Abstract
OBJECTIVE: Based on deep learning, the characteristics of food impaction with tight proximal contacts were studied to guide the subsequent clinical treatment of occlusal adjustment. At the same time, digital model building, software measurement, and statistical correlation analysis were used to explore the cause of tooth impaction and to provide evidence for clinical treatment. METHODS: Volunteers with (n = 250) and without (n = 250) tooth impaction were recruited, respectively, to conduct a questionnaire survey. Meanwhile, models were made and perfused by skilled clinical physicians for these patients, and characteristics such as adjacent line length, adjacent surface area, tongue abduction gap angle, buccal abduction gap angle, and occlusal abduction gap angle were measured. A normality test, differential analysis, correlation analysis of pathological characteristics of the impaction group, principal component analysis (PCA), and binary logistic regression analysis were performed. RESULTS: The adjacent line length, adjacent surface area, tongue abduction gap angle, buccal abduction gap angle, and occlusal abduction gap angle all met normal distribution. There were statistically significant differences in adjacent line length (p < 0.001), adjacent surface area (p < 0.001), and occlusal abduction gap angle (p < 0.001) between the two groups. After dimensionality reduction by PCA on characteristics, adjacent line length, adjacent surface area, buccal abduction gap angle, and occlusal abduction gap angle had a strong correlation with the principal components. Binary logistic regression analysis showed that adjacent line length and adjacent surface area had positive effects on impaction. The buccal abduction gap angle and occlusal abduction gap angle had a significant negative influence on impaction. CONCLUSION: Adjacent line length, adjacent surface area, buccal abduction gap angle, and occlusal abduction gap angle are independent factors influencing food impaction.