Description

Stress is something that all but the best of us undergo everyday and the increasing work hours is not exactly helping. Aman is a low cost connected e-textile based sock that comfortably and non obtrusively quantify stress and aid people with hypertension identify and manage their stressors.Aman does so by measuring Electrodermal Activity(EDA) and Heart Rate.
The problem with the existing Fitness trackers is the lack of context which sampling, Without context interpretation of the data becomes difficult. Aman measures context using environmental parameters like Temp,Hum& Motion.

EDA is a part of the commonly known "lie detector" and even misused in circles of scientology(E-meter), EDA however is an important parameter that would help us quantify stress. Various products elucidate the use of EDA such as the Empatica Embrace which is used for Epilipsy monitoring. Those along with some amazing research at MIT media lab by Professor Rosalind Picard has inspired our work.

Details

The Problem

Stress has become a growing concern in modern society as the job environment and education are becoming more competitive. Long term consequences of the same are often ignored leading to a plethora of psychological and physiological problems if left unchecked. Stress monitoring and periodic feedback can help people better manage their stressors by providing easy to use tools. Sweating is a biological mechanism which is controlled by physiological and psychological activity. Emotional stress can be correlated with sweating activity.

Existing Systems:

Existing devices have been bulky and have primarily been used for clinical diagnosis. To perform continuous monitoring however user’s comfort plays a major role during development of such a device.
The current commercially available EDA sensing tools are bulky and comprise of a separate amplier and Analog to digital convertor along with a data acquisition device.
The external electrodes used are uncomfortable for long use. The current systems include edaMove, Empatica Embrace, CPSpro EDA tool. The Embrace and edaMove have explored wearable wristworn designs and though an improvement to previous designs, they measure only EDA and are quite expensive for consumers. One of the major reasons for discomfort is the use of metal plate electrodes or gel type AgCl electrodes while measuring EDA.

Solution:

Aman: The machine learning assisted stress monitor

My goal is to quantify stress in a person by measuring Electrodermal Activity, Heart Rate and Heart Rate Variability continuously in a person invisibly in a Sock.
Along with this motion data along with ambient Temperature and Humidity is measured.
Use of continuous monitoring aids in preventative healthcare diagnostics and use of data sharing with loved ones can aid in forging better relationships and handle stress better.
This form of quantified stress tracking can be correlated specific events which when digitised allows not only raw records of the event but also how the person reacted to the same. Contextual information such as this can be viewed by the user later to reflect on the events and alter behavior using machine learning.
Dry fabric electrodes have been shown to improve comfort from the standpoint of the user while compared to standard metal electrodes and thus we aim to use it.

Photoplethysmography would be done on the big toe validated here, Textile connectors would snap into the toe sensor enabling quick removal before washing.
HRV and Heart Rate would be calculated by measuring the reflectance of the 535nm light in the toe. Piezoresistive strips would be used to provide information on gait and motion data.

Environmental information such as temperature, humidity would be sensed by a HTU21 sensor. Motion data and gait would measured using the Eeonyx Piezoresistive textile which would be lined along joints and the toe to enable better filtering of PPG against blood dynamics due to motion.

System Design:

Electrodermal Activity sensing:

The Electrodermal activity signal consists of two parts the tonic and phasic activity, the tonic portion of the EDA signal is slow varying and is called the Skin Conductance Level (SCL) which corresponds to physiological the phasic activity is quick reacting and are associated with short term events and in response to cognitive and sensory input the sudden peaks are known as Skin Conductance Responses(SCRs). Besides this the physiological activity is a contributing factor while monitoring EDA.
We have used conductive textile based sensors stiched inside the sock with two electrodes placed an inch away in the arch of the foot. This has been validated by research in Wearable's lab ETH.

For initial evaluation and testing, We just use a resistor divider along with a low pass filter ceramic capacitor before interfacing to the ADC. More work has to be done to add some active filters and amplifiers (Maybe some wheatstone resistance bridge?).

Heart Rate Sensor:

A pulse sensor amped from World Famous Electronics...

Files

Aman.pptx

Powerpoint presentation of the project

presentation - 2.91 MB - 10/03/2016 at 13:32

Download

Components

Project Logs

Collapse

Prototype 1.0 Finished
Naveen Sridharan • 10/03/2016 at 12:29 • 0 comments

Hi guys,
I finally got the first prototype up and running, The socks now can accurately measure my electrodermal activity and share it to thingspeak along with my heart rate & environmental temperature and humidity.

I will be adding the ADXL345 code along with some gait sensing soon enough.

I got these fancy leather wrap done to cover the electronics neatly and allow it to be detached easily when washing the rest of the sock. The prototype was a tad bit bigger than what I had in mind but the next revision would be way smaller in size by using the particle P1 SOC module.

Here is the image of the sock without the electronics attached.
Neat eh?

Now here it is, when electronics are attached

Next up we'll look at some data streamed to thingspeak.
This was the data I acquired for a short while, More data on the way.
Motion & Humidity Temperature Sensor
Naveen Sridharan • 09/28/2016 at 12:00 • 0 comments

Motion data is vital to give context and add tags while cataloguing&labeling the data,
The motion coefficient is measured using an ADXL345 I2C accelerometer sensor.

Dynamic motion is captured in a pedometer like function, Data is logged when there is sudden movement over a given period of time. This helps make it possible to catalog physical activity. Accelerometers are surprisingly accurate and inexpensive providing valuble insight in the user's life.

Fabric based gait sensor would also be used to monitor the user's gait and posture.
We'll be using pressure based fabric sensors below the surface of sock to monitor foot pressure.

(Plusea has some great documentation)

We measure Humidity and Temperature using the HTU21d sensor which is clipped to the shoe to measure the environmental parameters. Maybe make a shirt clip addon BLE version for measuring the environmental parameters.
Electrodermal Activity
Naveen Sridharan • 09/28/2016 at 06:34 • 0 comments

The core of the project revolved around development of an accurate and effective Electrodermal Acitivity sensor. EDA modulates the amount of sweat secretion from sweat glands. While sweat secretion is related to thermoregulation and sensory discrimination, changes in skin conductance in hand,foot regions are also triggered quite well by emotional stimulation. However both positive response (Joy) and (fear, sorrow) often illicit the same response. To this end after reading various research journals, I decided to use a conductive fabric based electrodermal sensor. The location I chose to measure was the arch of the foot to enable accurate sensing of dermal response and the lack of motion in that area made it ideal.

Using a bit of fabric glue to stick it to the inside lining of the sock I used conductive thread to stitch connection from the electrode to a snap.

The snaps enable the rest of the electronics to be disconnected.

The principle of electrodermal activity is relatively straightforward, Skin conductance is measured accurately. A low pass filter with a 0.1uF ceramic capacitor is used to give a clean signal.
I'm looking into a wheatstone bridge based sensor interface, Adding some active filtering is up on my list too.
I've gathered some data over serial for about 10 minutes and plotted it on excel
Photoplethysmography
Naveen Sridharan • 09/28/2016 at 04:04 • 0 comments

Photoplethysmography forms the back bone of today's wearables ranging from the Apple Watch to fitbits. It works by sensing the change in blood flow volume by measuring the transmittance or reflectance of the 535nm light(Green).

This wavelength is ideal as it as at this wavelength peak absorption of heamoglobin happens. I have hence chosen to use the simple yet effective Pulse Sensor Amped from WorldFamousElectronics cheaper ebay knockoffs work fine too.

Since I cant have a bunch of wires running through a socks and expect the user to be comfortable, I had to connect this to some fabric ribbon cables.

I ran into some issues initially as the wire would snap under the stress of my toe moving, This was an issue I would have to deal with later. But for now nothing a bit of E-6000 glue cant fix.

Now that the cable is all stable and rigid, I got to the conductive textile ribbon cable part.
I coincidentally had some FabricKit three channel cables lying around in my bench, These cables look and feel like textile but are actually solderable and conduct electricity.

As it comes with it's own snap like connectors this part became easy. I just soldered the pulse sensor cable to the connector snap.

Now I soldered the conductive ribbon cable to the male snaps.

After some struggle with some solder bridges, I managed to connect it properly.
Now for the moment of truth, Does it work?
Thankfully yes.
The Story
Naveen Sridharan • 08/07/2016 at 04:06 • 0 comments

I'm a final year electrical engineering student based out of Chennai,India. I love building and making gadgets and have strong desire to bring in change to the world.
I got inspired to work on Aman after seeing a close relative of mine suffer everyday due to stress, Every time I met him during family visits I found him to be worn out and he was aging rapidly before my eyes.
After taking an elective of Biomedical Engineering at college, I learnt of various sensing methodologies that could quantify physiological parameters.
We were being thought about Galvanic Skin Response one lousy afternoon in class when I suddenly was struck by an idea for stress monitoring. I immediately started sketching it out on my journal and started making a storyboard to better understand the problems faced.

Here's the storyboard which was compiled from various problems faced by people suffering chronic stress.

I realised such a device would have to be very comfortable to wear for long term use and got started doing some research on some fabric materials for sensing and textile connectors.
I also learnt that a lot of people were disappointed with existing "fitness" trackers and reading research into accuracy involved with measuring Heart Rate in wrist, I was not suprised that those device were inaccurate and limited in function.
I began sketching out a possible solution to solve the issues.