Cerebral Near-infrared Spectroscopy in Patients Undergoing Open Cardiovascular Surgery Using Artificial Intelligence Programs: A Methodological Study

Overview

The aim of the present study is to test the accuracy and reliability of AI tools in this regard by interpreting NIRS, which we routinely use for monitoring cerebral perfusion in cardiovascular surgery, with AI tools and comparing this interpretation with the interpretations of two clinicians.

First described more than 40 years ago, this technology is a non-invasive monitoring method based on the fact that near infrared light can penetrate biological tissue, providing real-time and non-invasive information about tissue oxygenation and metabolism. Furthermore, in a clinical setting, NIRS can provide clinicians with potentially valuable information in patients with systemic and/or cerebral microcirculation disorders. NIRS is used to assess brain oxygenation, as well as to monitor local tissue and muscle oxygenation and tissue perfusion. Clinical applications of this technology include cerebral and cardiac monitoring, detection of global or regional low cardiac output status (LCOS), evaluation of treatment response, and prognosis determination. Open heart surgery is a complex surgery requiring a multidisciplinary approach. During this surgery, the heart is usually stopped, and its function is performed by the heart pump. During this time and throughout the entire operation, vital organs must be closely monitored, and adequate tissue oxygenation must be ensured. The human brain is one of the organs with the highest oxygen consumption. Therefore, it is extremely sensitive to hypoxic conditions. Furthermore, it has been noted that decreased brain oxygen saturation may be associated with postoperative neurological dysfunction and worse neurocognitive outcomes. Cerebral NIRS is increasingly being used in clinical settings to monitor brain oxygenation. This new and non-invasive technique can also function as a warning sign of cerebral ischemia and hypoxia. Additionally, real-time brain oxygenation monitoring with cerebral NIRS can be useful in detecting ischemic events during cerebrovascular procedures. Furthermore, NIRS-based management strategies aim to promote tissue oxygen delivery (including optimizing oxygenation and hemoglobin concentration and the intelligent use of intravenous fluids and inotropic drugs) while reducing tissue oxygen consumption (by reducing metabolic demand, including analgesics, sedation, and temperature control). This method, evaluated and administered only by anesthesiologists, allows us to routinely monitor brain oxygenation during the procedure and interpret it clinically, enabling us to intervene accordingly. With the increasing use of artificial intelligence (AI) tools today, the literature contains studies analyzing the clinical use and suitability of the most frequently used AI tools, such as Chat GPT (GPT-5.5), Google Gemini (Gemini 3.1 Pro), and Microsoft Copilot (365 Copilot Premium). However, the number of such studies in the literature is quite limited. Therefore, the aim of the present study is to test the accuracy and reliability of AI tools in this regard by interpreting NIRS, which we routinely use for monitoring cerebral perfusion in cardiovascular surgery, with AI tools and comparing this interpretation with the interpretations of two clinicians.