Aim To investigate the feasibility of predicting dental implant loss risk with deep learning (DL) based on preoperative cone‐beam computed tomography. Materials and Methods Six hundred and three patients who underwent implant surgery (279 high‐risk patients who did and 324 low‐risk patients who did not experience implant loss within 5 years) between January 2012 and January 2020 were enrolled. Three models, a logistic regression clinical model (CM) based on clinical features, a DL model based on radiography features, and an integrated model (IM) developed by combining CM with DL, were developed to predict the 5‐year implant loss risk. The area under the receiver operating characteristic curve (AUC) was used to evaluate the model performance. Time to implant loss was considered for both groups, and Kaplan–Meier curves were created and compared by the log‐rank test. Results The IM exhibited the best performance in predicting implant loss risk (AUC = 0.90, 95% confidence interval [CI] 0.84–0.95), followed by the DL model (AUC = 0.87, 95% CI 0.80–0.92) and the CM (AUC = 0.72, 95% CI 0.63–0.79). Conclusions Our study offers preliminary evidence that both the DL model and the IM performed well in predicting implant fate within 5 years and thus may greatly facilitate implant practitioners in assessing preoperative risks.

No abstract

No abstract

We investigate the behaviour of tilting sheaves under pushforward by a finite Galois morphism. We determine conditions under which such a pushforward of a tilting sheaf is a tilting sheaf. We then produce some examples of Severi-Brauer flag varieties and arithmetic toric varieties in which our method produces a tilting sheaf, adding to the list of positive results in the literature. We also produce some counterexamples to show that such a pushfoward need not be a tilting sheaf. Tilting sheaves and base change2.1. Generation in derived categories. Let D be a triangulated category and S a set of objects in D. We denote by < S > the smallest full triangulated category containing all the objects in S. We denote by < S > κ the smallest thick triangulated containing all the objects in S. Note that thick subcategories are assumed to be full.An object C of D is said to be compact if Hom(C, −) commutes with direct sums. We denote by D c the full subcategory of compact objects.Given a set S of objects of D we define S ⊥ to be the full subcategory of D consisting of objects A with Hom D (E[i], A) = 0 for all E ∈ S and i ∈ Z. We say that S right spans D if S ⊥ = {0}.If D c right spans D we say that D is compactly generated. Theorem 2.1. (Ravenel and Neeman) Let D be a compactly generated triangulated category. Then a set of compact objects S right spans D if and only if < S > κ = D c . Proof. See [BV, Theorem 2.1.2].Let Y be a scheme. We denote the unbounded derived category of quasi-coherent sheaves on Y by D(Qcoh(Y )) and the bounded derived category of coherent sheaves by D b (Y ).Proposition 2.2. Let Y be a quasi-compact, separated scheme. Then D(Qcoh(Y )) is compactly generated.