The Impact of Artificial Intelligence on the Accuracy of Interproximal Reduction Simulation in Clear Aligner Therapy: A Scoping Review
This scoping review synthesizes evidence from 19 studies (2022–2026) to demonstrate that AI measurably influences interproximal reduction simulation accuracy across clear aligner therapy workflows, while highlighting the critical need for dedicated prospective studies comparing AI-simulated versus conventional IPR outcomes.
Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer
Imagine your mouth as a bustling city of teeth, where every tooth is a building standing shoulder-to-shoulder. Sometimes, to fix a crooked street or make room for a new park, the city planners need to shave off a tiny bit of the side of a building. This is called Interproximal Reduction (IPR). It's a delicate job: shave too little, and the buildings won't fit; shave too much, and you damage the structure, causing sensitivity or even collapse. In the past, a human architect (an orthodontist) would measure and plan this shaving by hand. But now, we have a new kind of architect: Artificial Intelligence (AI). Think of AI as a super-fast, super-smart robot that can look at a 3D map of your teeth and instantly tell you exactly where to shave.
The big question everyone is asking is: Is this robot architect actually better than the human one? Does it make fewer mistakes, or does it get overconfident and shave off too much? This is the story of a new scientific investigation that tried to find out if these digital robots are ready to take over the job of planning tooth shaving, or if they still need a human to double-check their math.
The Robot Architect's Report Card
A team of researchers decided to act like detectives, hunting down every study they could find between 2022 and 2026 that tested how well AI handles this "tooth shaving" planning. They didn't just look at one robot; they looked at 19 different studies involving various AI tools, software, and digital models. Their goal was to map out the landscape: Where does the AI shine, and where does it stumble?
Here is what they found, broken down into the four main ways AI tries to help.
1. The "Measuring Tape" Glitch
Before a robot can tell you how much to shave, it first has to measure the tooth. The researchers found that the AI's measuring tape isn't always perfect. In one study, an AI system called "DentOne" measured the back teeth as being wider than they actually were. It was off by about 0.84 to 0.87 mm. Compare that to a popular human-guided system called ClinCheck, which was only off by 0.47 to 0.56 mm.
Think of it like this: If you ask a robot to measure a cookie, and it thinks the cookie is slightly bigger than it really is, it might tell you to cut off a huge chunk to make it fit in a jar. In reality, you'd only need to trim a tiny crumb. The paper suggests that because the AI sometimes overestimates the width of teeth, it might accidentally prescribe too much shaving, potentially removing more enamel than necessary.
2. The "Virtual Setup" Shuffle
Once the AI measures the teeth, it builds a virtual model of how they should move. The researchers found that different AI programs don't always agree with each other. In one experiment, when the exact same instructions were given to four different software platforms, they all created different virtual setups. The differences were most obvious in cases of severe crowding, where the teeth are packed tight.
It's like giving four different chefs the exact same recipe for a cake; they might all use slightly different techniques for mixing or baking, resulting in cakes that look and taste a bit different. The study showed that the "linear errors" (how far the tooth moves) in these automated setups could range from 0.39 mm to 1.40 mm, and the angles could be off by 3.25 to 7.80 degrees. While some of this is acceptable, it means the robot isn't always consistent, and sometimes a human needs to step in to fix the plan.
3. The Crystal Ball (Predicting the Future)
AI is also being used as a crystal ball to predict what might go wrong during treatment, like if a patient will need a "refinement" (a second round of aligners) or if a gap might open up between the gums and teeth. The researchers found that AI is actually quite good at spotting these risks.
One study used a machine learning model to predict "open gingival embrasures" (those ugly gaps between teeth and gums). The AI model was correct 88% of the time (an AUC of 0.880), which was better than a model that only looked at human clinical notes. Another study found that AI could predict refinement risks with 93.94% accuracy. This suggests that while the AI might be shaky at planning the shaving, it's getting really good at predicting the consequences of that shaving.
4. The Magic Mirror Effect
Finally, the researchers looked at how AI visuals change what the human doctor decides to do. They found that when doctors could see a 3D AI-generated picture of the tooth roots and the bone underneath (instead of just the crown of the tooth), they changed their minds about the treatment plan 33% to 43% of the time.
Imagine looking at a house through a window versus looking at a blueprint that shows the hidden pipes and foundation. The blueprint makes you realize, "Oh, I can't knock down that wall!" Similarly, seeing the AI's 3D bone view made doctors more confident and willing to change their plans. However, the paper notes a catch: just because the doctor changed their mind based on the AI picture doesn't mean the AI picture was 100% accurate. No study actually checked if the AI's picture of the bone matched reality.
The Bottom Line: Trust, but Verify
So, is the AI robot architect ready to take over the world? The answer from this review is a cautious "not quite yet."
The paper concludes that AI definitely affects the simulation of tooth shaving at every step. It can measure, plan, and predict, but it carries measurable errors. Specifically, it tends to overestimate tooth width, which could lead to removing too much enamel. It also produces different results depending on which software you use.
Most importantly, the researchers point out that no study has directly compared AI-generated plans against human-generated plans to see which one actually works better in real life. We don't have a "gold standard" test yet. Because of this, the paper strongly suggests that dentists should not blindly trust the AI's prescription. They need to double-check the measurements, especially for back teeth, and treat the AI's 3D visuals as a helpful guide rather than a final truth.
In short, the AI is a powerful new tool that can speed things up and spot risks, but it's not a replacement for the human expert. For now, the best plan is to let the robot do the heavy lifting, but keep a human hand on the steering wheel to make sure the robot doesn't shave off too much of the cookie.
Drowning in papers in your field?
Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.