BIG DATA
UNO faculty & students mine bits and bytes
To help make sense of big mysteries
By Lori Rice
Let’s stretch the brain for a bit with a quick science lesson on … a teaspoon of water.
Never thought much about it? Think about this — each teaspoon of water contains 2^1023 water molecules. And given that each water molecule is comprised of 3 atoms, we can say there are more atoms in one teaspoon of water … than there are teaspoons of water in the Atlantic Ocean.
Whoa — mind blown.
Here’s another staggering factoid — best estimates suggest that at least 2.5 quintillion bytes of data is produced every day. That’s 2.5 followed by a staggering
18 zeros.
“Everything we do today is leaving a track,” says Parvathi Chundi, professor of computer science at UNO and a data analytics expert. “Everything we do today is being recorded in some way, is being logged in some way.”
Raw data — bits and bytes — by itself without purpose is white noise. The focus, though, should not only be on the amount of data, but the value that data can bring to organizations and how we can use this data to improve products and services. Quantitative and Qualitative.
UNO researchers are doing just that in a variety of ways, funded by various grants and supported by faculty, students and collaborations with other universities and healthcare systems across the country.
Seeing the whole picture
At UNO, Chundi’s focus is in the field of healthcare, where she is developing software that can help ophthalmologists understand the progression of age-related macular degeneration (AMD) and other retinal diseases. She works with doctors from the Byers Eye Institute at Stanford, along with faculty in the department of ophthalmology at the University of Nebraska Medical Center (UNMC).
In the case of AMD, ophthalmologists use many different instruments to take pictures of the eye, and each picture tells a different story about the eye and how healthy it is. Once the disease progresses, vision has already weakened. Dr. Chundi’s work uses enormous amounts of data from hundreds of patients that includes pictures and data from medical records to come up with predictions and correlations among data points. She builds the software that analyzes the data then predicts patient outcomes and is continually working on building the software in a way that it can crunch through the data at quicker speeds and come up with predictions that will help ophthalmologists.
“As we collect more data and as we analyze it – in the future we want to prevent AMD – but right now our focus is on prediction and correlations,” Chundi says. “Visually, we cannot see something, but algorithms can see something. The idea is to actually see that you are going to have this vision loss much earlier, before it actually happens.”
A growth in knowledge
Her work with big data also spans into another large project that involves researchers from five institutions across four states (Montana, Nebraska, Nevada and South Dakota) where they are looking for the growth of microbes and bacterial growth on materials.
“This project is really exciting and really different than what I have done before. This project is actually trying to see bacterial growth on materials,” Chundi says. “The idea of the project is: can we help material scientists in the lab to find materials where the microbe growth can be mitigated? We are using big data; in this case the genetic data from the microbes, the research data that is being brought about in the lab by researchers, and also papers that are published by the researchers in the field.
“Combine all these things and can we actually synthesize information that material scientists can take to the lab and devise experiments to discover new properties of materials and new materials themselves.”
The implications of these studies run the gamut from identifying the degradation of oil pipelines, bridges, and even materials used in spaceships.
“The idea is that materials we are using in daily life are prone to deterioration and destruction, and can we build materials that help us stay safe?” Chundi says.
Bridging a gap
Bridges, in particular, are complex structures with mass amounts of data to assimilate and understand bridge health.
“Bridges have a problem,” says Brian Ricks, assistant professor of computer science, and founder of UNO’s Bricks Lab (formed from his name, Brian Ricks). “Along with a long list of critical infrastructure in our country, they tend to be old.
“And when things get old, they break down.”
Ricks has been studying big data for years in an effort to improve the safety of bridges across Nebraska and the country. Funded by a $1,000,000 grant from the National Science Foundation, the program is a combined effort with students and professors from UNO and UNL. The research is done out of the Peter Kiewit Institute, which houses UNO’s College of Information Science and Technology, along with the College of Engineering and Technology at UNL.
“It’s been a great collaboration,” Ricks says. “We are trying to use big data to help our engineers choose the next bridge to repair and where they should put their resources.”
The collapse of bridges over the years, such as the tragic catastrophic failure of the I-35W Mississippi River Bridge in August 2007 that killed 13 people, have served as a catalyst for a greater reliance on big data to help identify weaknesses in critical components of major bridge structures.
“You really want to avoid the catastrophic failures,” Ricks says. “In order to do that – you have thousands of bridges in your state and a limited budget – you have to be repairing the right bridge next so you don’t have the catastrophic failures.”
UNL students collect data on bridges in various ways including gluing sensors to bridges that measure all sorts of movements and fluctuations and by using drones that are able to get to places where it is difficult for a human to go and examine bridges for cracks. UNO students then take all of this data and create programs that display patterns and trends that can anticipate and identify bridge priority in terms of inspections and repairs.
Doctoral student Akshay Kale uses machine learning to look through inspection records on more than 15,000 bridges in Nebraska and uses predictive analysis to look at patterns found in the survey data.
“We hope that machine learning and going through all the big data will lead us to the bridges that require the most attention,” Kale says.
Brick by brick
As part of the Bricks Lab, UNO graduate student Ryan Narducci works with Ricks and UNMC on a 3D hospital crowd simulation that virtually reenacts different scenarios and mass triage situations in emergency room settings.
“It’s also a cost-efficient and safe way to examine if there are any changes you can make without having to actually make them at the hospital just yet,” Narducci says.
Using these simulations can improve the quality of patient care and safety, and improve flow through the emergency room.
“We have so much data we don’t know how to process it,” Kale says. “If a human had to do it by manually looking into, it’s not even possible to understand what the data is all about.”
With the use of these cutting edge technologies, these researchers and students are beginning to crack the code of big data in a predictive way which will be of great use to society.
Like making it possible to drink the ocean, one teaspoon at a time.