The use of intraoral scanning (IOS) has revolutionized prosthodontics by enabling fully digital workflows that enhance procedural efficiency and patient comfort [[1], [2], [3], [4], [5], [6]]. A key component of this workflow is the precise capture of the maxillomandibular relationship, which ensures proper articulation of the arches in the maximum intercuspal position (MIP) [[7], [8], [9], [10], [11], [12], [13], [14]]. Modern IOS systems generate virtual occlusal records (VORs) by digitally registering the occlusion through bilateral buccal scans [[15], [16], [17], [18], [19], [20], [21]].

The trueness of VORs is influenced by multiple factors, including patient-specific anatomy [8,12,[22], [23], [24], [25]], variability in IOS hardware and software platforms [[26], [27], [28], [29]], length of the scanning span [28,30,31], registration method [[32], [33], [34], [35]], and environmental factors [11]. Although it has been shown that VORs have comparable accuracy to conventional techniques in cases of minor tooth loss [20], their reliability decreases markedly when extensive distal edentulism is present, particularly in the mandible [3,9,10,13,36,37]. This challenge is compounded by the natural changes in tooth positions over time, which can lead to occlusal variations in implant-supported prostheses that require careful long-term monitoring [38].

A persistent technical challenge in the construction of VORs is the occurrence of occlusal collisions, defined as the inter-penetration or intersection of opposing digital models. These artifacts can compromise the final restoration, leading to clinical inaccuracies such as inadequate occlusal contacts or under-contoured prostheses [7,20,28,36]. To mitigate this, several IOS platforms have integrated features of artificial intelligence-driven occlusal contact adjustment (AI-OCA) to automatically resolve these interferences during post-processing. Preliminary evidence indicates the promise of this approach, with one study reporting that AI-OCA improved VOR trueness in a fully dentate subject, supporting its potential clinical utility [10]. However, the broader reliability of AI use in occlusal applications is controversial. For instance, a study assessing the effectiveness of AI-based prediction of the MIP reported translational and rotational discrepancies of up to 1.3 mm and 1.5°, respectively [39]. Robust data on the performance of AI-OCA strategies are limited, particularly for challenging cases involving mandibular free-end partial edentulism treated with dental implants.

The present in vitro study was designed to evaluate the effect of AI-OCA on the trueness of VORs captured using two widely used IOS systems in clinical scenarios involving either free-end edentulism or single molar defects requiring implant rehabilitation. The null hypothesis posited that AI-OCA would not significantly affect VOR trueness in either scanner group or clinical scenario.