Technological innovations like Artificial Intelligence and Machine learning allow the detection of valuable patterns within large datasets which – after been subjected to enough data – allows the algorithms to perform predictions on previously unseen data subjects. Such intelligent software has been used extensively in different fields of the healthcare industry – including neurology, cardiology, and oncology – with the purpose of aiding medical personnel with disease diagnostics, disease prevention, and personalized medical treatment .
However, previous attempts to incorporate machine learning within anesthesiology – which is the field within the healthcare industry that focuses on providing perioperative care to patients – have been unsuccessful . This article will provide an overview of the difficulties that arise when automating the field of anesthesiology.
Difficulties with automating anesthesiology procedures
Generally, systems for the automation of anesthesiology procedures rely on a closed-loop feedback system which is able to successfully keep a quantifiable target measure – usually the bispectral index (BIS) when assessing depth of anesthesia – within a pre-defined range  . Using various drug administration rates – which depend on the measured BIS level – the patient’s level of consciousness can be controlled in an autonomous way.
Various studies have shown that the use of such closed-loop feedback systems could be beneficial in the context of keeping the patient’s level of consciousness within a pre-determined BIS range, with the additional benefit of providing a lower dose of anesthetic in comparison to the human-controlled case   . Whereas there is evidence that closed-loop feedback systems are feasible to assist in guaranteeing required anesthetic levels for both simple and more complex cases , they by no means are able to fully automate the – usually human-controlled – process.
However, innovations such as Artificial Intelligence – which implement a bottom-up rather than a top-down approach like rule-based feedback loops – are able to learn to take the required patient’s level of consciousness actions from real-world patient data without being explicitly programmed to. Whereas these algorithms are able to tackle tasks that are much more complex in comparison to rule-based systems, in practice they still possess flaws which require the need of a professional anesthetist during medical interventions:
- Artificial Intelligence is especially well-suited for performing cognitive tasks (i.e., carrying out accurate predictions and crunching large data sets). However, the technology is yet unable to deliver the dexterity-based labor that is involved with the field of anesthesiology. 
- Artificial Intelligence and Machine Learning – implemented in robotic devices – do not have the finesse to deal with complex tasks such as neural blockades, venous cannulation or tracheal intubation. 
- The field of anesthesiology is characterized by providing micro-doses in order to remain the required level of patient consciousness. However, patients are uncomfortable with the thought of replacing a human anesthetist with fully autonomous decision-making software without human control.
Whereas current procedures – such as rule-based systems or artificial intelligence – are yet unable to fully take over human anesthetic tasks, they are thought to play a major role in the future of anesthesiology. Computer software – powered by artificial intelligence – will ultimately aid in all decisions made by anesthetist and, when innovations in robotics allow it, take over dexterity-based labor as well.
 Murali, Nivetha & Sivakumaran, Nivethika. (2018). Artificial Intelligence in Healthcare-A Review. 10.13140/RG.2.2.27265.92003.
 Alexander, J. C., & Joshi, G. P. (2018, January). Anesthesiology, automation, and artificial intelligence. In Baylor University Medical Center Proceedings (Vol. 31, No. 1, pp. 117-119). Taylor & Francis.
 Kissin, I. (2000). Depth of anesthesia and bispectral index monitoring. Anesthesia & Analgesia, 90(5), 1114-1117.
 Brogi, E., Cyr, S., Kazan, R., Giunta, F., & Hemmerling, T. M. (2017). Clinical performance and safety of closed-loop systems: a systematic review and meta-analysis of randomized controlled trials. Anesthesia & Analgesia, 124(2), 446-455.
 Pasin, L., Nardelli, P., Pintaudi, M., Greco, M., Zambon, M., Cabrini, L., & Zangrillo, A. (2017). Closed-loop delivery systems versus manually controlled administration of total IV anesthesia: a meta-analysis of randomized clinical trials. Anesthesia & Analgesia, 124(2), 456-464.
 Puri, G. D., Mathew, P. J., Biswas, I., Dutta, A., Sood, J., Gombar, S., … & Arora, I. (2016). A multicenter evaluation of a closed-loop anesthesia delivery system: a randomized controlled trial. Anesthesia & Analgesia, 122(1), 106-114.
 Zaouter, C., Hemmerling, T. M., Lanchon, R., Valoti, E., Remy, A., Leuillet, S., & Ouattara, A. (2016). The feasibility of a completely automated total IV anesthesia drug delivery system for cardiac surgery. Anesthesia & Analgesia, 123(4), 885-893.
 Angie, D. (2018). 6 insights on how artificial intelligence could transform anesthesia. Becker’s ASC Review. Obtained from: https://www.beckersasc.com/anesthesia/6-insights-on-how-artificial-intelligence-could-transform-anesthesia.html