My experience at Dementia Hack 2017: Build something people want
This past weekend, Marcel O’Neil, Curtis Chong, Kevin Shen, good friends of mine and colleagues at the Toronto Hacker Club, attended the annual Dementia Hack in Toronto.
For those who don’t know me, I am a 15-year-old hackathon Enthusiast. I have been to 9 hackathons and won 5 prizes. In this hackathon, I didn’t win anything, (I didn’t even stay the entire time), but it has been my most unique and favorite hackathon experience so far. You must wonder why. Here is my story at Dementia Hack 2017.
Original Plan
To be honest, I was neither interested in Dementia Research, nor in Neurotech. We decided to attend just because we would like to try our Muse, a brain sensing headband. We looked at the name of the hackathon, Dementia. Umm, that’s kind of related to the brain, let’s try to program our Muse. That became the first mistake we made.
Mistake #1: We decided on the tool before looking at the problem.
The hackathon started, and we were assigned to Category 4: try to help scientists and researchers to research and cure Dementia. Hackernest invited about 50 mentors, including researchers, doctors, health software engineers, and even a dementia patient. We did some research on the internet and found that the EEG data that Muse collects might be useful to analyze Dementia. We assumed the problem is that doctors won’t have enough valuable brain data. Our original idea was formed; a database system for EEG data so that doctors can analyze Dementia visually. We were pretty confident, didn’t talk to any of the mentors there, and moved to a Second Cup nearby with stable high-speed wifi.
Mistake #2: We made our own assumption about the problem.
Mistake #3: Missed the chance to talk to professionals and users (researchers, biologists, doctors, patients). Lack of understanding of the status quo and users’ demand.
So we started programming the Muse and building the server. We faced many technical issues but we eventually solved them. At the end of Saturday, we made an app that reads data from Muse and a server running to store all the data.
Later that night, we went to The Knowledge Society and met with Navid Nathoo, the director of the program. We showed our product through a short demo. He asked us to reflect on the key features of the product and explain to him why it matters. We thought for a long time and couldn’t answer those two questions at all. Is our hack something that would be useful? Is it just a simple hack that collects data and stores them onto the server? What is the point of this? Why does it matter? Will anyone ever use it?
After thinking through these questions, we suddenly realized we messed up. We didn’t even know what the problem is, what features the users would like to have. Without knowing the demand, it is impossible for us to design a product people would actually use and would want to use.
A Change in Strategy
We realized our problem and we decided to change in the last half of the hackathon. We basically “restarted” the hackathon. On Saturday night, Kevin hurried back to MaRS to find any possible mentors to talk to (and of course, to get free dinner). Luckily, he found 2 mentors who provided some crucial suggestions.
“With some kind of remote scanning systems, patients can do memory test in their home now without being supervised by a doctor.” — from a doctor
We never thought about what the remote, convenience feature of our product could lead to. With our product, patients no longer need to travel a long way to hospitals now. Our program can also provide a platform for all kinds of customized integrations.
“We want cheaper and more consistent brain scanning system.” — from a doctor
The current brain scanning system only collect a short segment of data. The devices are usually large and complicated to operate. The $250-dollar Muse is far cheaper and easier to use (just press the button) and could scan EEG for a long time at one charge.
We all went home that night because we didn’t want our parents to get worried. The next morning, I came back and decided to ask more mentors and dig deeper into their demands.
I was afraid. I never talked to professionals in that field before, and I knew nothing about Dementia research nor the basic biology concepts. However, I decided to step up. I walked up to two judges and joined their breakfast talk. Getting encouraged by that, I waited at the mentor and judge check-in desk. As soon as a mentor or judge came to the Hackathon, I came up to him/her and started talking. In one single morning, I spoke to about 10 mentors, made notes, constantly brought our product for them to test, and asked for feedbacks.
“I want to know more about what happened to the patient when Dementia happens.” — From a dementia researcher
A bunch of EEG data doesn’t represent anything, unless there is some kind of reference to the surroundings. We decide to install a camera next to the Muse so that it could analyze the environment while Dementia EEG signal is scanned. This could provide some crucial reference information for research.
“Patients in early stage Dementia do not want to show others and admit they have dementia” — From a dementia researcher
“I want something comfortable and cheap. I don’t want it to change to my lifestyle. ” — From a dementia patient
We decide to hide the Muse headband (install it under a cap), so that patients are confident wearing it all the time, without feeling embarrassed. With our hack, patients can easily integrate medical scanning into their life without changing their lifestyle significantly.
“I want relevant data collected from uniform platforms and real-life scenarios.” — From a data scientist
Currently, different hospitals have different operators using different techniques and devices. EEG data is not collected from uniform sources and hence is difficult to analyze. Our product provides reliable real-life data from the same hardware powered by the same code. Implementing Machine Learning or Data Mining in our database would create meaningful results.
“Doctors don’t want to learn to operate a new complicated software system. They want some simple UI that anyone can understand. They also want something that could be easily integrated into their current software system. — From a medical software designer
We decide to design a nice UI for the doctor to access the database. They can also download a .csv file so that their existing system can easily import the data. We are also planning to put the service on an HIPAA-compliant platform in the future.
After talking with mentors, we established the adjusted functions of our hack:
- Scan patients’ EEG data by using Muse
- Transfer Muse data through Bluetooth Low Energy to an Android App
- Android App uploads the data to our server, add the data into database
- Researchers and Doctors can use a user interface to access, graph, and download data(.csv) in a particular time period they select.
- Stream Video to Doctor with analyzed environment (didn’t get to work well because we were underground and we had poor wifi)
The redesigned product solves problems researchers faced in these three aspects:
- Reduce the procedure, technical difficulty, and cost significantly of Dementia Research. The patients no longer need to come to hospitals and do long and expensive experiments and testing and the doctor can remotely control the patient. Rather, they can simply wear the device and experience normal life while testing. The cost of the research/scanning is only a 300-CAD Muse headband that you can use forever, much cheaper than going to hospitals constantly.
- Increase the consistency and reliability of the data. Because the data are collected by uniform hardware and code in real-life environments, rather than different equipment and techniques offered in the hospital, the data are likely to be analyzed with higher accuracy and consistency.
- Modularity and infinite possibilities. We provided such a platform that many features based on individual need could be added to. Such as adding a camera to let the patient do memory test remotely, and provide environmental references for significant brain signal changes.
In the last four hours, we finally started our stressful coding process. We successfully called post request in our Android App, built a server receives and reorganizes the data, and set up a website to access the database and graph real-time EEG data. Everything magically works well at the last quarter of the hackathon.
The demo also went super well. We described the problem that researchers and doctors faced today, and how our product fits their demand and provides a better solution. (Fun fact: Among the 6 judges judging us, I talked to half of them earlier in the morning. They were glad that we addressed and created solutions to the problems they faced.) The judges also got to wear our product and saw their real-time brain signal data. We talked little about the technical aspects and focused more on why our product matters.
Reflection
Dementia Hack is really different from usual college hackathons. During a typical college hackathon, you won’t have many useful insights of the problem that you are trying to solve because there is no limitation on topics. Dementia Hack is so unique. They invite 50 professionals in the industry and even patients to focus on just one problem — Dementia. You have all the professional insights and supports to help you design and adjust your product. The product you create here would mostly like to be valuable for the market.
If I attended Dementia Hack again, we would spend the first severals hours talk to researchers and doctors, learn about the status quo and their demand and opinion on dementia results. We used our standard coding or college hackathon method for this event in the beginning, which causes us create a product with useless feature and wasted lots of time. When we started, we made many assumptions about researchers’ demand, and none of them are proved to be right. The entire first-half, none of us used the generous mentorship resources, and we basically wasted the first 12 hours on non-relevant technical procedures. We realized the problem when Navid talked to us Saturday night. Luckily, we adjusted our hack in the last 8 hours, and we have learned so much in that final quarter of the hackathon. The evening after and next morning, Kevin and I went up and talked to about 10 mentors and a dementia patient for their opinion on dementia research and what features they would want most. We learned about their actual demand and changed our product focuses just 4 hours before hacking ends.
Key things I learned from this hackathon
- Consult your users. Utilize your mentors. Get other people to test.
- Design something people want and they would use.
- It is never too late to change. Don’t give up!
I really like the style of this hackathon. They gave you rich resources and mentorships, narrow topics, so you can actually design something that would make a difference in that field. I knew nothing about Dementia research nor treatment before the hackathon, but after the last 8 hours, I got to know a lot in all aspects. I particularly enjoyed the last 8 hours where I talked to many people and design product that fits their demand. I will definitely take this year’s lesson and implement it into next year’s dementia hack and future projects.
Again, big thanks to Hackernest for hosting such an inspiring hackathon. Dementia Hack is hosted every year. I highly encourage you to come next year!
Here is the detailed documentation of our hack on Devpost.
We also shared our hack with Muse. We are planning to visit them in April, and I will write a blog about that.
