trachAI

Half of the approximately 30,000 patients who are admitted to Danish ICUs every year require mechanical ventilation because they can’t breathe on their own. The standard procedure involves intubating the patient through their mouth to connect their lungs to a ventilator. If the patient is expected to require mechanical ventilation (MV) for an extended period, a tracheostomy is often placed. This involves cannulating the patient’s windpipe (trachea) to help them breathe while protecting their vocal cords and surrounding anatomy. While this procedure is more comfortable and mitigates the risks associated with intubation, it’s a highly invasive procedure that carries a non-trivial risk of complications.

The Inspiration Behind the Innovation

Physicians lack objective metrics to help them assess whether a patient will require MV for longer than a week. Therefore, the process of determining whether a tracheostomy should be placed is a subjective one that lacks precision. ICUs collect super-rich samples of real-world data from a variety of sources (e.g. ventilators, monitors, drug infusion pumps), creating the perfect foundation for machine learning and AI-based prediction models to assist in decision-making.  

The Innovation 

trachAI uses performant explainable AI software to learn from hundreds of parameters on tens of thousands of patients, and flag those who will require extended MV. The solution, which is accessible on a variety of devices, is available in standalone and web-based versions as well as an integrated version that automatically captures data from patient files. By enabling physicians to make better-informed decisions about the use of tracheostomies, trachAI should result in shorter and less taxing ICU stays, thereby ensuring the most efficient use of scarce ICU capacity while improving patient outcomes and reducing morbidity. The team believes that in time, the software could be extended with other AI models to inform decisions related to other patient-important outcomes.

The Team

Benjamin Skov Kaas-Hansen: MD, Postdoc; Department of Intensive Care, Copenhagen University Hospital – Rigshospitalet

Hans-Christian Thorsen-Meyer: MD and Specialist in Intensive Care, Postdoc; Department of Intensive Care, Copenhagen University Hospital – Rigshospitalet

Davide Placido: Biomedical Engineer, Postdoc; NNF Centre for Protein Research, University of Copenhagen

Søren Brunak: Full Professor; NNF Centre for Protein Research, University of Copenhagen

Anders Perner: MD and Senior Specialist in Intensive Care, Full Professor; Department of Intensive Care, Copenhagen University Hospital – Rigshospitalet