Healthcare Analytics, Population Health Management, Healthcare Big Data

Analytics in Action News

Using Machine Learning, Health IT to Improve Patient Safety

MedStar Health and the Pennsylvania Safety Authority are leveraging machine learning tools to mine patient safety data and more rapidly identify actionable insights.

Using machine learning, health IT to improve patient safety

Source: Thinkstock

By Jessica Kent

- Across the care continuum, all healthcare organizations are continuously seeking new and innovative ways to improve patient safety. 

Medication errors, hospital-acquired conditions, and preventable deaths have always topped the list of events to avoid, and as artificial intelligence and machine learning have crept into each part of the industry, health systems have started leveraging these tools to learn from past patient safety incidents. 

“For quite some time, our team has been really focused on developing more intuitive, advanced, and scalable ways of analyzing patient safety event data,” Raj Ratwani, PhD, director of the MedStar Health National Center for Human Factors in Healthcare, told HealthITAnalytics.com

“Most healthcare systems across the country collect this information. If it’s used well, the data can help organizations identify safety hazards, and then remedy those safety hazards quickly. But we've seen that people don't always have the necessary tools and skills to analyze all that data.”

Raj Ratwani, PhD Source: Xtelligent Healthcare Media

Pennsylvania’s Patient Safety Authority (PSA) is well-acquainted with these data-related challenges. Home to one of the largest patient safety databases in the country, the PSA collects information on every type of event, from those that have been shown to cause harm to those that could potentially cause harm.

READ MORE: New Partnership Uses Machine Learning to Improve Data Security

“We get about 300,000 events reported into our database every year,” said Regina Hoffman, executive director of the PSA. 

“The healthcare facilities that are recording these events face a lot of challenges, not just with patient safety but also with resources. They spend a tremendous amount of time reporting this information in our database, so it's critical that that we’re analyzing the data we have from these organizations as best we can.”

To better detect critical patient safety trends, PSA has joined forces with MedStar Health Research Institute (MHRI) to apply advanced machine learning tools to its vast database. 

“It's just not humanly possible for the PSA to read through every single event and try to identify patterns in the data,” Ratwani explained. 

“Machine learning provides the opportunity to have a human code or analyze a subset of this data, and then scale that information across millions of reports. We believe that process of semi-automated pattern recognition and trend identification will lead to substantial increases in patient safety.”

READ MORE: Using Big Data Analytics for Patient Safety, Hospital Acquired Conditions

Using advanced algorithms, MHRI will be able to analyze the free text of patient safety reports on a broader scale and at a faster speed, accurately categorizing issues and obtaining valuable information that otherwise may have been underutilized. The partnership will further expand on PSA’s efforts to boost patient safety efforts across the state, Hoffman said.

“We've been collecting event reports since 2004, and we've educated over 70,000 healthcare providers across the commonwealth based on that information,” she noted. “We've published close to 600 advisories. And now we're at the point where we’re ready to take the next step. What else is in that data that we're not seeing?”

Regina Hoffman, Executive Director of the PSA Source: Xtelligent Healthcare Media

Hoffman and Ratwani expect that advanced analytics tools will help answer that question. And they aren’t the only ones: Across the care continuum, providers, payers, and IT developers are buzzing about the potential for these technologies to revolutionize the industry, allowing organizations to draw more actionable conclusions from multiple sources of data. 

“There's huge promise for machine learning in healthcare,” said Ratwani.

“If clinicians use the technology on the front lines, it could lead to improved clinical decision support and new information at the point of care. But in order for that to happen, the insights generated through machine learning have to be presented to the clinician in a way that's intuitive. We need to provide information to the clinician that's going to effectively guide their decision-making.”

READ MORE: How Patient Safety Organizations Encourage Data Collection, Quality Care

Ratwani noted that this will depend heavily on how technology is designed, specifically that they’re user-centric.

“Well-designed products ensure that our clinicians can seamlessly interact with the technology, and help them deliver better, safer care,” Ratwani said.

“In contrast, a poorly designed piece of technology will force the clinician to think deeply about how they use that device. It increases the likelihood that the clinician is going to make some kind of error. And that has direct patient safety consequences.”

Data corroborates Ratwani’s stance.The 2019 ECRI Patient Safety Concerns list cited organizational challenges with EHR use, data stewardship gaps, and poor communication as the top issues facing health systems and providers, indicating that clinicians are still struggling to effectively leverage their IT tools.

In another study, published in JAMA last year, researchers found that EHR usability was often the culprit in cases of patient harm. 

Organizations and federal agencies like the Pew Charitable Trusts and ONC have consistently addressed health IT usability and patient safety issues, and discussions about bringing AI or machine learning into everyday clinical care usually end in conversations around protecting patient safety.

“Technology plays a huge role in patient safety,” Hoffman echoed. “EHRs have completely changed communication. Some changes are very positive, but new technology also comes with new challenges. Now, it’s not just making sure you connect with other providers face-to-face, it’s making sure that systems connect with each other, too.”

“As clinicians, I think this is one of the things that make us nervous. If you're a nurse, you may not know how to fix an IT problem. You may not even recognize that there is an IT problem.”

To overcome these challenges, MedStar and the PSA will work to develop technology that is intuitive for clinicians. 

“We're really looking to build tools and capabilities that are more useful for people, and to make sure that those solutions are sustainable,” said Ratwani. “To improve safety, we have to take a systems approach that is focused on making changes to the environment, as opposed to just retraining individuals.”

Ratwani noted that this will require user-centered design methods, which could incorporate workflow analysis, clinician usability testing, and continuous user support.

“If we know that the outcome from a machine learning algorithm is going to be used by clinicians, we make sure we get clinicians involved early in the process,” he said. 

“That might sound strange, because we’re developing algorithms, but if the clinicians are going to be the end user, they should shape what the outcomes of that algorithm look like. That's taking a true user-centered design approach.”

Developing health IT solutions with providers in mind will ensure machine learning and other AI tools augment clinicians’ abilities and act as complementary decision support. 

“We firmly believe that clinicians want to understand why a machine learning algorithm is coming to some conclusion. They generally don't want to follow a recommendation that's coming out of a black box,” Ratwani said. 

“Ultimately, the clinician is going to be responsible for the decision that they make. We have to disclose a little bit about how the technology is making a decision when we provide machine learning guidance. And it has to be done in a way that’s intuitive and fits into clinicians’ workflows.”

Through this partnership, both MedStar Research Institute and the PSA will aim to make strides in patient safety and care with advanced analytics technologies.

“Using machine learning, MedStar can bring new insights into the data we already have and help us make better decisions about how to improve care,” Hoffman said. 

“We're incredibly excited about the partnership with the PSA,” Ratwani added. 

“The PSA is thinking deeply about how to improve safety in the entire State of Pennsylvania, and they are taking advantage of advancements in computer science that are going to improve their abilities. Through this partnership, we can make tremendous improvements in patient safety.”

X

Join 25,000 of your peers

Register for free to get access to all our articles, webcasts, white papers and exclusive interviews.

Our privacy policy


no, thanks

Continue to site...