Raffaele Ornello, MD
Artificial intelligence (AI) is increasingly used in several aspects of everyday life and in medicine, as well. Stroke medicine, in which rapid decisions are required, can benefit from the implementation of AI in terms of decision making and patient safety.
This review by Mouridsen et al. focuses on AI applications in stroke imaging. Machine algorithms can be trained to improve the quality of imaging techniques and sparing radiations for diagnostic tools such as CT perfusion imaging; they can also date stroke onset and differentiate the ischemic stroke core from the ischemic penumbra, thus identifying the patients that can benefit the most from revascularization procedures. Machines can also help clinicians in selecting patients with large vessel occlusion, who benefit from endovascular treatments, and in predicting stroke outcomes, such as hemorrhagic transformation and 3-month outcomes.
In the future, machines will likely perform some tasks for which specific human figures are lacking or acting remotely. Besides, ‘big data’ obtained from current practice might lead to real ‘deep learning’ and innovative research on stroke clinical characteristics and physiopathology.
Machines already assist us in making clinical decisions. At the present time, we are telling machines what we decide, so that they can learn. The future goal is to have machine algorithms collaborate with us in making the best decisions for the patients. Not far from now, machines might become respected members of stroke care teams.
