At the World Economic Forum in late January, Patrick Hackett and Daniel Zagyva, students in Georgia Tech’s interdisciplinary Master of Science in Analytics (MSA) program, were part of UHAI, a team that participated in FutureHack. FutureHack was a 48-hour hackathon designed to address the United Nations’ Sustainable Development Goals. Out of 18 participating teams, UHAI went on to earn the top prize of $100,000 and the opportunity to present the team’s project to WEF attendees.
Afterward, UHAI became a start-up venture that has since on-boarded a team of marketing, branding, strategy, and business development professionals. The company is prepping to present offer documents to potential investors, and once funding is secured, UHAI will continue to build out its technology.
In the meantime, Hackett is preparing for his final semester in the MSA program and is currently working for NCR as an IT data analyst for the program’s Business Practicum project. In his role at NCR, he works with big data to perform analytics on repairs and services for ATMS, self-checkout lines, and other NCR products that he describes as “making the everyday easier around the globe.”
In the following interview, Hackett discusses what draws him to data science, the FutureHack experience, and what he’s learned from the hackathon and the MSA program.
What brought you to Tech’s MSA program?
With my background in neuroscience and nonprofit work, I found some of my most enjoyable experiences were being able to solve problems on the technical level and then bring them to the wider world. I feel like Georgia Tech – and in particular, ISyE – has always excelled at that. Both technical know-how and being able to understand people and the problems that companies face is much of what the MSA degree is about.
When you say you like to solve problems on the technical level, what does that mean exactly?
The MSA is a data science degree; I like taking data and finding answers in it. That means formatting the data in a way that you can look at and understand it, and then performing analysis on it using statistical modeling and machine-learning models. So, finding trends and relationships in the data, often with the goal of predicting or advising behavior, whether it be for a major corporation or a nonprofit.
This degree not only teaches us theory and skills, but has us work directly on projects that require us to think on our feet to create applicable and reliable solutions.
The people I get to work with every day make the program something special, because my cohort is very gifted. At least half of our students are international, and everyone is collaborative and hardworking. We work in teams a lot, and I love working with smart people who really care about each others’ success. I think that’s something Tech has in general.
Let’s talk about FutureHack. What was it exactly, and why did you want to participate in it?
As we’ve all been hearing in the news, the blockchain space and cryptocurrencies are becoming widely popular. Every day you’re hearing something new about Bitcoin. So, some of these self-made blockchain entrepreneurs decided to fund the first hackathon during the WEF in Davos as a parallel event. The goal was to envision blockchain in a way that serves humanity, specifically to address the United Nations’ Sustainable Development Goals (SDGs). That’s the basic premise: We had 48 hours to build something using blockchain that also addressed one of the SDGs. The team leader, Quang Vu Dang, brought me onto the team, which consisted of five people.
Can you explain how you developed your model for the hackathon?
Our team, UHAI (Universal Health Artificial Intelligence), created a decentralized platform that processes, stores, and unifies electronic medical information on the blockchain, using artificial intelligence (AI) to detect anomalies and predict health outcomes. At the hackathon, we specifically used breast cancer mammogram images to predict whether tumors were malignant or benign.
The blockchain interface of UHAI provides a distributed system for accessing the health care data that is secured on a database. Daniel and I built a neural network to analyze these images, training the neural net on historical images where we knew whether the tumor was malignant or benign.
Once the model is trained on the labeled data, it can predict whether a tumor in a new image is benign or malignant.
In the 48 hours of the competition, we were able to reach 70 percent accuracy on the test data set, which already reaches the average accuracy rate of radiologists. Based on current research, our goal is to reach 90 percent accuracy and provide an effective second opinion in the health care space, based on machine-learning algorithms built upon thousands of patients’ data.
What’s the benefit to the health care system of using a model like you created?
There are already computer-aided diagnoses in hospitals – we’re trying to improve that system and extend it to every cutting-edge technique that’s out there and put that technology directly into the hands of patients. At the same time, we’re creating an infrastructure for incentivizing data storage with UHAI, allowing our algorithms to continuously improve as we obtain more and more data.
The long-term idea is that this will reduce health care costs. For example, if you can improve diagnoses at the imaging step rather than needing biopsies to determine whether a tumor is malignant, this will cut down costs considerably. And there are many other areas where this applies.
What was it like to be on the winning hackathon team?
There were such bright teams participating -- from Russia and Germany, as well as a couple of other American groups who had already been very successful in blockchain, so they had a lot of funding and superstars on their team.
In the end, we felt honored to have won. We felt that we had built a comprehensive proof-of-concept model, and our goal was to build a platform to eventually be able have a start-up. It was a unique and special experience – being at Davos – which just made us all the more honored to have won.
How have the hackathon and what you’re doing with UHAI grown your data analytic skills?
It blows my mind! And I’ve experienced this in the MSA program as well: how much having to do a big project shoots you up multiple levels. I hadn’t even worked with neural networks before the hackathon, and I felt like I became someone who – at least, with a specific kind of neural network – could do professional-level work by diving deep and having to solve a problem.
That’s also one of the amazing aspects about the MSA program, especially the second semester. The format becomes less about introductory theory and more about challenging projects that require us, as students and professionals, to work together and compare multiple different approaches to identify the best model. Thanks to my coursework, I can see a big project and rather than feeling like I have no idea how to approach it, my attitude is that I can do the work, and do it well.
Stewart School of Industrial and Systems Engineering