# Parkinson Disease Wearable Tech

441

5

1

## Introduction: Parkinson Disease Wearable Tech

More than 10 million people worldwide are living with Parkinson’s disease (PD). A progressive nervous system disorder that causes stiffness and affects the movement of the patient. In simpler terms, many people suffered from Parkinson’s disease but it is not curable. If deep brain stimulation (DBS) is mature enough then there’s a chance for PD to be curable.

By addressing this problem, I will be creating a tech device that could possibly help hospitals to offer PD patients more accurate and practical medications.

I created a wearable tech device – Nung. It can accurately capture patient’s vibration value throughout the day. Tracking and analyzing recurring pattern to help hospitals make better medication decisions for each patient.
Not only does it provide accurate data to hospitals, it also brings conveniences to PD patients when they revisit their doctors. Usually, patients will recall their past symptoms and ask doctor for further medication adjustment. However, it is difficult to recall every single detail, thus making the medication adjustment inaccurate, and inefficient. But with the use of this wearable tech device, hospitals can identify the vibration pattern with ease.

## Step 1: Electronics

- ESP8266 (wifi module)

- SW420 (vibration sensor)

- Jumper wires

## Step 2: Vibration Monitor Website

By graphing this out, hospitals can visualize patient’s condition live.

1. SW420 captures the vibration data from the user

2. Save the time and vibration data to a database (Firebase)

3. The website will get the data stored in the database

4. Output a graph (x-axis - time, y-axis - vibration value)

## Step 3: Machine Learning Model

I’ve decided to use Polynomial Regression model to identify the user’s greatest average vibration value from different time period. Reason being my data points do not show an obvious correlation between the x and y-axis, polynomial fits wider range of curvature and more accurate prediction. However, they are very sensitive to outliers, if there are one or two anomaly data points, it will affect the result of the graph.

x_axis = numpy.linspace(x[0], x, 50) # range, generation y_axis = numpy.poly1d(numpy.polyfit(x, y, 5)) # draw x y, 5 nth terms

## Step 4: Assembly

At the end, I modify a few electronics and decided to use lithium polymer battery to power the wearable tech. This is because it is rechargeable, light weight, small and can move around freely.

I've solder all the electronics together, designed the case on Fusion 360 and printed it out in black to make the whole product look simple and minimal.

Participated in the
Battery Powered Contest

28 2.4K
5 843
147 12K
3 753