Corti AI

Published by Daniel at August 1, 2023

Corti was founded in 2016 by Andreas Cleve Lohmann, Lars Maaløe, and Michael Boesen, who felt that improving interactions between dispatchers and callers who use emergency services was necessary. These scientists had ample experience developing neural networks in Google and NASA, so they decided to implement machine learning to create a tool that empowered physicians to respond successfully to the challenges that arise in the highly stressful environment of an emergency call, where life-saving decisions are made in seconds. Today, Corti has headquarters in Copenhagen, and the technology has reached multiple countries, including Sweden, the USA, France, New Zealand, and Australia. (1,2,) The company aims to make a meaningful impact in healthcare, enabling physicians to count on practical and faster solutions that can significantly improve patients’ outcomes globally and reduce misdiagnosis of cardiac arrest cases in the prehospital period by 43%. (3)

Intel directly from the source

With a multidisciplinary team of machine learning specialists, engineers, mathematicians, and neuroscientists, Corti aims to build a unique platform that aids in diagnosing critical cases based on large amounts of clinical data. The company has gathered information directly from some of the best emergency departments worldwide, including the European Emergency Number Association, the Fire Department of Seattle, the Wellington Free Ambulance Service in New Zealand, and the SOS Alarm system in Sweden. (3,4,5)

The company has gathered information directly from some of the best emergency departments worldwide, including the European Emergency Number Association, the Fire Department of Seattle, the Wellington Free Ambulance Service in New Zealand, and the SOS Alarm system in Sweden. (3,4,5)

A resourceful tool in times of need

This company’s research focuses on neural network architectures for two essential aspects of natural processing language: image recognition and automatic speech recognition. They test Residual Networks, Highway Networks, and Densely Connected Networks. The software takes insights for models built by machine learning framework data-driven from the entire population to analyze different conversations’ components, including context, patterns, and background noises that correlate with historical data in the system. The algorithm was built to understand dialogue in a challenging acoustic environment and to improve End-To-End audio and imagining processing. After catching key aspects that correlate with the diagnosis, the software makes predictions and offers feedback to the dispatcher for case management. By this premise, Corti trained Audia, a platform that assists medical call-takers when detecting out-of-hospital cardiac arrest during emergency calls by integrating audio, video, and text in consultations.

The algorithm was built to understand dialogue in a challenging acoustic environment and to improve End-To-End audio and imagining processing. After catching key aspects that correlate with the diagnosis, the software makes predictions and offers feedback to the dispatcher for case management. By this premise, Corti trained Audia, a platform that assists medical call-takers when detecting out-of-hospital cardiac arrest during emergency calls by integrating audio, video, and text in consultations. This tool has proven very efficient in trials by making triage 25% faster, decreasing errors by 50%, optimizing performance by reducing call-handling time, improving medical recommendations, and understanding health trends; it also complies with GDPR and HIPAA. (3,6,7,8)

Three solutions for several problems

Corti offers three leading solutions developed in multiple cities in Denmark and recently incorporated in other cities around Europe, increasing the amount of clinical data available for the algorithm. These solutions are call center triage, analytics, and the newest addition, a specialized COVID-19 center. The triage department was the first tool offered by Corti and includes live recommendations, protocols, and symptom detection for any clinical cases, from Lyme Disease to Cardiac Arrest. The call analytics section tracks performances and quality assurance with each call, using AI technology to collect information that ensures resource maximization accurately.

Finally, COVID-19 solutions offer a triage questionnaire associated with COVID-19 symptoms. Simultaneously, the Audia platform analyzes patient consultations in different formats with text, video, and audio to detect whether the user is at a high risk of COVID-19 and points to the correct management for each case. This section has been used at Harborview Medical Center in Seattle since the beginning of the pandemic and has a high success rate with patients. (2,4,9)

The triage department was the first tool offered by Corti and includes live recommendations, protocols, and symptom detection for any clinical cases, from Lyme Disease to Cardiac Arrest. The call analytics section tracks performances and quality assurance with each call, using AI technology to collect information that ensures resource maximization accurately. Finally, COVID-19 solutions offer a triage questionnaire associated with COVID-19 symptoms. Simultaneously, the Audia platform analyzes patient consultations in different formats with text, video, and audio to detect whether the user is at a high risk of COVID-19 and points to the correct management for each case. This section has been used at Harborview Medical Center in Seattle since the beginning of the pandemic and has a high success rate with patients. (2,4,9)

Tackle the problem before it arrives

While Corti has expanded into different sections since 2016, heart disease remains the most common cause of death worldwide, and it is important to identify cardiac arrest in the prehospital setting successfully. A recent study suggested that each minute lost in a heart attack without beginning CPR leads to a 7-10 % drop in survival rates. Accordingly, Corti can determine factors that suggest a cardiac arrest, signaling the dispatcher to advise the caller to begin CPR. The technology can diagnose cardiac arrest emergency calls with 92% accuracy in less than 50 seconds, reducing the error rate. Annually, in Denmark, there are approximately 3,500 out-of-hospital cases of cardiac arrests, 30,000 in the UK, and above 350,000 in the USA. Clinical trials suggest that Corti has accurately diagnosed 93% of those compared to the 73% accuracy achieved by human operators. It is a clear example of how technology could transform the Emergency Department and save hundreds of lives. (10,11,12 )

Annually, in Denmark, there are approximately 3,500 out-of-hospital cases of cardiac arrests, 30,000 in the UK, and above 350,000 in the USA. Clinical trials suggest that Corti has accurately diagnosed 93% of those compared to the 73% accuracy achieved by human operators. It is a clear example of how technology could transform the Emergency Department and save hundreds of lives. (10,11,12 )

A second opinion in line

Audia and the other solutions provided by Corti were developed to act as virtual assistants for healthcare providers worldwide and do not aim to replace human interaction by any means. This technology will significantly benefit physicians and patients as it improves triage flow, reduces the burden on healthcare systems, tracks key symptoms to optimize protocols, provides remote monitoring, and boosts the patient experience. All these are essential factors in spreading the advances of machine learning equally across society. (13)

Under the spotlight in Europe

This innovative company has earned several awards and distinctions right from the start, including the “The Danish AI Startup Award” in 2017 by Global Startup Awards, the “Best Global AI Innovation” by VentureBeat in 2018, and the award to the most profitable corporation in Europe by DIGITALEUROPE in 2020. (13,14)

References