Project

# Title Team Members TA Documents Sponsor
23 Wearable Pediatric Eczema Tracker
Instructors Award
Dong Hyun Kim
Hyoungjo Hahm
Sung Hoon Lee
Jonathan Hoff design_document3.pdf
final_paper1.pdf
other1.pdf
photo1.jpg
photo2.png
presentation1.pdf
proposal2.pdf
video
# Team Members (NetID):
Sung Hoon Lee (slee528) | Dong Hyun Kim (dkim294) | Hyoungjo Hahm (hahm4)

# Problem
---------------------------------------------------------------------------------------------------------------------------
Eczema, also known as atopic dermatitis, is one of the most widespread skin conditions affecting nearly 20% of the world population and is more common and severe to young children and infants. Its main symptoms are severe itching and skin cracking which eventually damages the skin. Until this day there is no known universal cure and what dermatologists suggest is to find the causing “triggers” which may involve consumed food, environmental exposures, or chemicals and to isolate the patient from it. The biggest problem is that it is nearly impossible to track down such triggers as symptoms of itchy skin rash may arise long after 24-48 hours from making contact, and since infants and children have limited means to convey their symptoms objectively for dermatologists to diagnose.


# Solution Overview
---------------------------------------------------------------------------------------------------------------------------
We propose making a low cost wearable eczema tracking sensor specialized in pediatric population that helps dermatologists treat their patients using scratch monitor. It is equipped with bioimpedance tomography, accelerometer and temperature sensors for high detection accuracy. Patients will wear the sensor on the wrist of either arm with best comfort in a specific time of interest such as during sleep, where scratching events are detected directly through inference model in SoC and logged and give indication in green, yellow or red LED embedded in the device, depending on the severity of scratching through number and duration of scratches per timeframe so that guardians can notice right away. Users will also be able to log their daily information such as consumed foods, location and skincare appliances so when superimposing the scratch severity graph with these information, dermatologists can easily find pattern and track down which substance or environmental factor was the “trigger”.


# Solution Components
------------------------------------------------------------------------------------------------------------
## Scratch detection: Electrical Impedance Tomography (EIT) and Accelerometer Sensor
By measuring multiple bio-impedance simultaneously using 8-32 electrodes, tomographic representation of the specific part of the body can be derived. Recently there has been attempts to detect hand gestures using EIT as it can track the movement and dynamics of muscles with high accuracy that EMG or accelerometer cannot deliver. Scratching involves applying periodic force to one or more fingers which EIT can pick up and be used to determine the number and severity of scratching.

## Arm band and power system
The device will have a form factor of arm band with part involving electrodes flexible in order to have them as adequately fit as possible and enhance signal quality. As the device is intended for everyday use, we will use battery system with single switch for use with minimal effort.

## Bluetooth Low Energy (BLE)
Bluetooth will allow the sensor for wireless monitoring. Using nRF52832 SoC which has bluetooth 5.0 with 2Mbps Bluetooth Low Energy mode, it will allow for -96dbm sensitivity to the sensor. This BLE mode on nRF52832 chip is ideal for Device Firmware Update (DFU) which brings total flexibility and control to the firmware running in the sensor. It also has enough bandwidth to run machine learning inference model for scratch detection or for streaming both EIT and accelerometer data directly to the application. Depending on the implementation, this BLE SoC will communicate regularly with the GUI for real time monitoring and gives users better flexibility when moving around wirelessly using relatively low energy.

## Application for scratch logging and visualization, and Machine Learning Model
The application will have two main components- first, to log daily activity including which food the patient has consumed, places visited and skincare appliances used and second, give graphical information by superimposing these information with scratch severity graph for dermatologists to find patterns and track down which substance was the trigger. To minimize the data driven nature of certain inference models, we suggest performing preprocessing such as FFT and autocorrelation to capture the periodicity, combined with other feature extraction and use linear SVM to lower the computation load.


# Criterion for Success
---------------------------------------------------------------------------------------------------------------------------

## Hardware
The sensor precisely measures the scratch activity time using the accelerometer sensor and body temperature using the temperature sensor. By attaching multiple electrodes on the sensor around the wrist, bio-impedance is measured in designated channel and pattern monitoring of the patients’ wrist muscle activity as well as scratch detection. EIT should come out similarly without lowest noise possible and should pick up parts of the muscle used when scratching

## Software
A python based GUI will be able to monitor data from the sensor using BLE mode from the nRF52832 chip. It should represent proper real time graph of EIT around the wrist. Device firmware should allow the communication between the sensor and the GUI without lagging.

## Design
The device is stretchable to some extent in order to fit patients’ wrist and the electrodes should stick right on to the skin in order to minimize the noise. Depending on subject type, 8~32 electrodes are placed evenly around the device for even spatial resolution.

## Ethics and Safety
According to IEEE Code of Ethics, technologies by all means must avoid causing human injury. [1] Since the product is a wearable technology that directly touches and wraps around a human body with electric circuitry, there are many possible safety concerns that must be handled in order to comply with the code. For example, the circuitry contains battery, which might cause skin burn when the electric flow is poorly designed or handled. The electric current must be strictly controlled and cut off in necessary situations where any malfunctioning is detected. Other than electric hazards, the wrist band might also affect blood flow if worn too tightly. A way to properly adjust band should be provided to prevent adverse effect on the wearer's body while ensuring signal quality to be high enough for smooth processing.

Easy Cube Clock

Allan Englehardt, Jason Luzinski, Benjamin Riggins

Featured Project

Today's alarm clock market is full of inexpensive, but hard to use alarm clocks. It is our observation that there is a need for an alarm clock that is easy to set, and turn on and off with little instruction. Imagine an alarm that is set with the intuitive motion of flipping the clock over. When the alarm is on, you can see the alarm time on the top of the clock. To turn off the alarm, you simply flip it over to hide alarm display. Out of sight, out of mind. The front face of the clock will always show the current time, and will flip to the correct orientation.