Stevens Digital Healthcare Hackathon

This project is my approach to tackling a widespread, yet very serious problem – alcoholism. No one knows exactly how many alcoholics there are in the world, but reputable estimates for the number of alcoholics in the US vary from between 12 million to 18 million people. In the US, which has a population of about 326 million people, this means that about 3.6% to 5.5% of the population abuse alcohol heavily.

After watching someone close to me struggle with alcohol for over two decades, I decided to do something about it. My university, Stevens Institute of Technology, hosted a Digital Healthcare Hackathon during October 2017 aimed at providing better solutions in tele-health and managing Big Data for medical applications. I deviated a bit from the recommended categories and decided to tackle this problem, which I had been mentally chewing on for quite some time. I brought my kit of electronics prototyping parts, formed a group with Sharan Juangphanich and Ryan O’Shea, and got to work.

Over the course of one weekend we had 6 hours to brainstorm with random people and find a team to work with, and 36 hours to work, including sleep. During that time, we:

1. Built a hardware prototype of the wearable device (with Bluetooth)
2. Built a website to collect sensor data from the device
3. Made a presentation

Rather than merely forming a team of friends to work on something with, the aim of the hackathon was to find others whose passions were aligned with your own. This was actually quite helpful, as it allowed us to form teams of people whose skillsets meshed well together, and made us work more efficiently together.

This project can be found on Devpost as well.

Background

There are multiple different existing solutions for alcoholism recovery. Some of these solutions include detox and rehabilitation centers, where alcoholics can take special pharmaceutical drugs to ease off of the withdrawal. Withdrawal from alcoholism can easily be deadly if done too quickly and without the assistance of drugs, as the body has formed a physiological addiction to ethanol. Once detoxification is complete, alcoholics typically undergo a multiple-month long program where they live together with other alcoholics and try to reassess their own situation and how they can restructure their lives in a way that doesn’t involve alcoholism. It is quite often the case that many alcoholics are suffering from mental trauma, depression, or anxiety, and use alcohol as a means of drowning out the source of their own despair – which unfortunately, comes from within their own thoughts. By living in a constant state of drunkenness, alcoholics are seeking reprieve from their own thoughts by turning off their prefrontal cortex so that they physically cannot think about what troubles them. These rehabilitation centers, which place very tight restrictions on what the patients can do during the day, are attempting to help the patient find the root cause of their mental illness while they are in a sober state.

After undergoing these programs, it is common for recovering alcoholics to then join Alcoholics Anonymous (AA), an organization which connects recovering alcoholics to mentors who have themselves endured the struggle. Through regularly repeated interaction, perhaps meeting a few times a week or even every day, the alcoholics can keep their mind firmly set on their goals. Additionally, a very important element of having a mentor is that there is a much greater sense of accountability – imagine having a coach that you really don’t want to disappoint. If the alcoholic-mentor relationship is a good match, then the alcoholic should be afraid of going back to his or her old ways.

While these programs sound great, the sad truth is that the relapse rate for alcoholics after finishing these programs is about 90%.  The transition from a tightly controlled, restricted environment to the freedom of the real world is a rough one. Once leaving the programs, alcoholics find themselves right back in the world where alcohol is cheap and readily available almost anywhere in the country. Who would find out if I just had a drink? Just one. I’ll hide the rest. Maybe I’ll stash the bottle in attic. Or maybe under the shed in the backyard. Who’s gonna know?

The fact is, it might take a few days before anyone really notices that the alcoholic is back to drinking again. It tends to be a gradual process where the alcoholic can sneak a drink here or there, and is drunk for only a short period of each day. As the days progress, sooner or later the alcoholic will be back to drinking around the clock, and it isn’t this point that anyone else really notices. But by this point, it’s already too late to encourage them to stop. It’s back to square one.

The Mission

This project aims to target recovering alcoholics coming out of these rehab programs. For families concerned about their loved one, it would be extremely valuable to have instant feedback on what the alcoholic was up to, and whether or not they had started drinking again. By having feedback within an hour of the moment they started drinking, family, friends, and mentors could step in and remind them of why they started on this journey, and what is at stake if they continue on this path.

Ethyl is a wearable worn by the alcoholic, taking regular breath alcohol measurements (BAC), once every hour. These readings are reported to a website where friends, family, and mentors can view the alcoholic’s progress on their campaign to be alcohol-free. On this website, people can provide encouragement to stay on the right path. The website would open a clear line of communication between the recovering alcoholic and those who care about him or her, and would create a high degree of accountability.

Each night, the wearable will detect that the user is sleeping, and will not ask for breath measurements. When the user wakes, the unit will begin to ask for breath measurements once again, once every hour (unless the user pre-configures the device not to ask during a certain time, which can be approved by others).

The device also has sensors to detect whether or not it is being worn. When the device is taken off, this information will be reported to the website, so that others may know what is happening.

Frequently Asked Questions

Q. How can we know that the alcoholic isn’t tricking the sensor?

A. With a variety of different sensors, including continuity and impedance, temperature sensors, and pressure sensors, the device can recognize when the user is trying to make someone else take a measurement for them, or if they are attempting to trick it with other means. Besides, even if the user can convince someone to take a breath measurement for them, it would be incredibly difficult to get someone to do that for every single reading of the day!

Q. Why would knowing any sooner that the alcoholic is drinking make a difference for relapse rate? Haven’t they already relapsed?

A. During the first hour of drinking, the alcoholic has not fully relapsed. During this time, concerned individuals can contact the alcoholic and remind them of their mission to stop. The first 2-3 hours of the relapse is a very critical time period – just moments before the relapse, the alcoholic made the impulse decision to drink. They are in a fragile state of mind, and if we act fast enough, we can convince them to stop before it’s too late.

 

Hardware Prototype

Figure 1. Initial Prototype of the Device

Looks sloppy in the picture, right? Considering that we only had 36 hours to work, I’d say it turned out pretty well!

Most hackathons aren’t really geared well for building hardware devices, and are instead focused on creating software prototypes to different concepts. The development process for hardware devices can very often be a long and drawn out process, as there are many different design considerations and compromises to make in order to make a viable product. We decided to use some pre-existing hardware infrastructure, such as the NodeMCU internet-of-things device which utilizes the ESP8266 module. This microcontroller has built-in Wi-Fi, so it formed a perfect platform on which to build the wearable.

We cobbled together a device using the microcontroller, a Lithium Polymer battery, and an alcohol gas sensor to form the basic prototype. A boost converter board was used to convert the 3.7 V battery voltage to 5 volts to drive the NodeMCU.

For the sensor to verify whether or not the device was still being worn, we used a continuity checker. Two copper pads on the bottom of the device were in constant contact with the skin. By measuring the resistance between the two pads, we could tell if the device was being worn or not. This information was then reported to the website. Although an infrared sensor probably would have done a better job of telling whether or not the device was being worn, we were making do with the tools that we had on hand.

Lastly, a third sensor was used to tell if the user was actually blowing into the device, or if they had convinced someone else to blow into it for them! A copper electrode on the alcohol sensor would make contact with the user’s lips as they blew into the alcohol sensor. The resistance between this electrode and the user’s wrist was used to determine if the user had handed the device off to someone else to blow into it.

 

Software

The software was split in two different domains:

1. NodeMCU Firmware, written in C++ using Arduino IDE
2. Website Design, using a Bootstrap template along with JavaScript for interactive web elements

For the NodeMCU code, the wearable would connect to the user’s smartphone via a mobile hotspot. The device would make a connection to Adafruit.io’s servers, and upload data from the sensor to the Adafruit.io database. This service made it quite easy to store data from the wearable, so that we didn’t have to provide our own server. From there, the data could be retrieved from our website, which ran as a Github.io page.

The website can be found on https://atomicantimatter.github.io/Ethyl/web/.

Bootstrap and stock photos sure help a lot in making a snazzy looking website!

On the Device Page, the data from the wearable can be seen, and the recovering alcoholic’s profile page is shown. Here is an example with one of our team members (who is not actually an alcoholic). All of the dynamic content on this page was provided using JavaScript. The script would fetch data from the Adafruit.io server, and display it on the page dynamically, updating every hour. For demonstration purposes, we made it update every 10 seconds.

 

 

Presentation

 

Presenting in front of judges from Bristol Meyer-Squibb, Health Recovery Solutions, and others. It was quite stressful presenting in front of the CEO of a huge pharmaceutical company!

Receiving our award. I really wasn’t expecting first place; we were up against some awesome teams, and I’m really grateful to Stevens for the opportunity to participate in this event.