Human error is a major contributor to surgical risks. In fact, according to a 2019 JAMA Network Open study, more than half of adverse events occurring during surgical procedures resulted from human error, with just over half of these errors being cognitive in nature. Since the level of training and confidence have a direct correlation with the surgical outcome, the EU-funded BESAFE project aims to enhance surgical technologies. To do this, it is using artificial intelligence (AI) to identify and flag high-risk procedures based on real-time analysis of the user´s patterns of behaviour. The specific goal of the project was to reduce the risk of intra-operative accidents using add-on software that embeds AI into intraoperative neuromonitoring (ION) systems. ION systems are surgical devices that alert surgeons about ongoing damage to nerves, thus allowing them to quickly alter the procedure and avoid causing irreversible harm.
Learning from individual behaviour
BESAFE builds on the work done by the EU-funded IBSEN FET project, which developed AI tools based on unsupervised machine learning algorithms to discover patterns in human decision-making. By integrating the AI developed in the IBSEN FET project into the BESAFE software, the project hopes to be able to prevent surgical accidents. “The goal is for the software is to learn from the behaviour of individuals using ION systems and the outcomes of the surgical procedures,” says a spokesperson for the project. “It will then use this knowledge to automatically detect high-risk actions by members of the surgical team while in the operating room.” BESAFE is unique in that it observes the user’s clicks on the system’s touch screen to gauge the probability of human-caused accidents involving the ION technology.
Opening the door to new opportunities
The BESAFE concept was evaluated using a mock-up application. From this evaluation, researchers confirmed that embedding AI into a touch screen user interface could distinguish between users operating under an inadequate cognitive load and those who are over-stressed or under-trained. “Those who use the touch screen following certain patterns are recognised by the AI-enhanced software module as being high risk and corrective measures are proposed to anticipate and avoid accidents,” adds the project spokesperson. Based on these positive findings, researchers have developed a comprehensive business plan for bringing the solution to market. “This project opened the door to new opportunities, not only from the industrial side, but also as an exciting line of the research in academia,” concludes the spokesperson. “The path towards a final product has been laid out – a path that can ultimately benefit the well-being of patients.”
BESAFE, intraoperative neuromonitoring, artificial intelligence, AI, surgical procedures, surgeons, machine learning