Tag Archive for Analytics

At Issue: Analytics

“A popular government without popular information or the means of acquiring it, is but a prologue to a farce, or a tragedy, or perhaps both.” — James Madison

Analytics to solve civic problems is not — in its simplest form — anything new. In 1854, for example, cholera broke out in an area of London, eventually killing some 600 residents. Doctors thought that cholera was spread by “miasma,” a fouling of the air. But one physician, John Snow, drew a map of the area and put a mark where each of the afflicted lived. Snow discovered that most of the sick lived within a few blocks of the Broad Street public well, but there were some exceptions. He investigated the exceptions and found that some stricken children that did not live near the pump attended school in the Broad Street area, and had used the suspect pump. And some who lived near the pump but were unaffected drank only beer.

By mapping the outbreak and investigating those exceptions, Snow came to the conclusion that the water was somehow responsible for the cholera outbreak. The pump handle was removed and the outbreak faded away. It was later discovered that sewage pits beneath houses had overflowed into the water supply. The actual bacterium responsible for cholera was spotted — also in 1854 — but was not widely known until much later. So a simple analysis of data led to a conclusion that not only saved lives in the 1854 outbreak, but provided information useful to cities throughout the world.

A more recent example of information in the service of public safety is something called Comstat or Compstat. Back in 1990, a New York City subway cop named Jack Maple began to map where crime occurred in the subway. Which stops, what time of day, etc. These maps, which he called “Charts of the Future” were done on 55 feet of butcher paper. The charts helped predict where and when crimes were most likely to occur, so officers could be assigned accordingly. Between 1990 and 1992, they helped cut subway felonies and robberies by nearly one third. NYPD Commissioner William Bratton later incorporated the system into all NYPD operations and today, police departments around the world use Compstat. The original butcher paper charts of the future were not very sophisticated by today’s standards, but they showed where and when subway crime was likely to occur, and they changed policing forever.

Then in the mid-2000s, a doctor in Camden, N.J., fed up with violence in the city, took a page from Comstat and with a student intern, began mapping hospital data on accidents and injuries. In a lengthy interview for the Frontline television program, Dr. Jeffrey Brenner described the same kind of meticulous data collection and mapping as Dr. John Snow had conducted in long-ago London.

“I went to the hospital that I worked for,” said Brenner in the interview, “and submitted a proposal to collect patient-level information for everyone who had had an accident or injury, so this included people who had been shot, people who had been assaulted or fallen down the stairs. We got the data and began … We mapped it, graphed it, charted it, and it was just an unbelievable data set … It pretty quickly became clear that there were hot spots of everything. There were hot spots by disease, hot spots by patient; there were certain patients who had been [admitted] over and over and over. There were hot spots by ZIP code and by neighborhood … And because I knew the city so well, you could begin to take the data and tell stories with the data. And that’s an incredibly powerful tool for making change.”

Brenner widened his criteria, asking more hospitals for claims data for a full year. The patterns showed “hot spot” buildings where elderly and disabled people lived, who made repeated emergency room visits. Not only is an emergency room visit expensive — from $1,000 to $3,000 — but repeated ER visits are not good health care, and without follow up, patients often ended up in the hospital again within a few weeks.

Brenner discovered that one percent of the residents of Camden were responsible for 30 percent of the hospital and emergency room costs. And the reasons for those visits were, in order of occurrence, head colds, ear infections, sore throats, asthma and stomach viruses. Surprisingly, other groups of patients, including the insured, had a similar pattern.

Now that he had the data, Brenner went into action, figuring if that one percent could be helped outside the ER, it would mean a huge drop in costs, and more personalized care. In 2007, he built a team of home health workers to look in on the patients and help with seemingly minor preventive care such as filling prescriptions for asthma medication, which resulted in huge drops in medical costs, ER visits and hospitalizations, and provided better more personalized health care for those individuals.

The common denominator in each of these cases was the collection and analysis of data which resulted in a strategy leading to action to solve some problem. This approach — now called “analytics” — is increasingly of use in cities and counties to focus on some of local government’s biggest problems such as health, crime, infrastructure, and other difficult challenges.

Recently, however, revelations of the massive collection of surveillance data by U.S. security agencies has caused a backlash against “big data” which could change the dynamics of analytics. For more information on big data and analytics, watch for a special section in the September issue of Government Technology magazine.

Photo: At Pacific Northwest National Laboratory, the science of cyber analytics supports better predictions and guides adaptive responses of computers and computer networks. Courtesy of Flickr/Pacific Northwest National Laboratory.

View the original article here

Opinion: Can Google Analytics Replace Public Health Workers?

As big data becomes more sophisticated, the question about whether analytics tools — like those developed by Google — could someday replace data compiled by federal agencies.

A case in point is the Google Flu Trends website, which could someday become a suitable replacement for much of the work now performed by the Center for Disease Control (CDC).

That day, however, is not likely to come for a least a few years, according to a recent opinion article in the National Journal, although many agree there is great potential in such technology if it can be honed.

A comparison of Google’s flu trend graph and the CDC’s data shows a discrepancy in findings:

Google Flu Trends map
Google Flu Trends chart

CDC flu trends chart

While Google shows the current flu outbreak as being the worst of the past six years, the Center for Disease Control shows that the current outbreak is bad, but not as bad as outbreaks of at least two years past. Furthermore, the CDC graph shows that the outbreak is already on the decline. The discrepancy can be accounted for if one looks at the simplistic nature of Google’s data, according to the article.

The CDC uses a combination of data: reports of sickness across many disease control centers along with tweaks made by public health experts with years of experience. Google’s data is based off of search results that do their best to filter out search noise, but with limited success. The things that’s missing, according to the article? The human factor. And Google admits the tool is still in an early phase of development; it has a ways to go before it can compete with human data analysis.

“We intend to update our model each year with the latest sentinel provider [influenza-like illness] data, obtaining a better fit and adjusting as online health-seeking behavior evolves over time,” Matt Mohebbi, a Google software engineer recently wrote for Forbes. “With respect to the current flu season, it’s still too early to tell how the model is performing.”

But in the future, we could see a combination of the two, said Lynnette Brammer, a flu epidemiologist with the CDC. While there may never be a substitute for human decision-making, technology could save public heath workers a lot of time, she said. “We want the data transmission to be as easy for the people providing it to us as possible,” she said. “But the thing we don’t want is to lose the connection we have with those people. Even if you have really good data coming in, you’re always going to have questions about what it means.”

When comparing the two systems, one primarily run by people and the other by a machine, it comes down to understanding complexity. “It’s really hard, certainly for us at CDC, to understand what’s causing that change,” Brammer said. “They’re seeing pretty much record levels of influenza-like illness. And while ours are high, they’re not at historical limits by any means. We just have a lot more flexibility and ability to track down and ask additional questions and find the answers to those questions.”

Main photo courtesy of Shutterstock

You may use or reference this story with attribution and a link to

View the original article here