A new project at Penn Medicine in Philadelphia is trying to use big data to keep lung cancer patients out of the emergency room.
By using a variety of measures such as recent lab tests and data culled from electronic medical records, Penn Medicine hopes to come up with a formula and a related score that will predict the chances of a lung cancer patient ending up in the ER. Think of it as a form of proactive intervention with the help of data-crunching computers. The project is just in the pilot stage, but right now the formula is able to predict about one out of every three cancer patient visits to the ER.
“We’re trying to get an actual score for a patient so we know their risk of ending up in the ER, and such a risk calculator is continuously running in the background,” said Dr. Abigail Berman, an assistant professor of radiation oncology at Penn Medicine, in a recent phone interview.
To get to that score is a two-fold process. First, there’s a retrospective look at historical data of each patient’s past trips to the ER to assess why those trips happened. Then that past activity is used to create a model to estimate a patient’s risk of going back to the ER. If chances for a return trip to the ER are high, the patient’s physician is notified electronically.
“The main thing we’re trying to do is train a computer algorithm that would recognize patterns in a patient’s EMR, a pattern of activity that seems to precede people going to the ER in the historical data,” said Dr. Peter Gabriel, the chief oncology informatics officer at Penn Medicine’s Abramson Cancer Center. “This computer algorithm is constantly running, and once it picks up on a pattern, it sends an alert to that patient’s physician.”
In this case, that’s the reason for blending cancer medicine with big data, the buzzword that signifies a copious amount of data that’s difficult for humans to comb through, but simple for computers to make mincemeat out of.
A Deep-dive Into Specialty Pharma
A specialty drug is a class of prescription medications used to treat complex, chronic or rare medical conditions. Although this classification was originally intended to define the treatment of rare, also termed “orphan” diseases, affecting fewer than 200,000 people in the US, more recently, specialty drugs have emerged as the cornerstone of treatment for chronic and complex diseases such as cancer, autoimmune conditions, diabetes, hepatitis C, and HIV/AIDS.
“The computer could see something that the physician might not pick up,” Dr. Gabriel said.
Why lung cancer patients are the first to try out this new system owes itself to the severity of this cancer. According to Drs. Berman and Gabriel, lung cancer patients have more acute problems, like shortness of breath, and therefore tend to go to the ER more than people with prostate or breast cancer, for instance. In addition, patients with lung cancer sometimes have other underlying conditions like heart disease or diabetes that land them in the ER while undergoing treatment for lung cancer.
“Lung cancer patients are among the sickest of cancer patients,” Dr. Gabriel said. “Part of what we want to learn from the project is to characterize the reasons why people go to the ER and be able to make some judgments about which ones are preventable and which ones aren’t.”
For that reason, the pilot project is operating in concert with the recently opened Oncology Evaluation Center, which is meant to act as an interim step between a patient with an urgent problem and the ER. Essentially a hybrid between a clinic and an ER, the OER is a clinic space within Penn Medicine where patients can report to for urgent evaluation.
Say a doctor participating in the project is alerted that their patient has a high chance of winding up in the ER soon. A visit to the OER would be the next step. Nurse practitioners work closely with physicians to run tests on lung cancer patients who arrive with the aim of being able to send a patient home at day’s end instead of sending them on to the ER.
“For all the patients we do know, we take it very seriously to have them live their normal life, which means not in the hospital,” Dr. Berman said. “We want to keep those patients out of the ER if at all possible.”
The OER is reminiscent of the step the Johns Hopkins Hospital in Baltimore, Maryland, took several years ago. In 2014, Hopkins opened an urgent care center specifically for cancer patients, who call a hotline and walk in if nurses can’t address their problems over the phone. But Penn Medicine is going an extra step by pairing that goal with a big-data approach to anticipate when those visits to the OER are necessary and then notifying medical personnel.
“This lung cancer project is an example of how we can take the EMR system and make it more useful so that it presents information that’s more helpful to physicians, which will help them make better decisions,” Dr. Gabriel said.
Using big data informatics in oncology treatment the way Penn Medicine is doing is still a fairly new concept, More typically the intersection of big data and oncology happens in genomics, where the goal is to figure out which genes are important to the development or progression of different types of cancers and what treatments will work.
Whether at Penn Medicine or Johns Hopkins, however, the endgame is the same: Get cancer patients urgent help while preventing them from sitting through long waits at the ER.
“We’re trying to anticipate the problem before it manifests at 3 in the morning,” Dr. Gabriel said.
