Big Data, Big Future

From left: Drs. Lietman, Nair, Deiner, Porco, Acharya, and Stewart

UCSF vision scientists use the mountain of digitized information called “big data” to advance sight for all.

Technological innovations have converged to create big data. UCSF’s vision scientists now routinely use massive, diverse, and searchable datasets to help answer vital research questions more quickly and clearly than ever.

 

Big Data Expertise

The Francis I. Proctor Foundation for Research in Ophthalmology and the Department of Ophthalmology are jointly establishing a big data group, led by data scientist Michael Deiner, PhD. “This reflects UCSF leadership, experience, and commitment in this specialty,” says Tom Lietman, MD, the Foundation’s director.

 

Google, Twitter, and Global Epidemics

Billions of people use their phones, tablets, and laptops to look up symptoms or seek information about health and illness. Google alone provides what biostatistician Travis Porco, PhD, calls “oceans of data,” with about 5.5 billion searches per day. Twitter users generate about 500 million tweets each day.

Drs. Porco, Deiner, and Lietman collaborate on two major big data studies. They use data mining, statistical spatiotemporal analysis, data visualization, machine learning, and topic modeling to investigate whether internet-based search terms, social media posts, and large clinical datasets can assist in detecting and tracking contagious outbreaks and epidemics.

The teams’ machine classification of Twitter posts yielded common terms related to infectious pink eye.

“Poor Baby Has Pink Eye.” For example, through machine classification of millions of social media posts about pink eye, the team found common phrases such as “poor baby has pink eye.” These phrases guide them towards relevant online content to identify possible outbreaks.

Each year in the United States, over 3.5 million school days and 8.5 million workdays are estimated to be missed due to pink eye, with clinical costs estimated to exceed $800 million. Although the redness, burning, itching, pain, and/or swelling often clear up on their own, untreated complications can lead to sight loss.

“If information from social media, online forums, and internet searches can complement traditional approaches for identifying infectious eye outbreaks more quickly and accurately, health agencies may be able to intervene more effectively,” says Dr. Deiner. Alerting people to stay at home or take other simple measures can stem the tide of eye contagion, keeping more children and adults happy and healthy.

 

Trachoma Forecast

The team’s second big data project is aimed at tracking epidemics of trachoma. To more accurately forecast if and when the contagious blinding disease can be brought under control worldwide, Drs. Leitman, Deiner, and Porco are developing the most in-depth meta-analysis yet by mining all major databases that track trachoma. UCSF’s Proctor Foundation has long worked to banish the disease from deeply affected rural areas in Africa.

 

Genetic Targets to Halt Glaucoma

By the time Christie Hastings noticed blurring in her visual field, her sight already had been permanently compromised. For patients like Christie, glaucoma treatments that target retinal cells to protect the optic nerve from degeneration promise to transform outcomes. Toward this aim, geneticist and glaucoma researcher Saidas Nair, PhD, and his team leverage a powerful database that links the genetic profiles of more than 110,000 individuals with their health records.

The database was codeveloped by the UCSF Institute for Human Genetics and Kaiser Permanente Northern California. With a median patient age of 65 years, the database is ideal for investigating age-related diseases like glaucoma. In a broad initiative to advance understanding of glaucoma and novel solutions to save sight, Dr. Nair’s team joins clinician scientists, computer scientists, biologists and epidemiologists. Eric Jorgenson, PhD, and Ron Melles, MD (Kaiser Permanente Northern California), were instrumental in characterizing the glaucoma patient groups, which facilitated Dr. Nair’s genetic studies of the retina.

Analysis has already yielded a list of common genetic variants that predispose individuals to develop glaucoma. Combining these findings with their research on laboratory glaucoma models, Dr. Nair’s team identified a set of genes that works through retinal nerve cells, thus contributing to optic nerve degeneration. This group is now exploring how these genes operate to induce damage to the optic nerve. Better understanding will suggest novel targets for glaucoma treatment.

 

Can a Diabetes Drug Help AMD?

Grateful patient Betty Toal with Dr. Jay Stewart

When patient Betty Toal was diagnosed, a major research breakthrough allowed vitreoretinal specialist Jay Stewart, MD, to treat her wet age-related macular degeneration (AMD). “I am grateful because he saved my sight,” says Betty.

No treatments yet exist for most AMD patients – those with the much more common dry type. Untreated, AMD slowly reduces the detailed central sight needed for facial recognition and reading.

“There is a pressing need for novel therapies to halt AMD,” says Dr. Stewart. He hypothesizes that the anti-inflammatory effects of metformin, a common diabetes treatment, may help, since inflammation is believed to play a role in AMD. His curiosity and commitment to eye patients drive the research.

AMD is the leading cause of blindness in people 55 and older in developed countries.

Big data provides a window for understanding metformin’s potential for AMD. The Intelligent Research in Sight (IRIS) Registry (41 million health records) allows Dr. Stewart and coinvestigator Jeremy Keenan, MD, MPH, to identify a huge number of patients with diabetes and AMD and identify patients that were treated with metformin.

The team will harvest new knowledge to inform AMD treatment from any differences between patients that took metformin and those that did not.

“If the investigation shows metformin to be promising, we will build on this inquiry,” says Dr. Stewart. “We’re keenly aware that preserving sight increases people’s choices in old age.” As part of the growing wave of aging Americans with AMD, Betty is eager for solutions that will help more seniors. Never one to sit on the sidelines, she has made a bequest to That Man May See in support of Dr. Stewart’s research.

 

An Eye on Shingles

Older people fear shingles (herpes zoster) for good reason—the painful viral infections can be long lasting, and risk increases with age. Ocular shingles can lead to chronic or blinding conditions.

“Big data lets us answer key questions about how available vaccines affect patients with ocular shingles,” says uveitis specialist Nisha Acharya, MD. The latest vaccine requires two shots, prevents more than 90 percent of infections, and reduces duration, pain, and discomfort in cases that still occur. Her team will use the Optum database (81 million health records) to compare shingles cases in vaccinated and unvaccinated patients. New understanding of the impacts may mobilize public health agencies to more actively promote shingles vaccinations.

 

Power Tools for Discovery

Many UCSF vision researchers now capitalize on big data, and methodologies range from metagenomic deep sequencing (Thuy Doan, PhD) to deep learning neural networks (Luca Della Santina, PhD).

“No matter the eye disease, big data offers new pathways for understanding and intervening, personalizing treatment, and advancing healthy sight worldwide,” says Department Chair Stephen D. McLeod, MD.

Research support is provided by the National Institutes of Health, Research to Prevent Blindness, OptumLabs, American Academy of Ophthalmology, Foundation Fighting Blindness, and That Man May See.