AI Revolutionizing Anesthesiology: From Machine Learning to AnestCopilot

Key takeaways#

Core concepts

• Modern AI uses machine learning to learn patterns from clinical data and support prediction, classification, or recommendation.
• Generative AI produces new content (text, image, audio, or multimodal) from prompts and/or clinical data.³
• LLMs operate on natural language and can summarize, explain, structure, and retrieve knowledge from large volumes of text.

Impact and limitations in healthcare

• Concrete and regulated applications exist; the FDA publishes annual reports on approvals and authorizations.⁴
• For generative AI, there is ongoing discussion about efficiency alongside risks: incorrect content, biases, and the need for rigorous governance and validation.²

Why this matters in anesthesiology

• The specialty requires real-time decision-making in high-risk settings, integrating pharmacology, physiology, mechanical ventilation, monitoring, and surgical context.¹⁻²
• AnestCopilot is proposed as a "copilot" to organize knowledge, check risks, and reduce cognitive load, aligned with quality and clinical safety criteria in AI governance.²

Introduction#

Artificial intelligence (AI) has evolved from a niche computing topic to a cross-cutting accelerator of medicine.

A key aspect of this transformation is that modern AI doesn't rely solely on "programmed rules"; it primarily depends on machine learning—building statistical models that "learn" patterns from data (e.g., vital signs, laboratory tests, images, clinical text) to predict, classify, or recommend based on prior examples.

What is generative AI and why has it become central

In recent years, a subset of ML has taken center stage: generative AI.

Unlike classical predictive models (which answer "what is the probability of X?"), generative models are designed to produce new content—text, image, audio, or multimodal combinations—from a prompt and/or clinical data.³

The World Health Organization describes this group as large multimodal models (LMMs) capable of receiving different types of input and generating diverse outputs, with expected broad applications in healthcare, research, and drug development.

In practice, this family includes large language models (LLMs), which operate on natural language and can summarize, explain, structure, and retrieve knowledge from large volumes of clinical and scientific text.

AI in healthcare: concrete applications, regulation, and risks

In healthcare, these technologies already appear in concrete and regulated applications.

In the "predictive/analytical AI" axis, regulatory bodies such as the FDA publish information and annual reports on approvals and authorizations of health products.⁴

In the "generative AI" axis, the initial impact has been marked in clinical documentation and information synthesis, with growing discussions in the literature about potential efficiency gains (reducing administrative and cognitive burden), while also highlighting typical risks of generative models, such as production of incorrect content, biases, and the need for rigorous governance and validation.²

Why this is particularly relevant for anesthesiology

This background is particularly relevant for anesthesiology: a specialty that requires real-time decision-making in high-risk environments, where rapid integration of pharmacology, physiology, mechanical ventilation, monitoring, and surgical context is fundamental.¹⁻²

The proposal for tools like AnestCopilot arises precisely from this intersection between a high-pressure decision environment and AI capable of organizing knowledge, checking risks, and reducing cognitive load.

This type of implementation is described as aligned with quality, transparency, and clinical safety criteria, as well as best practices and recommended governance for AI in healthcare.²

AnestCopilot: AI as a copilot in anesthetic practice#

AnestCopilot is a specialized clinical intelligence platform for anesthesiologists.

Each tool functions as an AI copilot, combining advanced algorithms with a knowledge base curated by human specialists. In essence, it offers rapid answers (in seconds) to clinical questions in the operating room, always with evidence-based content.

The platform brings together a set of 11 integrated tools covering everything from preoperative preparation to intraoperative emergencies, eliminating blind spots and providing safety when every second counts.

AI Copilot: Assistant and Diagnostic modes#

The AnestCopilot intelligent copilot is a virtual assistant specialized in anesthesia, capable of understanding clinical questions in natural language and providing contextualized answers.

It operates in two modes:

• Assistant Mode: for general consultations about approaches in clinical scenarios.
• Diagnostic Mode: the AI focuses on diagnostic analysis from clinical cases or presented symptoms.

Users can even dictate their question or case description verbally, as the platform incorporates voice transcription to streamline interaction in the operating room environment.

If the answer isn't in the content curated by anesthesiologists, it automatically performs a search in the medical literature, bringing a synthesis of available evidence.

In other words, the anesthesiologist has access to a "virtual consultant" that elaborates an evidence-based answer in seconds, saving precious minutes that would otherwise be spent leafing through books or articles during surgery.

Preoperative medication management tool#

One of the major challenges in preoperative preparation is deciding which patient medications to maintain, adjust, or discontinue before surgery.

With the proliferation of new therapeutic classes—immunobiologics, oral anticoagulants, innovative antidiabetic agents, among others—it has become humanly difficult to memorize the recommended approach for each drug.

For example, the pace of new medication launches is accelerated: in 2021 alone, the FDA approved 50 innovative drugs (new molecular entities), more than double the annual average from a decade ago.⁷

AnestCopilot's Medication Discontinuation tool solves this problem, offering clear and up-to-date guidance on preoperative management of over 5,000 drugs and supplements.

Simply search for the medication name to get an objective recommendation in seconds.

Recommendations are evidence-based and follow best practices (including society guidelines), being constantly updated as new data emerges.

This means that even for that unfamiliar medication, the anesthesiologist can rely on AnestCopilot to guide their decision.

Drug interactions tool#

Another field where AI brings significant impact is in checking drug interactions.

During anesthesia, a myriad of drugs are administered (intravenous and inhalational anesthetics, opioids, neuromuscular blocking agents, prophylactic antibiotics, vasopressors, etc.), in addition to the chronic medications the patient was already taking.

Each combination carries the potential for interaction—pharmacodynamic or pharmacokinetic—that can alter the expected effect or cause adverse reactions.

In fact, the risk of drug interactions is probably higher in anesthesia than in other areas of medicine, given the number and variety of agents used in a short time interval.⁶

Many interactions are beneficial and deliberate (for example, using synergistic drugs to reduce individual doses), but others can be dangerous.

Manually identifying all potential interactions from the patient's medication list and the anesthesiologist's drugs is a complex task.

AnestCopilot's Drug Interactions tool functions as an anesthesia-specialized checker, analyzing the entire planned drug combination in seconds and flagging important risks in real time.⁶

Arterial blood gas analysis (GasoAI)#

Arterial blood gas (ABG) analysis is a critical examination for rapid assessment of the patient's respiratory and metabolic status during surgery and in the ICU.

However, manually interpreting the test takes time and attention, especially if it's necessary to calculate anion gap, delta ratio, estimate compensations, or identify mixed disorders.

In emergency situations or patient instability, every minute counts, and waiting 3 to 5 minutes for a physician to thoroughly analyze the ABG can delay interventions.

With this in mind, AnestCopilot incorporated GasoAI, an AI tool that interprets complete arterial blood gas in just 3 seconds, instead of 3 minutes.

The operation is simple: the anesthesiologist can photograph the printed ABG result or enter the values, and GasoAI immediately processes the data.

In its response, the system provides a comprehensive clinical interpretation, already indicating which acid-base disorders are present and presenting diagnostic suggestions compatible with that pattern.

This automated and instant analysis allows confident action even under pressure, guiding ventilatory adjustments or metabolic interventions as needed according to the patient's condition.

Mechanical power ventilatory analysis#

In the mechanical ventilation domain, AI also makes a notable contribution through the Mechanical Power analysis tool.

Mechanical power is a concept that integrates various ventilatory variables—tidal volume, driving pressure, PEEP, and respiratory rate—to express the total energy transmitted to the lungs per minute during ventilation.⁵

Studies in intensive care have shown that elevated mechanical power is associated with a higher risk of ventilator-induced lung injury (VILI) and worse outcomes.⁵

For example, even patients ventilated with "low" tidal volume or controlled driving pressure showed increased mortality when accumulated mechanical power was high.

This suggests that mechanical power functions as a unifying marker of mechanical stress imposed on the lungs and deserves careful monitoring.⁵

In anesthesiology, similarly, elevated mechanical power (above 12.9 joules/min) has also been associated with increased postoperative respiratory complications.⁸

However, manually calculating mechanical power at the bedside is not trivial and involves applying formulas unfamiliar to anesthesiologists.

AnestCopilot radically simplifies this process: with the Mechanical Power tool, the anesthesiologist can simply take a photo of the ventilator screen, and the AI extracts the displayed ventilatory parameters (such as volumes, pressures, and frequency) to automatically calculate mechanical power.

In seconds, the tool reports the calculated value (in Joules/min) and correlates it with the risk of lung injury based on published threshold values.

Quick overview of AnestCopilot tools#

Conclusion#

The incorporation of artificial intelligence in anesthesiology, exemplified by AnestCopilot and its tools, represents a paradigm shift in how anesthesiologists access information and make decisions.¹⁻²

In a scenario where time is critical and the volume of knowledge is immense, having a digital "copilot" capable of providing rapid, accurate, and evidence-based answers becomes a valuable differentiator.

Tools like AnestCopilot don't replace human clinical judgment but enhance it: they reduce cognitive overload, increase patient safety, and allow the anesthesiologist to focus on what matters most—direct patient care—with the peace of mind of having reliable support for other intellectual tasks.

The AI revolution in science and medicine is already underway,³ and in anesthesiology, it translates to more informed decisions, safer practices, and continuous learning.¹⁻²

References#

1. Minehart RD et al. Artificial Intelligence Supporting Anesthesiology Clinical Decision-Making. Anesth Analg. 2025;141(3):536-539.
2. Hashimoto DA et al. Artificial intelligence in anesthesiology: current techniques, clinical applications, and limitations. Anesthesiology. 2020;132(2):379-394.
3. Karpatne A et al. AI-enabled scientific revolution in the age of generative AI. NPJ Artificial Intelligence. 2025;1:18
4. U.S. Food and Drug Administration. Artificial Intelligence-Enabled Medical Devices. FDA. https://www.fda.gov/medical-devices/software-medical-device-samd/artificial-intelligence-enabled-medical-devices
5. Gattinoni L, Tonetti T, Cressoni M, et al. Ventilator-related causes of lung injury: the mechanical power. Intensive Care Med. 2016;42:1567-1575.
6. Silva A et al. New Perspective for Drug-Drug Interaction in Perioperative Period. J Clin Med. 2023;12(14):4810.
7. Mullard A. 2021 FDA approvals. Nat Rev Drug Discov. 2022;21(2):83-88.
8. El-Khatib et al. Intraoperative mechanical power and postoperative pulmonary complications in low-risk surgical patients: a prospective observational cohort study. BMC Anesthesiology (2024) 24:82

Related reading#

• Etomidate vs Ketamine for RSI in Critical Patients
• HES 6% in Surgical Fluid Therapy: Does PHOENICS Rewrite the Story?
• Perioperative Management of GLP-1 and GIP Receptor Agonists