Key Challenges
Longer call times and increased call load using a dated protocol that no longer reflected best practice.
Key Results
Used Corti to reduce call time by 21% through enforcing and continuously optimizing a leaner medical protocol.
SOS Alarm is Sweden’s nationwide emergency-service provider. They are responsible for more than three million calls annually, across emergency and nursing lines.
A rising population in Sweden, combined with more demand for telehealth and a worker shortage have exacerbated already-existing needs to support call takers through better protocol guidance.
Call takers at SOS Alarm use specific criteria to assess patients, such as health status, breathing, consciousness, and other critical pieces of information.
Before engaging Corti, SOS had moved from paper-guided to digital-guided protocols. Much of their protocols came from one of the world’s top medical institutions. Having taken these protocols wholesale from paper to digital, however, they wanted greater insight into their effectiveness and to identify where they could resolve any weaknesses.
They knew they wanted a new decision-support process and homed in on security, safety, efficiency, and precision as key areas of improvement.
Armed with research into novel technological solutions, SOS Alarm knew that Artificial Intelligence (AI) would be key to solving their needs while maintaining patient safety. This research quickly led them to Corti, a company with a strong track record for implementing AI within high pressure medical consultations across the US and Europe.
Once they had engaged Corti, we used machine-learning and natural-language processing to structure the dialogue and metadata of nine million historic and current SOS Alarm calls in order to provide insights at scale.
We looked at both protocol and call-taker weaknesses. Done manually, this would have taken 10 analysts 250 years. It took us only 15 minutes.
We then used the collected data to optimize protocol, ensuring that questions asked in calls were as effective as possible. With a laser focus on key areas, questions were added or removed and advice was re-structured. Together, we used A/B split-testing to understand when protocol changes drove their desired outcome. This work led to an overall reduction in call time by 21%, translating not only into significant efficiency improvements but also more important, a better experience for patients in need.
These findings don’t need to languish in hours of staff training before SOS Alarm can act on them. Jannice Mattsson, COO of SOS Alarm notes, “Our medical staff use Corti to make protocol changes and distribute those to all operators at the click of a button, saving us time and money on continued staff training.”
Meanwhile, our automated speech recognition discovered where SOS Alarm could strengthen protocol adherence among call takers. We found that nurses had been frequently delivering advice and selecting priorities for calls manually rather than via protocol. This armed SOS Alarm with better focus for training, and also suggested to us where processes could be improved to keep nurses on track.
