PATIENTS receiving care after cardiac arrest can benefit from a web-based application, where doctors can enter patient information to learn how thousands of similar patients have progressed. Three such systems of artificial intelligence (AI)-based decision support have been developed by researchers at the University of Gothenburg, Sweden. In the future, this could have a major impact on doctors’ work.
The clinical prediction model SCARS-1 is one of these decision support tools; it is now published and downloadable free of charge from the Gothenburg Cardiac Arrest Machine Learning Studies website. The app accesses data from the Swedish Cardiopulmonary Resuscitation Register, which contains tens of thousands of patient cases. SCARS-1 suggests whether a new patient case resembles other previous cases, where 30 days after their cardiac arrest, patients had either survived or passed away. The algorithms consider numerous factors relating to, for instance, the cardiac arrest, treatment provided, previous ill health, medication, and socioeconomic status.
The model showcased unusually high levels of accuracy. Based solely on the ten most significant factors, the model had a sensitivity of 95% and a specificity of 89%. The model had an AUC-ROC value of 0.97, where ROC refers to the receiver operating characteristic curve for the model and AUC refers to the area under the ROC curve. An AUC-ROC of 1.0 is the highest possible value, and 0.7 is the threshold for a clinically relevant model.
Araz Rawshani, researcher at the University of Gothenburg’s Sahlgrenska Academy, and resident physician in cardiology at Sahlgrenska University Hospital, Gothenburg, oversees the research group working on decision support for cardiac arrest. Rawshani stated: “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 researchers state that they hope to succeed in further developing the prediction model and enhance its precision, while also highlighting that the prediction model can already support doctors in identifying factors with an important bearing on survival among cardiac arrest patients who are to be discharged from hospital. Although the results from the algorithm have to be interpreted by trained professionals, AI-based decision support is actively advancing in many healthcare sectors.
