AI to assist cardiac arrest decision-making

Doctors can now access an AI-based decision support tool to assist in managing cardiac arrest patients. By entering relevant data in a web-based app, appropriately skilled physicians can discover how thousands of similar patients have fared, thanks to researchers at the University of Gothenburg, Sweden.

Three systems of decision support have been developed and it is hoped that these may make a significant difference to doctors’ work in the future. The clinical prediction model, known as SCARS-1, is presented in The Lancet’s eBioMedicine journal and available to download free of charge from the Gothenburg Cardiac Arrest Machine Learning Studies website.

The app accesses data from the Swedish Cardiopulmonary Resuscitation Register on 55,000 patient cases. The Gothenburg researchers used an advanced form of machine learning to teach clinical prediction models to recognise various factors that have affected previous outcomes. The algorithms take into account numerous factors relating to the cardiac arrest, such as treatment provided, previous ill health, medication and socioeconomic status.

Research head Araz Rawshani, a researcher at Gothernburg’s Sahlgrenska Academy and resident physician in cardiology at Sahlgrenska University Hospital, said, “Both I and several of my colleagues who treat emergency patients with cardiac arrest have already started using the prediction models as part of our process for deciding on the level of care.

“The answer from these tools often means we get confirmation of views we’ve already arrived at. Still, it helps us not to subject patients to painful treatment that is very unlikely to be of benefit to the patient, while saving care resources.”

The model indicates whether a new patient case resembles previous cases and offers information about whether the previous patient had survived or died 30 days after their cardiac arrest.

The researchers said the model’s accuracy is unusually high. Based on the 10 most significant factors, the model has a sensitivity of 95% and a specificity of 89%.

The ‘AUC-ROC value’ (ROC being the receiver operating characteristic curve for the model and AUC the area under the ROC curve) for this model is 0.97. The highest possible AUC-ROC value is 1.0 and the threshold for a clinically relevant model is 0.7.

This decision support element was developed by Fredrik Hessulf, a doctoral student at the Sahlgrenska Academy and anaesthesiologist at Sahlgrenska University Hospital/Mölndal.

“This decision support is one of several pieces in a big puzzle: the doctor’s overall assessment of a patient. We have many different factors to consider in deciding whether to go ahead with cardiopulmonary resuscitation,” Hessulf said.

This form of support is based on 393 factors affecting patients’ chances of surviving their cardiac arrest for 30 days after the event, but 10 factors have been found to be most significant in predicting survival; the most important was whether the heart regained a viable cardiac rhythm again after the patient’s admission to the emergency department.

The second decision support tool published, SCARS-2, has been presented in the journal Resuscitation and will be launched shortly. This tool is based on data from patients who survived their out-of-hospital cardiac arrest until they were discharged from hospital.

The predictive models are based on 886 factors in 5098 patient cases from the Swedish Cardiopulmonary Resuscitation Register. This tool, developed by research doctor Gustaf Hellsén, is partly aimed at helping doctors identify which patients are at risk of another cardiac arrest or death within a year of discharge from hospital.

It also aims to highlight which factors are important for long-term survival after cardiac arrest — an aspect of the subject area that has not been well studied.

“The accuracy of this tool is reasonably good. It can predict with about 70% reliability whether the patient will die, or will have had another cardiac arrest, within a year. Like Fredrik’s tool, this one has the advantage that just a few factors can predict outcome almost as well as the model with several hundred variables,” Hellsén said.

A third decision support tool, SCARS-3, is also planned and this will offer support for doctors treating patients who experience an in-hospital cardiac arrest.

Image credit: iStock.com/Marcus Millo