetooth Low Energy
Closest projects by meaning, across every fair and year in the corpus.
Browse more like this
Source: California Science & Engineering Fair public projects