How to design a health emergency alert device using MPU6050

Not long ago, when I was browsing the news, I came across such a news: a 90-year-old elderly woman likes to wander around at night very much, but at the same time, her health is getting worse at the age of 90. Just recently, she accidentally fell to the ground while walking at night, and her family spent a night and finally called the police to find her.

I shared this story with my friends from the same team at LAHacks last weekend. After brainstorming, we thought, why not design a wearable device that can detect health emergencies in an elderly wearer and alert his or her family and friends? That’s how Project Cura came about.

Research

In my research, I found that many older adults have common concerns about health emergencies. The two main emergencies are falls and arrhythmias (unusual heart rhythms), which are caused by physical and organ deterioration.

According to the CDC, approximately 36 million seniors fall each year, resulting in 3 million emergency department visits and more than 32,000 deaths, while 70% of seniors have high blood pressure, which can lead to stroke, heart failure, and even sudden death, according to NHANES.

In addition, a study by BMC Emergency Medicine showed that 60 minutes after trauma is the maximum chance of survival if medical care is given. This proved our idea that we could save the lives of the elderly by reporting their health emergencies promptly, so we started building it.

Build

We use MPU6050 accelerometer, pulse sensor, ESP-826612-eWiFi module, button, LED bulb and 3.7V battery. MPU6050 and pulse sensor need to be soldered first.

Detecting falls

We use an accelerometer to detect falls because falling is a "lower version" of free falling, giving us a smaller acceleration due to gravity than usual (G < 9.8 - fall threshold). Some projects use the DPS310 to detect falls by measuring the wearer's drop in height, but we found this to be inaccurate (e.g. going down a slope or stairs) and not as intuitive as the acceleration due to gravity.

We made fall detection more accurate and avoided false alarms by incorporating code from this project. The accelerometer must receive a lower G during a fall, a reverse spike in G upon impact with the ground, and no change in G for a period of time, indicating that the user has fallen and is stationary on the ground.

Detecting arrhythmias

We use a pulse sensor to measure the wearer's heart rate. Research results show that a heart rate above 200 or below 27 beats per minute (BPM) is life-threatening. Considering that the wearer may be exercising or sleeping, a heart rate in this dangerous range is abnormal regardless of the wearer's exercise state.

Emergency Button

Since stroke and heart attack emergencies may not have noticeable heart rate changes, we added a panic button to allow the wearer to send an alert directly if they feel unwell.

When any of the three above trigger an emergency, the emergency LED bulb illuminates. The light indicates to potential people around the wearer that they are in an emergency situation, not to be confused with the belief that the wearer is sleeping or simply lying down.

Send SMS reminder with location

Once an emergency is triggered, the mainboard outputs a HIGH pin to the ESP8266, which then sends a text message to the wearer's emergency contacts informing them of the wearer's location.

Twilio SMS API

Twilio supports sending SMS over WiFi connection using ESP8266. We followed this document to implement this feature.

UnwiredLab Geolocation API

Through research, we found that the WiFi Positioning System (WPS) allows us to use information from the connected network to obtain the wearer's current location. UnwiredLab provides a Geolocation API that utilizes WPS to return the location. Given the location in the text message, emergency contacts can find the wearer before their vital signs deteriorate.

Future improvements possible

Using Arduino Nano R3 and ESP8266-01 to make the device smaller

Adding 3D printed enclosures for hardware components

Press the emergency button again to cancel the emergency and send another text message

Add health features such as steps and activity time measured by the accelerometer

Add a monitor or implement a Blynk app to view health data, past emergency alerts, and edit emergency contacts