Doctor dentist working in office, using laptop, making notes on clipboard. Dentistry, medicine, hygiene, treatment, dental health care concept

AI-driven drug dosing algorithms

Doctor dentist working in office, using laptop, making notes on clipboard. Dentistry, medicine, hygiene, treatment, dental health care concept
iStock

Artificial intelligence (AI) is transforming healthcare practice by helping to enhance decision-making and personalize treatments. As general dentists increasingly manage anxious patients, pediatric populations and medically complex individuals requiring sedation, understanding the current capabilities of AI-based systems can enhance drug safety, monitoring accuracy and treatment planning. AI-driven dosing and decision-support tools are starting to show promising results in anesthesiology by predicting physiological instability, guiding anesthesia depth, and improving analgesic titration.1-4 Although still early, these advances have clear relevance for dental practices especially where sedation services are routinely provided for behaviourally challenging patients. This review summarizes current evidence and future implications on AI-assisted monitoring and dosing, AI-based behavioural support and decision-support tools highlighting the impact in dentistry.

AI systems in anesthesia and sedation

AI-driven algorithms work by training on large datasets to recognize patterns and make decisions without being explicitly programmed for every scenario. In anaesthesiology, this can help clinicians to predict drug effects and patient responses. AI-driven drug dosing refers to systems that assist or autonomously adjust anesthetic, analgesic or sedative drug delivery based on physiologic inputs or patient factors. They are generally categorized into the open-loop systems, which provide recommendations to the clinician without directly administering the medication, and the closed-loop system, which automatically titrates drug infusions to maintain a desired endpoint such as target depths of anesthesia or analgesia.3,4

The closed-loop systems use real-time inputs such as EEG indices, photoplethysmographic nociceptive signals or hemodynamic variability to adjust dosing more frequently than a clinician without real-time algorithmic feedback.3 Open-loop systems may predict the likelihood of postoperative pain, risk of hemodynamic instability or appropriate drug doses based on clinical patterns extracted through machine learning.5-7 Between both models, AI aims to reduce variability, anticipate instability and provide individualized care which is highly relevant to office-based sedation.

Related article: AI Doctor: The rise of artificial intelligence in healthcare

AI monitoring, dosing, and safety

Studies demonstrate that AI-enhanced monitoring may significantly improve the precision and timing of analgesic and anesthetic dosing. The Nociception Level (NOL) index, an AI-derived parameter combining photoplethysmography, skin conductance and heart rate, has shown promise in guiding opioid titration. In a systematic review, NOL-guided analgesia reduced postoperative pain scores by almost one-third compared to standard care without increasing opioid consumption.5 A randomized controlled trial by Fuica et al. demonstrated that NOL-guided fentanyl dosing during abdominal surgery produced lower postoperative pain scores with similar total opioid use which suggests improved timing rather than quantity of administration.6 Morisson et al. also showed that intraoperative nociceptive patterns predicted patients at risk of moderate to severe postoperative pain with good discriminative ability.7 These findings highlight that AI may soon allow dentists and anesthesiologists to anticipate pain trajectories and tailor perioperative dosing accordingly.

AI has also shown capabilities in physiologic monitoring and predictions. Tools like the Hypotension Prediction Index (HPI) can forecast hemodynamic instability before onset using arterial waveform features which allow pre-emptive vasopressor or fluid management.2 Although this requires arterial lines, it highlights a conceptual relevance for sedation where undetected hypotension and respiratory depression is dangerous. AI-augmented respiratory monitoring during moderate sedation has also shown improved sensitivity for early hypoventilation or airway obstruction which could be used for training or office-based IV and oral sedation.8

AI for behavioural support and communication

Beyond intraoperative dosing, AI systems can enhance perioperative behavioural management, particularly in children. Social robots, virtual-reality platforms and interactive AI-driven tools have been shown to reduce anxiety and improve cooperation in pediatric dental settings.9-11 These interventions reduce behavioural resistance and perceived pain which can lower the need for pharmacological sedation.9 In a randomized controlled trial, a humanoid robot improved cooperation during IV induction in children undergoing anesthesia.11 Another study demonstrated improved postoperative anxiety and mobilization with an interactive robot.10 Behavioural AI may represent a meaningful adjunct to pharmacologic dosing, particularly for dentists managing anxious or special needs children.

Large language models (LLMs), including ChatGPT, have also begun to influence perioperative planning and patient communication. However, the current evidence highlights important limitations. Abdel Malek et al. found that ChatGPT-4 generated coherent anesthetic management plans but occasionally made unsafe recommendations, underestimated risk and omitted necessary airway and postoperative considerations.1 Reader and Drum also reported that ChatGPT responses to questions about local anesthesia were generally accurate but sometimes lacked details and included incomplete information.12 A randomized controlled trial by Yahagi et al. demonstrated that a ChatGPT-based educational tool reduced pre-operative anxiety but contained inaccurate facts such as nonexistent drugs and suggestions not adhering to guidelines.13 These studies collectively highlight that LLMs may enhance communication but cannot be relied upon for autonomous dosing decisions without expert oversight.

Clinical relevance, limitations and challenges

For general dentists, AI-driven dosing and monitoring systems may provide practical advantages in several ways. AI-enhanced respiratory and nociceptive monitoring has the potential to function as real-time safety nets during nitrous oxide or moderate sedation.8 Predictive models may help identify pediatric or medically compromised patients who are at higher risk of postoperative pain, anxiety or behavioural issues that allow earlier referral.7,9,11 LLM-based tools may help reduce anxiety and improve understanding when used with clinician supervision.13

Despite optimistic progress, there remains significant concerns. AI-driven dosing algorithms are often built on small, single-centre datasets that limit their generalizability to diverse patient populations.14 Transparency also remains a major challenge as many algorithms are complex and difficult for clinicians to evaluate reliability or bias.3,4 LLMs may generate fabricated but clinically plausible statements which raise medicolegal concerns if used without verification. Integration into office workflows such as compatibility with monitors, infusion systems, electronic records remain a barrier as well.1,4,13 Ultimately, clinicians must bear full responsibility for dosing decisions and treatment planning regardless of AI assistance.

Future directions and conclusion

The next decade may see significant advances in sedation AI tools, including models trained on dental and ambulatory anesthesia datasets and low-cost monitors capable of feeding real-time physiologic data into dosing algorithms. Generative AI may also support the design and manufacturing of novel sedative agents with improved pharmacokinetics for short dental procedures.15 More studies are required alongside regulatory frameworks that define clinician responsibilities.

AI-driven dosing algorithms are part of a promising evolution in transforming healthcare practice. While current systems require careful oversight, these tools may eventually improve patient safety and optimize individualized care. 

Oral Health welcomes this original article.

References

  1. Abdel Malek M, van Velzen M, Dahan A, Martini C, Sitsen E, Sarton E, et al. Generation of preoperative anaesthetic plans by ChatGPT-4.0: a mixed-method study. Br J Anaesth. 2025 May 1;134(5):1333–40.
  2. Shaw KM, Safavi KC. The Role of Artificial Intelligence in Anesthesia Monitoring and Surveillance. Anesthesiol Clin. 2025 Sept 1;43(3):577–85.
  3. Hashimoto DA, Witkowski E, Gao L, Meireles O, Rosman G. Artificial Intelligence in Anesthesiology: Current Techniques, Clinical Applications, and Limitations. Anesthesiology. 2020 Feb;132(2):379–94.
  4. Ye J, Bronstein S. Artificial Intelligence in Anesthesiology Drug Administration. Anesthesiol Clin. 2025 Sept 1;43(3):431–51.
  5. Agrawal S, Sihman Bharattej Rupavath RVS, Prasad Jalaja P, Ushmani A, Mishra A, Bodapati NVSB. Artificial Intelligence (AI)-Driven Approaches to Manage Postoperative Pain, Anxiety, and Psychological Outcomes in Surgical Patients: A Systematic Review. Cureus [Internet]. 2025 May 16 [cited 2025 Dec 9]; Available from: https://www.cureus.com/articles/370296-artificial-intelligence-ai-driven-approaches-to-manage-postoperative-pain-anxiety-and-psychological-outcomes-in-surgical-patients-a-systematic-review
  6. Fuica R, Krochek C, Weissbrod R, Greenman D, Freundlich A, Gozal Y. Reduced postoperative pain in patients receiving nociception monitor guided analgesia during elective major abdominal surgery: a randomized, controlled trial. J Clin Monit Comput. 2023 Apr;37(2):481–91.
  7. Morisson L, Nadeau-Vallée M, Espitalier F, Laferrière-Langlois P, Idrissi M, Lahrichi N, et al. Prediction of acute postoperative pain based on intraoperative nociception level (NOL) index values: the impact of machine learning-based analysis. J Clin Monit Comput. 2023 Feb;37(1):337–44.
  8. Mincer MN, Mincer RC, Sung EC. Augmenting patient monitoring during intravenous moderate sedation with artificial intelligence: A pilot study. Spec Care Dentist. 2024 Nov;44(6):1788–92.
  9. Acharya S. Role of artificial intelligence in behavior management of pediatric dental patients—a mini review. J Clin Pediatr Dent. 2024;48(3):24.
  10. Topçu SY, Semerci R, Kostak MA, Güray Ö, Sert S, Yavuz G. The effect of an interactive robot on children’s post-operative anxiety, mobilization, and parents’ satisfaction; randomized controlled study. J Pediatr Nurs. 2023 Jan;68:e50–7.
  11. Lee-Krueger RCW, Pearson JR, Spencer A, Noel M, Bell-Graham L, Beran TN. Children’s Pain During IV Induction: A Randomized-Controlled Trial With the MEDi® Robot. J Pediatr Psychol. 2021 Aug 19;46(8):991–1000.
  12. Reader A, Drum M. A Review of ChatGPT as a Reliable Source of Scientific Information Regarding Endodontic Local Anesthesia. J Endod. 2025 May;51(5):571–6.
  13. Yahagi M, Hiruta R, Miyauchi C, Tanaka S, Taguchi A, Yaguchi Y. Comparison of Conventional Anesthesia Nurse Education and an Artificial Intelligence Chatbot (ChatGPT) Intervention on Preoperative Anxiety: A Randomized Controlled Trial. J Perianesth Nurs. 2024 Oct;39(5):767–71.
  14. Serna-Muñoz C, Ortiz-Ruiz AJ, Gómez-Ríos I, Saura-López V, Pérez-Silva A. Application of machine learning for data analysis in paediatric dentistry: a systematic review. Eur J Paediatr Dent. 2025;(Early Access):1.
  15. Julius A, Bowness JS. Generative AI models: the next anaesthetic agent? Br J Anaesth. 2025 July;135(1):21–5.

About the author

Dr. Jason Lee is a PGY1 dental anaesthesia resident at the University of Toronto. He earned his Bachelor of Science and Doctor of Dental Medicine degrees from the University of British Columbia in Vancouver. Prior to entering residency, Dr. Lee practiced general dentistry for three years in rural Alberta. His academic interests include pain management and the delivery of care to diverse patient populations..