Category: Statistics

J Prosthet Dent. 2026 Aug 15:S0022-3913(26)00516-0. doi: 10.1016/j.prosdent.2026.07.012. Online ahead of print.

ABSTRACT

STATEMENT OF PROBLEM: The design outcomes of dental crowns generated by artificial intelligence (AI)-based computer-aided design (CAD) systems have not been sufficiently evaluated in comparison with those produced by conventional CAD software programs.

PURPOSE: The purpose of this study was to compare the precision, morphology, and structural stability of dental crowns designed using AI-based software programs and a conventional CAD software program.

MATERIAL AND METHODS: A digital cast of the maxillary first molar was created from a typodont. Crowns were designed with 3 CAD software programs: a conventional software program (3Shape Dental Designer; 3Shape A/S) (CC group) and 2 AI-based software programs (3Shape Automate; 3Shape A/S) (AA group) and (Dentbird; Imagoworks Inc) (AD group) using the same prepared tooth standard tessellation language (STL) file. Sample size had been determined using an a priori power analysis with α=.05, 1-β=.80, and 10 specimens per group were fabricated using a 5-axis milling machine (BX5 PLUS; Dental Plus). The designs were evaluated for precision using root mean square (RMS) deviation, and the cusp angle of the functional cusp was analyzed for morphological comparison. Finite element analysis (FEA) was performed to compare the stress distribution among the groups. Statistical significance was assessed using the Kruskal-Wallis test and 1-way analysis of variance (ANOVA) (α=.05).

RESULTS: The CC group showed the highest precision, presenting the lowest RMS deviation (13.9 ±3.8 µm), followed by the AD and AA groups, with significant intergroup differences (P<.001). For morphological evaluation, the AA group exhibited the lowest functional cusp angle (64.4 ±3.5 degrees), which was significantly smaller than those of the CC and AD groups (P<.001). FEA demonstrated that all crowns experienced maximum stress under 0-degree loading and minimum stress under 90-degree loading conditions. The AA group showed the highest von Mises stress (839.8 MPa), whereas the AD group exhibited the smallest displacement and the most homogeneous stress distribution.

CONCLUSIONS: Conventional CAD exhibited higher precision than AI-based crown design systems, while AI-based designs showed altered cusp angle and occlusal morphology. FEA demonstrated design-dependent differences in stress distribution and displacement.

PMID:42601314 | DOI:10.1016/j.prosdent.2026.07.012

Oral Surg Oral Med Oral Pathol Oral Radiol. 2026 Jun 25:S2212-4403(26)00256-7. doi: 10.1016/j.oooo.2026.06.009. Online ahead of print.

ABSTRACT

BACKGROUND: There is increasing interest in MRI for diagnostic purposes in dentistry, resulting in development of a novel dental-dedicated MRI (ddMRI) system. This study compared 2D and 3D acquisitions of the mandibular third molar (MTM) region using ddMRI.

METHODS: Twenty MTMs of 20 patients were randomly selected from participants in clinical trials scanned using ddMRI (MAGNETOM Free. Max, 0.55 T, Siemens Healthineers AG, Forchheim, Germany). The 2D and 3D acquisitions were obtained for each patient. Three observers evaluated both 2D and 3D image datasets of each patient for overall quality and visibility of relevant anatomic structures. Results focused on descriptive statistics, inter- and intraobserver reproducibility and inferential statistics for comparison of 2D and 3D images within each observer.

RESULTS: Most anatomic structures were recorded as visible in every dataset for both 2D and 3D acquisitions. The 2D images showed superior quality and lingual nerve visibility across all observers compared to 3D volumes. Inter- and intraobserver reproducibility ranged vastly, from -0.167 to 1 and -0.667 to 1.

CONCLUSIONS: Most anatomic features in the MTM region are adequately depicted with both 2D and 3D acquisitions. The 2D acquisition quality was sufficient for diagnostic purposes of the MTM region, while diagnostic capability of the 3D volumes remains uncertain.

PMID:42601287 | DOI:10.1016/j.oooo.2026.06.009

Int J Oral Maxillofac Surg. 2026 Aug 14:S0901-5027(26)00338-3. doi: 10.1016/j.ijom.2026.07.017. Online ahead of print.

ABSTRACT

Surgical navigation can improve orbital implant positioning, yet current commercial systems are limited by the feedback interpretation and its timing after implant positioning. This study evaluated how different visualization methods impact the efficiency and interpretability of real-time feedback during orbital reconstruction. A standardized in-vitro setup was used to simulate orbital reconstructions by six surgeons of varying experience levels. Each surgeon performed implant positioning tasks using six different visualization modes, starting from three predefined misaligned positions. Interpretation accuracy and time to correct implant positioning were recorded. Visualization methods varied in the presence of computed tomography (CT) images, visualization of three-dimensional (3D) models, and use of implant-to-target distance visualization (distance mapping). Across a total of 108 reconstructions, no statistically significant differences in positioning time or interpretation accuracy were found between visualization methods. However, surgeon feedback revealed strong preferences: anatomical CT data and additional coronal view at the ledge improved spatial understanding, while views relying solely on 3D models or instrument orientation were seen as less intuitive. Notably, surgeons reported a consistent two-step strategy: initially 3D views were used for initial spatial alignment, followed by two-dimensional slices for precise adjustments. These findings suggest that while performance metrics may not differ across visualization modes, surgeon usability and interpretability are critical for clinical implementation. Future development of a clinically viable navigation workflow should include real-time feedback with both anatomical context and tailored visual support to promote more intuitive navigation in orbital surgery.

PMID:42601279 | DOI:10.1016/j.ijom.2026.07.017

Dig Liver Dis. 2026 Aug 14:S1590-8658(26)00881-9. doi: 10.1016/j.dld.2026.07.084. Online ahead of print.

ABSTRACT

BACKGROUND: Gastrointestinal stromal tumors (GISTs) are tumors with malignant potential. This research aims to develop an artificial intelligence (AI)-based system for analyzing endoscopic ultrasonography (EUS) visuals and generating risk scores.

AIMS: This system is designed to better predict GIST risk levels by identifying risk factors.

METHODS: An internal dataset comprising 504 EUS images from 226 patients with pathologically confirmed GISTs was collected from Yuzhong Hospital and Jiangnan Hospital of the Second Affiliated Hospital of Chongqing Medical University, Sichuan Provincial People’s Hospital between 2018 and 2024. Multiple AI-based machine learning models were developed and tested. Six machine learning models were compared, and the optimal diagnostic model was selected based on statistical results.

RESULTS: For the best-performing model, XGBoost, the internal test set results were as follows: overall accuracy 83.17%, sensitivity 75.68%, specificity 93.72%, positive predictive value (PPV) 73.21%, negative predictive value (NPV) 93.52%, F1 score 0.74, and area under the curve (AUC) 0.97. The external validation set results were: overall accuracy 80.68%, sensitivity 71.53%, specificity 92.72%, PPV 70.44%, NPV 92.78%, F1 score 0.70, and AUC 0.94.

CONCLUSIONS: This model utilizes machine learning algorithms on EUS images to accurately predict GIST risk stratification.

PMID:42601260 | DOI:10.1016/j.dld.2026.07.084

Sci Total Environ. 2026 Aug 14:182202. doi: 10.1016/j.scitotenv.2026.182202. Online ahead of print.

NO ABSTRACT

PMID:42601255 | DOI:10.1016/j.scitotenv.2026.182202

J Hazard Mater. 2026 Aug 14:143292. doi: 10.1016/j.jhazmat.2026.143292. Online ahead of print.

NO ABSTRACT

PMID:42601240 | DOI:10.1016/j.jhazmat.2026.143292

Arch Dis Child. 2026 Aug 14:archdischild-2026-330279. doi: 10.1136/archdischild-2026-330279. Online ahead of print.

ABSTRACT

Directed acyclic graphs (DAGs) are increasingly recommended or required by journals for observational studies making causal claims yet few paediatric observational studies present DAGs to support their analytical approach. This practical guide addresses the implementation gap between methodological guidance and research practice by clarifying when DAGs are needed (causal questions only, not descriptive or predictive research), demonstrating how to construct them for paediatric studies, and explaining how to use them to identify appropriate statistical adjustment strategies. We illustrate common pitfalls, including adjusting for colliders and mediators, address practical challenges such as temporal ordering ambiguity, and provide a worked example from recent paediatric literature. Proper use of DAGs makes causal assumptions explicit, prevents common analytical errors and strengthens causal inference in observational paediatric research.

PMID:42601214 | DOI:10.1136/archdischild-2026-330279

Br J Ophthalmol. 2026 Aug 14:bjo-2026-329939. doi: 10.1136/bjo-2026-329939. Online ahead of print.

ABSTRACT

BACKGROUND/AIMS: To investigate glaucoma prevalence and diagnostic discrepancies-undiagnosed glaucoma, missed diagnostic opportunities and false-positive diagnoses-in South Korea and to identify factors contributing to these errors.

METHODS: This nationwide cross-sectional study used data from the Korea National Health and Nutrition Examination Survey 2019-2021. Participants aged ≥20 years who had gradable ophthalmic examinations were included. Glaucoma was adjudicated by specialists using International Society of Geographical and Epidemiological Ophthalmology criteria. Survey-weighted logistic regression analyses were performed to identify factors associated with undiagnosed glaucoma and missed diagnostic opportunities.

RESULTS: The final analytical cohort comprised 18 710 participants. The weighted prevalence of glaucoma was 3.9% (95% CI 3.6% to 4.2%). The awareness rate was only 17.1%, leaving 82.9% (95% CI 80.4% to 85.4%) of cases undiagnosed. Notably, 44.7% (95% CI 41.0% to 48.3%) of undiagnosed cases represented a missed diagnostic opportunity. Multivariable analysis identified high myopia as the strongest predictor of a missed diagnostic opportunity (OR 23.53, 95% CI 5.65 to 98.09, p<0.001), alongside younger age (OR 0.98, p=0.039), rural residence (OR 1.97, p=0.047) and thicker retinal nerve fibre layer (OR 1.09, p<0.001). The lowest household income quartile was associated with undiagnosed glaucoma (OR 1.87, p=0.019).

CONCLUSION: Diagnostic discrepancies present a dual burden: profound underdiagnosis coexists with frequent false-positive diagnosis. High myopia, milder disease phenotype and rural residence are central drivers of missed diagnostic opportunities. Reducing discrepancies will therefore require efforts at both the clinical-professional and health system levels.

PMID:42601192 | DOI:10.1136/bjo-2026-329939

Stat Med. 2026 Aug;45(18-19):e70700. doi: 10.1002/sim.70700.

ABSTRACT

The FDA initiated Project Optimus and issued guidance for dose optimization, recommending randomized parallel dose-response cohorts to generate additional data at promising dose levels and implying that different dosages may be needed for different indications. In addition to dose optimization, with recent advancements in precision medicine and cancer biology, the development of cancer treatments has shifted toward the search for agents targeted to specific molecular profiles that may appear in more than one type of cancer. The basket trials are clinical trial designs that enable the simultaneous assessment of a new treatment in multiple indications. Concerning the FDA’s dose optimization perspective and the recent trend of basket trials in early-phase clinical trials, this paper proposes a dose-ranging basket trial design based on a Bayesian model-averaging approach considering efficacy and toxicity outcomes, where indications and dose levels define baskets. A key benefit of the proposed approach is that it explicitly accounts for the possible heterogeneity of response rates among baskets. Our simulation study shows that the proposed approach outperforms other methods, offering higher statistical power, better control of Type I error rates, precise optimal dose selection, and sample size savings in various scenarios with heterogeneous treatment effects between baskets.

PMID:42601178 | DOI:10.1002/sim.70700

BMJ Open Gastroenterol. 2026 Aug 14;13(1):e002278. doi: 10.1136/bmjgast-2026-002278.

ABSTRACT

OBJECTIVE: Patients with inflammatory bowel disease (IBD) have an increased risk of cardiovascular and cerebrovascular events, but the underlying mechanisms remain unclear. We aimed to assess subclinical carotid atherosclerosis in IBD in a large population-based cohort.

METHODS: We performed a cross-sectional analysis of the Paracelsus 10 000 cohort in Salzburg, Austria. Adults aged 40-77 years underwent standardised clinical evaluation, laboratory testing and bilateral carotid ultrasonography. The primary endpoint was carotid plaque burden; secondary endpoints included plaque presence, intima-media thickness, carotid stenosis, coronary artery calcium and polygenic risk scores. Multivariable ordered logistic regression models were adjusted for age, sex and comprehensive cardiovascular risk measures including Systematic Coronary Risk Evaluation 2 and Life’s Essential 8.

RESULTS: Among 9723 participants, 70 had IBD and 9653 served as controls. Patients with IBD had higher levels of high-sensitivity C reactive protein (median 0.15 vs 0.12 mg/dL; p=0.039) and lower ferritin levels (86 vs 117 ng/mL; p=0.018). Carotid plaque prevalence (40% vs 38%; p=0.79) and plaque burden distribution (p=0.82) were similar between groups. Intima-media thickness, carotid stenosis, coronary artery calcium score categories and polygenic risk scores were also comparable. In multivariable analysis, IBD was not associated with higher plaque burden (adjusted OR 1.49; 95% CI 0.88 to 2.55; p=0.139).

CONCLUSIONS: In this population-based cohort, IBD was not associated with increased subclinical carotid atherosclerosis. These findings suggest that cardiovascular risk in IBD may not be fully explained by atherosclerotic burden alone and may involve inflammation-related and prothrombotic mechanisms beyond atherosclerosis.

PMID:42601172 | DOI:10.1136/bmjgast-2026-002278