European Journal of Cardiovascular Medicine

The history of anesthesia began in 1846 when William Morton used ether for surgery. Since then, anesthesia has grown into a safe and advanced specialty. In the 20th century, tools such as pulse oximetry and capnography improved monitoring, making surgeries much safer. In recent years, we are now seeing another big change with the use of artificial intelligence (AI).

AI is the science of making machines simulates human thinking-learning, reasoning, and decision-making. In healthcare, it covers many technologies:

• Machine learning (ML): Finds patterns in patient data and predicts outcomes.
• Deep learning (DL): Uses layered neural networks to study complex signals like EEG or imaging.
• Natural language processing (NLP): Reads and analyzes text in medical notes and electronic records.
• Computer vision: Interprets medical images and video for procedures such as airway assessment or ultrasound-guided blocks.

Anesthesia produces a huge amount of data. Monitors give second-by-second updates on heart rate, blood pressure, oxygen, and carbon dioxide. Records contain lab results, imaging, and notes. Human doctors may struggle to process so much data, but AI can quickly analyze it and provide useful predictions.

Already, AI-based tools like the Hypotension Prediction Index (HPI) can predict low blood pressure before it happens. Platforms like MySurgeryRisk can estimate the chance of complications. Despite these advances, many challenges remain, including privacy, bias and questions about who is responsible if something goes wrong.¹

This review explains the foundations of AI in anesthesia, its applications before, during, and after surgery, and its role in ICU, clinical decision support, challenges, and possible future directions.

Foundations of AI in Anesthesia

1. Machine Learning (ML)
   ML algorithms analyze perioperative data such as patient age, comorbidities and intraoperative variables. These models can predict complications, cluster patients into groups or adjust drug doses. Reinforcement learning a special branch of ML allows AI to learn optimal dosing by trial and error over time.²
2. Deep Learning (DL)
   DL uses multi-layer neural networks to analyze complicated data. Convolutional neural networks (CNNs) can recognize facial patterns for airway risk. Recurrent neural networks (RNNs) can capture time-based changes in vital signs or brain signals.³
3. Natural Language Processing (NLP)
   Anesthesiologists produce large amounts of text in preoperative notes, intraoperative records and ICU documentation. NLP helps extract useful details such as detecting adverse events, summarizing records, and improving research databases.⁴
4. Computer Vision
   Computer vision allows machines to see. In anesthesia, it can analyze ultrasound images for nerve blocks, video laryngoscopy for airway management, or bronchoscopy feeds for lung inspection.^5,6

Together, these tools support better prediction, monitoring, and automation in anesthesia.

Applications of AI in Anesthesia

1. Preoperative Assessment and Risk Stratification

Before surgery, doctors must assess risks. Traditionally, ASA classification and clinical judgment are used. AI improves this by using larger datasets.

• Risk Prediction: The MySurgeryRisk platform predicts problems like sepsis, kidney injury, or death better than traditional scores.⁷
• Airway Prediction: AI systems can analyze face images or ultrasound scans to predict difficult intubation with more than 85% accuracy.⁸
• Personalized Plans: AI combines genetic, lab, and imaging data to create tailored anesthesia strategies.⁹

2. Intraoperative Monitoring and Decision Support

• Depth of Anesthesia: Traditional BIS monitors are limited. AI can analyze EEG more accurately to reduce the risk of awareness and optimize drug use.¹⁰
• Hemodynamics: AI tools like HPI can predict low blood pressure up to 15 minutes before it happens, giving anesthesiologists time to act.¹¹
• Ventilation: AI can predict oxygen desaturation and adjust ventilation strategies to prevent hypoxemia.¹²

3. Automated Drug Delivery Systems

Closed-loop anesthesia systems are among the most exciting uses of AI.

• Propofol and Remifentanil: AI-based systems titrate these drugs automatically, keeping BIS levels stable and reducing variability.¹³
• Neuromuscular Blockade: AI optimizes muscle relaxant dosing and reversal, lowering the chance of residual paralysis.¹⁴

4. Robotic Airway Management

AI-assisted robots can help in difficult intubations by guiding tubes based on computer vision analysis.¹⁵

5. Regional Anesthesia and Pain Management

AI enhances ultrasound imaging by identifying nerves and tracking needles, improving success and reducing dependency on operator skill.¹⁶

For postoperative pain, AI-driven pumps adjust opioid doses according to patient responses, reducing side effects while keeping pain under control.¹⁷

6. Postoperative Complication Prediction

AI can forecast complications such as nausea, delirium or even death. Neural networks have been shown to predict mortality more accurately than current scoring methods.¹⁸

7. AI in Intensive Care and Perioperative Medicine

In the ICU, AI predicts sepsis, respiratory failure, and helps in ventilator weaning.

It can also suggest optimal sedation and pain control plans.^19,20

CDSS are tools that provide real-time advice to anesthesiologists.

• Drug Guidance: Suggests the right drug and dose using patient comorbidities and intraoperative data.²¹
• Predictive Alerts: Warns about upcoming hypotension, hypoxia or arrhythmias.²²
• Workflow: Improves operating room efficiency by predicting surgery duration and reducing cancellations.²³
• ICU Integration: Supports decisions on sedation, ventilation, and sepsis care.²⁴

Challenges in Implementing AI in Anesthesia

1. Data Privacy: Patient data must be protected under laws such as HIPAA and GDPR. Large datasets need encryption and anonymization.²⁶
2. Bias: If datasets lack diversity, predictions may be less accurate for minorities leading to unequal care.²⁷
3. Validation: Many AI tools work well in retrospective studies but fail in real clinical use. Prospective trials are needed.²⁸
4. Medico-Legal Issues: Doctors remain responsible for outcomes, raising questions about liability when AI is involved.²⁹
5. Integration: Hospital IT systems often cannot easily connect with AI platforms, slowing adoption.³⁰
6. Cost: Developing and maintaining AI is expensive, and poorer countries may be left behind.³¹
7. Training: Doctors must learn the basics of AI to use these systems properly.³²

Future Directions of AI in Anesthesia

1. Personalized Anesthesia: Using genetics and metabolism data, AI could tailor exact drug doses for each patient.³³
2. Fully Autonomous Systems: Prototypes like McSleepy show that fully automated anesthesia delivery is possible, though ethical and legal issues remain.³⁴
3. AR and VR Integration: AI-guided overlays could highlight anatomy in real time during blocks or intubations. VR simulators could train residents effectively.³⁵
4. Federated Learning: Allows AI to train across hospitals without sharing raw data, protecting privacy.³⁶
5. Continuous Learning Systems: Adaptive AI can keep improving with every case, making it more reliable over time.³⁷
6. Medical Education: AI tutors and simulators can personalize resident training and speed up learning.³⁸

Artificial intelligence is changing anesthesiology in major ways. From risk prediction and monitoring to drug delivery and ICU care, AI offers safer, more accurate, and more efficient solutions. But challenges in privacy, fairness, cost and responsibility must be solved before AI can be fully trusted in everyday practice.

The future will likely bring personalized anesthesia, autonomous delivery, and advanced AR/VR guidance. With careful planning, validation, and ethical use, AI can support anesthesiologists in providing safe, compassionate, and patient-centered care.

1. Deo RC. Machine learning in medicine. Circulation. 2015 Nov 17;132(20):1920‑30.
2. Komorowski M, Celi LA, Badawi O, Gordon AC, Faisal AA. The Artificial Intelligence Clinician learns optimal treatment strategies for sepsis in intensive care. Nat Med. 2018 Nov;24(11):1716‑20.
3. LeCun Y, Bengio Y, Hinton G. Deep learning. Nature. 2015 May 28;521(7553):436‑44.
4. Kreimeyer K, Foster M, Pandey A, Arya N, Halford G, Jones SF, et al. Natural language processing systems for capturing and standardizing unstructured clinical information: a systematic review. J Biomed Inform. 2017 Sep;73:14‑29.
5. Esteva A, Kuprel B, Novoa RA, Ko J, Swetter SM, Blau HM, Thrun S. Dermatologist‑level classification of skin cancer with deep neural networks. Nature. 2017 Feb 2;542(7639):115‑18.
6. Yoon CH, Torrente‑Rodriguez RM, Muñoz‑Price LS, Korley FK. Machine learning in anesthesiology: current techniques, applications, and future directions. Anesth Analg. 2020 Oct;131(4):998‑1015.
7. Bihorac A, Ozrazgat‑Baslanti T, Ebadi A, Motaei A, Madkour M, Pardalos PM, Lipori G, Hogan WR, Hobson CE. MySurgeryRisk: development and validation of a machine‑learning risk algorithm for major complications and death after surgery. Ann Surg. 2019 Sep;270(3):484‑93.
8. Arulkumaran N, Harrison EM, Naidu B. Artificial intelligence in perioperative medicine: current status and future directions. Anaesthesia. 2021 Jan;76 Suppl 1:60‑67.
9. Johnson KB, Wei WQ, Weeraratne D, Frisse ME, Misulis K, Rhee K, Zhao J, Snowdon JL, Denny JC. Precision medicine, AI, and the future of personalized anesthesia care. Anesth Analg. 2021 Mar 1;132(3):711‑18.
10. Chazot T, Huybrechts I, Law‑Koune JD, Barvais L, Fischler M, Sessler DI. Closed‑loop co‑administration of propofol and remifentanil guided by bispectral index: a randomized multicenter study. Anesth Analg. 2011 Feb;112(2):546‑57.
11. Wijnberge M, Geerts BF, Hol L, Lemmers N, Mulder MP, Berge P, Schenk J, Terwindt LE, Hollmann MW, Vlaar AP, Veelo DP. Effect of a machine learning‑derived early warning system for intraoperative hypotension vs standard care on depth and duration of intraoperative hypotension during elective noncardiac surgery: the HYPE randomized clinical trial. JAMA. 2020 Jan 14;323(11):1052‑60.
12. Lundberg SM, Nair B, Vavilala MS, Horibe M, Eisses MJ, Adams T, Liston DE, Low DK, Newman SF, Kim J, King‑Wai Low D, Lee SI. Explainable machine‑learning predictions for the prevention of hypoxaemia during surgery. Nat Biomed Eng. 2018 Oct;2(10):749‑60.
13. Rinehart JB, Liu N. Closed‑loop vasopressor administration: science or fantasy? Anesth Analg. 2017 Jan;124(1):340‑45.
14. McKinley BA, Bailey JM. Computer‑controlled closed‑loop drug infusion in anesthesia and critical care. Anesth Analg. 1999 Jul;89(1):141‑52.
15. Fiadjoe JE, Kovatsis PG. Robotic‑assisted intubation: the next frontier in airway management? Anesthesiology. 2019 Aug;131(2):209‑11.
16. Shibata SC, Yoon JH, Park EY, Lee DK. Application of artificial intelligence to regional anesthesia: current state and future perspectives. Korean J Anesthesiol. 2021 Dec;74(6):495‑504.
17. Yoon S, Lim YJ, Jeon YT, Park HP. Artificial intelligence in postoperative pain management: present and future. Korean J Anesthesiol. 2020 Feb;73(1):12‑21.
18. Futoma J, Simons M, Panch T, Doshi‑Velez F, Celi LA. The myth of generalisability in clinical research and machine learning in health care. Lancet Digit Health. 2020 Sep;2(9):e489‑e492.
19. Komorowski M, Celi LA, Badawi O, Gordon AC, Faisal AA. The Artificial Intelligence Clinician learns optimal treatment strategies for sepsis in intensive care. Nat Med. 2018 Nov;24(11):1716‑20.
20. Topol EJ. High‑performance medicine: the convergence of human and artificial intelligence. Nat Med. 2019 Jan;25(1):44‑56.
21. Sutton RT, Pincock D, Baumgart DC, Sadowski DC, Fedorak RN, Kroeker KI. An overview of clinical decision support systems: benefits, risks, and strategies for success. NPJ Digit Med. 2020 Jan 30;3:17.
22. Sendak MP, Ratliff W, Sarro D, Alderton E, Futoma J, Gao M, Nichols M, Rehder K, Sandhu H, Balu S. Real‑world integration of a sepsis deep learning technology into routine clinical care: implementation study. JMIR Med Inform. 2020 Jul 15;8(7):e15182.
23. Hashimoto DA, Witkowski E, Gao L, Meireles O, Rosman G. Artificial intelligence in anesthesiology: current techniques, applications, and future directions. Anesthesiology. 2020 Feb;132(2):379‑94.
24. Rajkomar A, Dean J, Kohane I. Machine learning in medicine. N Engl J Med. 2019 Apr 4;380(14):1347‑58.
25. Price WN, Cohen IG. Privacy in the age of medical big data. Nat Med. 2019 Jan;25(1):37‑43.
26. Obermeyer Z, Powers B, Vogeli C, Mullainathan S. Dissecting racial bias in an algorithm used to manage the health of populations. Science. 2019 Oct 25;366(6464):447‑53.
27. Wynants L, Van Calster B, Collins GS, Riley RD, Heinze G, Schuit E, Bonten MM, Dahly DL, Damen JA, Debray TP, de Jong VMT. Prediction models for diagnosis and prognosis of COVID‑19: systematic review and critical appraisal. BMJ. 2020 Apr 7;369:m1328.
28. Gerke S, Minssen T, Cohen G. Ethical and legal challenges of artificial intelligence‑driven healthcare. In: Artificial Intelligence in Healthcare. Academic Press; 2020. p.295‑336.
29. Kelly CJ, Karthikesalingam A, Suleyman M, Corrado G, King D. Key challenges for delivering clinical impact with artificial intelligence. BMC Med. 2019 Dec;17(1):195.
30. Wahl B, Cossy‑Gantner A, Germann S, Schwalbe N. Artificial intelligence (AI) and global health: how can AI contribute to health in resource‑poor settings? BMJ Glob Health. 2018 Aug 1;3(4):e000798.
31. Chan KS, Zary N. Applications and challenges of implementing artificial intelligence in medical education: integrative review. JMIR Med Educ. 2019 Jun 15;5(1):e13930.
32. Hemmerling TM, Terrasini N. Robotic anesthesia: not science fiction anymore. Curr Opin Anaesthesiol. 2012 Dec;25(6):736‑42.
33. Pottle J. Virtual reality and the transformation of medical education. Future Healthc J. 2019 Jun;6(3):181‑85.
34. Rieke N, Hancox J, Li W, Milletarì F, Roth HR, Albarqouni S, Bakas S, Galtier MN, Landman BA, Maier‑Hein K, Ourselin S. The future of digital health with federated learning. NPJ Digit Med. 2020 Sep 14;3:119.
35. Wiens J, Saria S, Sendak M, Ghassemi M, Liu VX, Doshi‑Velez F, Jung K, Heller K, Kale D, Saeed M, Ossorio PN. Do no harm: a roadmap for responsible machine learning for health care. Nat Med. 2019 Dec;25(9):1337‑40.