Deborah Estrin, Computer Science at Cornell Tech and Healthcare Policy and Research at Weill Cornell Medicine, wants to make useful the vast streams of data produced and collected by the world’s devices. She was the founding director for the National Science Foundation Center for Embedded Networked Sensing (CENS), which ran from 2001 to 2011 and pioneered the development of wireless sensing systems that collected and analyzed data about the physical world. Today, she leverages “small data” to benefit our health.
Big data is a popular term used in the technology industry to describe giant data sets that, when analyzed, can provide useful and important information. Small data, in comparison, is about an individual.
“Small data are generated continuously on our mobile phones and through our online activities: walking and location patterns, as well as shopping, communicating, and web surfing,” says Estrin. It is the various data traces we each generate every day, just by living our day-to-day routine: checking email, taking the bus to work, going grocery shopping, walking home, and more.
The Small Data Lab at Cornell Tech, led by Estrin, works toward turning all of this small data into big insights for the individual about his or her health and wellbeing.
An App for Managing Pain
Estrin’s lab has developed an app that will help people manage pain from conditions such as rheumatoid arthritis, or more general, chronic pain such as lower back pain.
When the app is installed on a user’s smartphone, it passively collects a user’s location patterns and activity—the equivalent of what a fitness tracker would collect. It gathers far too much raw data, however, so the app’s main focus is in analyzing the information and presenting it in a way that is actionable by the patient and his or her clinician.
“To create useful behavioral and functional biomarkers for clinical care we need to develop the statistical techniques that derive meaning out of that overly plentiful and noisy data,” says Estrin. “For small data analytics, we’re not looking for patterns that emerge across a large population of patients, but instead the effective modes that the individual patient is in over time.”
Estrin and her team work closely with medical domain experts to apply new technologies to specific conditions. Estrin’s group collaborates with Weill Cornell Medicine College professors and physicians, as well as with physicians at the Hospital for Special Surgery, Pfizer, and UnitedHealth Group.
People with rheumatoid arthritis, for example, have flares and remissions. The flares do not occur according to any known, regular cycle, so there is a lot of interest in achieving control of these flares on an individual level. One of the group’s research goals is to study mobility patterns before, during, and after flares to develop models for flare prediction based on subtle changes in a patient’s physical motion. The app, in essence, would capture a given patient’s particular signature for the onset of a flare.
Orthopedic surgeons are working with Estrin to develop better measurements and metrics for surgical recovery. Some people recover quickly while others don’t. With an app that tracks their progress and their physical therapy (PT), it could be possible to ensure that patients receive the right amount of PT for the best recovery possible after their surgery, says Estrin.
The researchers are also experimenting with new ways for people to report their pain levels in the app. “How much pain someone feels versus how much the pain interferes varies from person to person,” says Estrin. “Instead of the standard pain scale, we implement an easy-to-use image-based approach that allows users to identify pain interference pictorially.” The users can indicate the activities they can’t do on their bad days and notify the app when they are having more interference from their pain.
Healthier Habits with the Advent of Online Grocery Shopping
The Small Data Group is interested in preventative behaviors that can improve and maintain health. One sector they’ve chosen to tackle is household grocery shopping habits.
Estrin says that she found an interesting opportunity with the advent of online grocery shopping.
“This project allows a nutritionist, dietician, or wellness coach to engage their clients to create an improved food environment at home,” she says. “Online grocery shopping is planned as opposed to impulsive.”
“For small data analytics, we’re not looking for patterns that emerge across a large population of patients, but instead the effective modes that the individual patient is in over time.”
The planning is incorporated into a prototype system called Pushcart. Once a user signs up for Pushcart, the service automatically analyzes receipts from online grocery purchases. It then seamlessly creates a nutritional analysis of the purchased food, displayed through data visualizations.
Users can set personal health goals from Pushcart’s library of options, which were created with the help of nutritionists, physicians, and cognitive behavioral therapists. Pushcart will then offer advice for future grocery purchases, like swapping out items for healthier versions.
The goal is to guide people toward a healthier behavior and then help them sustain the behavior by having objective quantitative data, Estrin explains. “It gives people the ability to be more analytical about their own behaviors,” she says.
What the Language of Email Can Reveal
A longer-term research effort out of Estrin’s group is to mine the language used in emails for health measurements. Our emails could possibly reveal fluctuations in cognitive performance, fatigue, side effects of medication or poor sleep, and other conditions and treatments that are typically self-reported and self-medicated.
“How do you take the technology of natural language processing and apply it to a very rich, but very noisy data stream?” is the question Estrin and her colleagues aim to answer.
From the computer science side, the researchers are developing the tools to securely access and effectively process a large, personal data set of emails and to figure out which language features to pull out. It could be grammar, repetition of words, sentiment analysis, or something else entirely.
What matters will also depend on what a patient or clinician is looking for. Estrin says that the group is working with neuroscience experts in aging to determine what signs might point to cognitive impairment and age-related diseases such as dementia and Alzheimer’s. At the same time, natural language processing of emails could be used to understand, and in turn better manage, attention-deficit/hyperactivity disorder.
The researchers are in the beginning stages of the project, which will require large amounts of work and collaborations. Estrin’s lab is currently seeking funding for the project.
A Health Tech Hub at Cornell Tech
Estrin is the founder of the Health Tech Hub at Cornell Tech, which launched an MS degree in Health Tech in August 2015. The hub and the degree bring in people with technical depth (typically Computer Science backgrounds) and provide the contextual information around healthcare that is needed to make useful products. Naturally, given its home, it’s an entrepreneurial program.
“Those three elements are so needed in this space in particular,” says Estrin. “They’re needed in New York City and across the country; it’s going to be a truly unique program, and I believe that it will be successful enough that it won’t be unique for long as others will replicate the model.”
Estrin says that she first became interested in the health tech space in 2007 because she saw so many opportunities for improvement.
“Given that the population has embraced using digital tech to help run their lives from entertainment to productivity, we have an opportunity to use that same technology to help manage health,” she says. “That is, however, much easier said than done. You can’t just say, I can see a way to do it, let me go write some code. The problems are harder than that and just a simple app isn’t going to help solve a problem.”
That’s why Estrin works closely with healthcare providers to identify the most important challenges on which technologists such as herself and her lab will have the most impact. The team collaborates with experts across the health-care industry, including clinicians, providers, researchers, payers, insurance companies, pharmaceutical companies, and patient groups.
Estrin says that Cornell Tech is an ideal home for the new Health Tech Hub because of the vision of the campus: creating a research and educational context that has entrepreneurship and external engagement at its core.