Cincinnati Predictive Analytics Project Takes Aim at Emergency Medical Services

By identifying and predicting when and where EMS calls are likely to occur, the city will be better positioned to improve its offerings.

by / August 9, 2016
Members of Cincinnati city staff meet with members of the University of Chicago's Data Science for Social Good during a project focused on improving emergency medical service through predictive data analytics. Courtesy University of Chicago Data Science for Social Good Program/City of Cincinnati

From the initial 911 call to the emergency response that follows, emergency dispatching is a well-choreographed process designed to ensure that each incident has the attention and equipment needed for saving lives.

But even with constant practice and training, emergency response amounts to an educated guess — in which lives are on the line. The dispatcher’s questions and the responses they trigger are the basis for everything that happens thereafter. While it may seem like simple resource management from an outside perspective, doing it well is a battle against time and the limitations that play into each call for help.

But Cincinnati is working toward a better way to deploy critical emergency services across the city through its collaborative data analytics project.

The Big Idea

With the help of the University of Chicago’s Data Science for Social Good program, city officials are hoping to take some of the uncertainty out of the dispatch and emergency response process, and provide citizens with even more reliable services.

Unlike some fire departments who contract services out to a third party, Cincinnati assumes the duties of first responders and medical transport for what Assistant Chief Anson Turley said equates to higher quality service.  

For perspective, the city of roughly 300,000 people has 12 medical transports, otherwise known as ambulances, and 26 engine companies positioned throughout the city. Each engine company is also staffed with paramedics. But these staffing levels may not be enough in the long run, Turley said — especially in relation to emergency medical services (EMS).

“EMS has become the lion share of our workload. Back in the '60s and '70s, fire departments mainly fought fires, right? Well, progressively, fires have decreased to such an extent that for the Cincinnati Fire Department, emergency medical service is more than 80 percent of our workload,” he said. “We have a finite resource, though, in the fact that we only have so many first responding companies, those 26 engine companies, and we only have 12 medical transport units, yet our workload keeps growing year over year.”

The overarching goal of the partnership with the university’s data science program will be to look at raw data collected by the city and translate it into operational opportunities.

By identifying and predicting when and where calls are likely to occur based on historical data, or callers known by the department as “frequent flyers,” the city will be better positioned to improve their offerings.

Paul van der Boor is the university’s project manager and sees the project as a way to aide departmental efficiencies while reducing the number of inaccurate medical transport calls by predicting which instances are likely to require them.

Through a close working relationship that includes regular communication and a ride along with first responders, the university team and the city outlined the important questions hiding in the unformatted data.

“The reason that is an interesting question to them, or an important question to them, is that they have a scarce resource, which is the medical transport units," he said. "There are only 12 of them, and they get deployed on various incidents. And sometimes they get called in when a transport to the hospital isn’t required, and sometimes they don’t get called in on time when a medical transport is required."

But identifying response requirements is just one benefit of the project. Cincinnati’s Chief Performance Officer Leigh Tami said the department will also likely see institutional improvements related to staffing and retention as a result.

One of the many problems facing EMS and first responders is burnout. Unbalanced workloads and frequent runs make them an easy victim for high turnover rates. Tami is hopeful the predictive models provided by the data science team will also translate into better retention.

“The other way the fire department really reaps benefit from this is medic retention. A huge, huge problem with the medic on medic transport units is burnout. They are really only supposed to be doing between eight and 11 or 12 runs a day…” she said. “When your burnout rates for medics are high, then you need to find a way to recruit more paramedics, you have to take paramedics off engines, it's really hard on people who literally see people in their worst state, and that’s what they do all day. So we are trying to make sure the … paramedics and the EMTs with them are not overburdened…”

The data has already offered some insight into the future, according to Assistant Chief Turley. As he sees it, the city will likely need to hire more medical transport units to combat the increasing workload and burnout factor.

According to city statistics, medical transport incidents increased from 45,000 incidents in 2012 to 54,000 in 2015. Turley said these numbers are expected to increase even more.

“One thing that we’ve learned from the data is that at some point in time, we are probably going to have to add more medic transport units because of the workload, because of the burnout that Leigh talked about," he said, "and the fact that the work just keeps increasing and increasing."

Both Turley and City Manager Harry Black said that the collaboration could very well bring economic benefits in the form of cost savings, though they agree that public safety is the primary motivation.

“The idea is to optimize the deployment of these resources because we want to improve response times and all of the thing associated with improving response times, i.e., we save more lives, we provide better care etc., etc.,” Black said. “Savings is not really a focus of this exercise. If we generate some economic savings, that’s great, but that’s not why we are doing this. This is all about service delivery optimization.”

It’s Worked Before

This project won’t be the first collaboration between the city and Data Science for Social Good. Last year, the academic team and city officials studied the indicators of urban blight and created a model that could identify the early warning signs of inner-city decay.

Black said predictive tools allow governments to look ahead proactively and highlight problems in advance, effectively “shaping a result to be more positive.”

“It’s about being able to predict streets or street segments that are vulnerable to becoming blighted way, way before it actually happens," he said. "You look at a lot of different factors. You look at health department data, our buildings and inspections department, you look at the fire department and other departments, and what they were able to do for us last year is to help us to develop a digital model … where if one key initial event occurs, we are almost certain that 15 other events are likely to occur. It gives us the opportunity to recognize it and intervene."

According to Black, predictive analytics represents the next wave in local government and what he calls the “ultimate value added” proposition.

As it stands, the university team is constructing models for a late August handoff. As with the blight prediction project, van der Boor said the team will remain in close contact with the city of Cincinnati to ensure operational usability.

Eyragon Eidam Web Editor

Eyragon Eidam is the Web editor for Government Technology magazine, after previously serving as  assistant news editor and covering such topics as legislation, social media and public safety. He can be reached at eeidam@erepublic.com.