How can we optimize treatments? How do we ensure doctors are following best practices? Ahus knew that its clinical records could hold the answers, but lacked a way to unlock the power of its data.
Working with Capgemini, Ahus is using cognitive content analytics from IBM to mine thousands of doctors’ reports and confirm that teams are meeting high standards around patient care and safety.
6,800+radiology reports analyzed in-depth for the first time
99%content analytics accuracy rate leads to new diagnosis and treatment insights
Revealsdiagnosis and treatment insights via structured and unstructured data analytics
Business Challenge story
Digging deeper into clinical data
The healthcare industry is deluged by data, and clinical records—from vitals and charts to test results and patient histories—are growing explosively. This wealth of information holds valuable insights that can be used to drive care decisions, but sifting through it to identify effective treatment options can take up enormous amounts of time for busy medical professionals.
For Ahus, one of Norway’s premier teaching hospitals, this is a familiar challenge. The organization gathers huge volumes of data on patients and treatments, but as much of this information was locked away in unstructured, textual reports, it was too difficult and time-consuming for teams to extract meaningful insights from the data.
Dr. Petter Hurlen Chief Physician Clinical ICT, Division for Diagnostics and Technology at Ahus, elaborates: “We treat tens of thousands of patients every year, and our top priority is to provide each patient with the very best care. To achieve this target, it is important that our doctors follow best practices when it comes to diagnosing and treating patients.
“Our team is made up of seasoned healthcare professionals, and we trust that they deliver a very high standard of patient care. At the same time, we were looking for way to back up our doctors’ experiences with hard data. This meant looking at our patient records and doctors’ reports in detail to confirm that quality standards were being followed properly and identify areas where diagnosis and treatment could potentially be improved.”
He continues: “The challenge was that we lacked an easy way to perform such quality assurance checks. Gathering the necessary information required combing through thousands of complex clinical documents—a task that was impossible for our team to complete manually. This meant that valuable insights were potentially going unnoticed, robbing us of the opportunity to optimize care delivery.”
Turning to cognitive content analytics
After a timely encounter with IBM Platinum Business Partner Capgemini, Ahus found a solution to its challenges. Dr. Hurlen recalls: “Ahus has enjoyed a close relationship with Capgemini over the years. I happened to be visiting their offices in Oslo on the day that Capgemini announced a new initiative centered around IBM Watson solutions, and I instantly knew that Ahus had to get involved.”
He continues: “We had been testing some open-source analytics solutions to help with our quality assurance work, but they came up short in many areas. When I learned more about the capabilities of IBM Watson Explorer, I was convinced that it offered exactly what we were looking for. What’s more, we could draw on Capgemini’s in-depth technical knowledge to make sure we got the most out of the solution. The combination of IBM Watson’s cognitive capabilities and Capgemini’s expertise was a very powerful proposal for us.”
After discussing its requirements with Capgemini, Ahus identified an initial use case for IBM Watson Explorer: analyzing when CT scans were performed on pediatric patients in emergency situations, and whether this usage fell within recommended guidelines.
CT scans can be life-saving in critical circumstances, but the radiation can also be potentially harmful, making it important for Ahus to avoid overuse of scans. By using Watson Explorer, Ahus was able to gather unstructured content from radiology reports, then apply machine-learning and natural language processing techniques to discover how often CT scans were undertaken and the findings of those scans.
Ahus extracted over 5,000 anonymized radiology reports from the hospital’s records system and transferred them to a local server in a secure information zone. By following these steps, the university ensured compliance with strict data privacy and governance rules.
“We worked with Capgemini to create a unique lexicon for Watson, training the solution to understand medical terminology and specific Norwegian words,” says Dr. Hurlen. “We also created a classification schema, teaching Watson, among other things, to distinguish files that reported positive scan results and those that reported negative results, and categorize the data accordingly.
“After several tests, we reached an accuracy level of 99 percent for content classification in Watson Explorer. The final analysis confirmed that the frequency of our CT scanning is at an acceptable level, and that we are striking the right balance between the probability of positive gains in relation to the potential harmful effects.”
Following the success of this pilot project, Ahus recently kicked off a new quality assurance study, looking at treatment pathways for prostate cancer.
Dr. Hurlen explains: “In 2015, the Norwegian government introduced the Cancer Patient Pathways program, which aims to drive a more efficient and consistent approach to diagnosis, treatment and cure. We wanted to investigate how effective the new inclusion criteria were at determining whether patients actually had cancer. If the diagnostic criteria are too broad, then too many patients might be wrongly included on the pathway—wasting valuable resources and causing unnecessary worry for patients who might be led to believe they have cancer.”
Choosing prostate cancer pathways as a test case, Ahus and Capgemini are using Watson Explorer to analyze more than 1,800 radiologists’ reports—using the results of CT and MRI scans and medical records to group patients according to their likelihood of having cancer. The team will then compare these diagnostic results against the number of patients included on the prostate cancer pathway to determine whether the inclusion criteria are being applied correctly and consistently.
Revealing new insights
Thanks to its collaboration with IBM and Capgemini, Ahus is breathing new life into its clinical data and opening up fresh insights for the hospital.
“It simply would not have been possible for us to conduct these kind of quality assurance studies without Watson Explorer,” confirms Dr. Hurlen. “It would take a team of people months and even years to analyze the same amount of data that IBM Watson can process in minutes. Watson Explorer allows us to take very large sets of unstructured data and examine it in an intelligent, efficient way. This enables us to give our content a new lease on life, and find fresh meaning in masses of clinical data that previously went untapped.”
By making it feasible for Ahus to take on in-depth quality assurance studies, cognitive technologies are empowering the hospital to maintain very high standards of patient care and safety.
Dr. Hurlen notes: “The work we are doing with IBM Watson Explorer and Capgemini has given us a very systematic way of seeing how our medical teams approach diagnosis, treatment and patient interaction. Armed with these insights, we can confirm that our people are following recommended best practices, and more easily identify areas for improvement. All of this helps Ahus to make sure that its resources are being utilized in the most effective way and that we are delivering the very best possible quality of care to the patients we serve.”
He concludes: “We have only scratched the surface of what is possible with cognitive content analytics and we are keen to explore new opportunities. By bringing together the experience of Capgemini with the power of IBM Watson, we can unlock deeper insights into our clinical data, and use that knowledge to shape smarter operations and world-class patient experience.”
About Akershus University Hospital
Akershus University Hospital (Ahus) is a Norwegian public university hospital serving approximately 500,000 inhabitants around Oslo. A teaching hospital, Ahus is affiliated with the University of Oslo and employs 9,500 people.
- Watson Discovery
Take the Next Step
To learn more about IBM Watson Explorer, please contact your IBM representative or IBM Business Partner, or visit the following website: ibm.com/marketplace/content-analytics
A global leader in consulting, technology and outsourcing services, Capgemini employs more than 190,000 people and maintains a presence in over 40 countries. Together with its clients, Capgemini creates and delivers business, technology and digital solutions that fit their needs, enabling them to achieve innovation and competitiveness.