In the United States, every 11 seconds, a senior is treated in the emergency room for a fall. Every 19 minutes, an older adult dies from a fall, directly or indirectly. Deteriorating balance is one of the direct causes of falling in seniors. This epidemic will only increase, as the senior population will double by 2060. While we can’t prevent the effects of aging, we can slow down this process of deterioration. Our mission is to create a solution to senior falls with Smart Soles, a shoe sole insert wearable and companion mobile app that aims to improve senior health by tracking balance, tracking number of steps walked, and recommending senior-specific exercises to improve balance and overall mobility.

What it does

Smart Soles enables seniors to improve their balance and stability by interpreting user data to generate personalized health reports and recommend senior-specific exercises. In addition, academic research has indicated that seniors are recommended to walk 7,000 to 10,000 steps/day. We aim to offer seniors an intuitive and more discrete form of tracking their steps through Smart Soles.

How we built it

The general design of Smart Soles consists of a shoe sole that has Force Sensing Resistors (FSRs) embedded on it. These FSRs will be monitored by a microcontroller and take pressure readings to take balance and mobility metrics. This data is sent to the user’s smartphone, via a web app to Google App Engine and then to our computer for processing. Afterwards, the output data is used to generate a report whether the user has a good or bad balance.

Challenges we ran into

Bluetooth Connectivity Despite hours spent on attempting to connect the Arduino Uno and our mobile application directly via Bluetooth, we were unable to maintain a steady connection, even though we can transmit the data between the devices. We believe this is due to our hardware, since our HC05 module uses Bluetooth 2.0 which is quite outdated and is not compatible with iOS devices. The problem may also be that the module itself is faulty. To work around this, we can upload the data to the Google Cloud, send it to a local machine for processing, and then send it to the user’s mobile app. We would attempt to rectify this problem by upgrading our hardware to be Bluetooth 4.0 (BLE) compatible.

Step Counting We intended to use a three-axis accelerometer to count the user’s steps as they wore the sole. However, due to the final form factor of the sole and its inability to fit inside a shoe, we were unable to implement this feature.

Exercise Repository Due to a significant time crunch, we were unable to implement this feature. We intended to create a database of exercise videos to recommend to the user. These recommendations would also be based on the balance score of the user.

Accomplishments that we’re proud of

We accomplished a 65% success rate with our Recurrent Neural Network model and this was our very first time using machine learning! We also successfully put together a preliminary functioning prototype that can capture the pressure distribution.

What we learned

This hackathon was all new experience to us. We learned about:

  • FSR data and signal processing
  • Data transmission between devices via Bluetooth
  • Machine learning
  • Google App Engine

What's next for Smart Soles

  • Bluetooth 4.0 connection to smartphones
  • More data points to train our machine learning model
  • Quantitative balance score system

Built With

Share this project: