Simulation and Artificial Intelligence in Rhinoplasty: A Systematic Review

Overview

Rhinoplasty is challenging because of the complexity of nasal structure and the effect the operation has on both appearance and breathing. The authors note artificial intelligence and simulation systems are increasingly used in plastic surgery, and this review explores their potential uses in rhinoplasty. Five databases were searched and the review followed the PRISMA reporting standard.

Sections of note

• Systematic review across PubMed, CINAHL, EMBASE, Scopus and Web of Science; the journal assigns level of evidence III.
• Simulation models were described that predict a nasal shape aesthetically matching the patient's face.
• Other models indicate implant size in augmentation rhinoplasty.
• Others construct three dimensional facial images from two dimensional photographs.
• Machine learning was used to learn individual surgeons' rhinoplasty styles and simulate their outcomes.
• Deep learning was used to predict rhinoplasty status and to analyse factors associated with increased facial attractiveness after surgery.
• A deep learning model predicted patients' apparent age before and after rhinoplasty, which the authors state showed the procedure made patients look younger.
• The authors conclude these systems can assist in planning, decisions during surgery and evaluation afterward.

What it means for a patient

• Most of these tools support planning and simulation rather than performing any part of the surgery.
• A system trained on a particular surgeon's past cases reproduces that surgeon's style, not an objective ideal.
• The conclusion that rhinoplasty made patients look younger comes from a model estimating apparent age from photographs, not from any clinical measure.
• The review describes what has been built; it does not test whether any of it improves outcomes.

Why this paper matters

Simulation is already routine in rhinoplasty consultation and automation of it is advancing quickly. This review maps what has been attempted across the field. None of the described systems has been shown to change surgical results or patient satisfaction.

Terms

• Simulation model: software producing a preview of an intended surgical result.
• Machine learning: computer methods that find patterns in data to make predictions.
• Deep learning: a form of machine learning using layered neural networks.
• PRISMA: a standard method for reporting systematic reviews.
• Three dimensional reconstruction: building a three dimensional model from flat images.