Doctor Stories
AI Algorithm Can Predict and Help Treat Hypotension in the OR and ICU
04.14.2025
A new AI-driven device developed by a Stanford Medicine team aims to support the prediction and prevention of hypotension in the OR and ICU to improve patient outcomes. Hypotension is a regular occurrence in up to 99% of surgeries and can lead to higher levels of morbidity and mortality after surgery. While hypotension is one of the only modifiable risk factors during surgery, overtreatment can also lead to poor outcomes.
Louise Sun, MD, MS, chief of cardiothoracic anesthesiology and professor of anesthesiology, perioperative, and pain medicine at Stanford University School of Medicine, pioneered the use of informatics techniques to help surgical teams tailor their blood pressure management approaches for individual patients. As a section lead for the Anesthesia Patient Safety Foundation, an expert consensus group responsible for defining perioperative hypotension, Dr. Sun and colleagues recently provided recommendations for managing perioperative hemodynamic instability characterized by abnormal blood pressure.1
CARE AT STANFORD
One of the nation's best for Cardiology, Heart & Vascular Surgery.
Visit Cardiovascular Health
Louise Y Sun
Chief of Cardiothoracic Anesthesiology and Professor of Anesthesiology, Perioperative and Pain Medicine
Throughout her career, Dr. Sun has been instrumental in helping define hypotension in both noncardiac and cardiac surgeries. 鈥淚n our studies, we defined the critical thresholds and durations of hypotension for death and many end-organ complications, including kidney failure and stroke,鈥 said Dr. Sun.
In noncardiac surgeries, her team defined intraoperative hypotension in association with acute kidney injury post-surgery that has been widely adopted into clinical practice. In cardiac surgeries, her team found that hypotension during the bypass period resulted in a higher risk of postoperative stroke,3,4 whereas hypotension after the bypass period was associated with higher risks of acute kidney injury, dialysis,5 and death.6
鈥淣ow, we鈥檙e looking at how to predict those episodes of low blood pressure before they happen, so we can do something to prevent them,鈥 said Dr. Sun. 鈥淲e also wanted to have a personalized way of supporting patients鈥 blood pressure so we can provide timely and appropriate treatment without overtreating.鈥
To predict what a patient鈥檚 blood pressure will be on a minute-by-minute basis during and after surgery, Dr. Sun鈥檚 team developed a personalized algorithm to predict future blood pressure values, with the help of AI and deep learning. The versatile algorithm can be readily integrated into a standalone monitoring device (see illustration) or into existing algorithms for integrated heart-brain risk prediction and telemonitoring.
The device also allows clinicians to tailor warning messages for hemodynamic deterioration based on each patient鈥檚 typical blood pressure before surgery, which is especially important if their normal blood pressure is outside autoregulation thresholds.
鈥淭he device gives you a warning message, telling you what the predicted blood pressure is and if there is risk of any related health event,鈥 said Dr. Sun. 鈥淭here are different ways the alarm can be set so that you can receive them remotely and only need to know when it鈥檚 a particularly bad episode.鈥
The patent for the device, called Oculus, is currently being examined in multiple countries. In the future, Dr. Sun expects the algorithm will help determine the cause of predicted hypotension. It can then provide an advisory message with suggestions for the clinician to consider, to complement their clinical judgement.
Read more about the future of cardiovascular anesthesiology at Stanford Medicine.
1 Scott, MJ; the APSF Hemodynamic Instability Writing Group. Perioperative Patients With Hemodynamic Instability: Consensus Recommendations of the Anesthesia Patient Safety Foundation. Anesthesia & Analgesia 138(4):p 713-724, April 2024. | DOI: 10.1213/ANE.0000000000006789
2 Sun LY, Wijeysundera DN, Tait GA, Beattie WS. Association of Intraoperative Hypotension with Acute Kidney Injury after Elective Noncardiac Surgery. Anesthesiology 123(3):p 515-523, September 2015. | DOI: 10.1097/ALN.0000000000000765
3 Sun LY, Chung AM, Farkouh ME, van Diepen S, Weinberger J, Bourke M, Ruel M. Defining an Intraoperative Hypotension Threshold in Association with Stroke in Cardiac Surgery. Anesthesiology 129(3):p 440-447, September 2018. | DOI: 10.1097/ALN.0000000000002298
4 Rios-Monterrosa J, Sun LY. Hypotension and Perioperative Strokes in Cardiac Surgery: How Big Data Can Help Answer Big Questions. Semin Thorac Cardiovasc Surg. S1043-0679(25): 00013-9, March 4, 2025. | DOI: 10.1053/j.semtcvs.2025.02.004. Epub ahead of print.
5 Ngu JMC, Jabagi H, Chung AM, Boodhwani M, Ruel M, Bourke M, Sun LY. Defining an Intraoperative Hypotension Threshold in Association with De Novo Renal Replacement Therapy after Cardiac Surgery. Anesthesiology 132(6):p 1447-1457, June 2020. | DOI:10.1097/ALN.0000000000003254
6 Ristovic V, de Roock S, Mesana TG, van Diepen S, Sun LY. The Impact of Preoperative Risk on the Association between Hypotension and Mortality after Cardiac Surgery: An Observational Study. J Clin Med. 9(7):p 2057, June 30, 2020. | DOI: 10.3390/jcm9072057