Performance of ChatGPT in dental implant treatment planning: evaluation using the modified DISCERN, Global Quality Score, and accuracy–safety score

Large language models (LLMs) such as ChatGPT are increasingly used to access health-related information, including in dental implant treatment planning. However, the reliability, quality, and clinical accuracy of information provided by these AI systems remain uncertain, raising potential concerns for patient safety. This study aims to systematically evaluate ChatGPT’s performance in dental implant treatment planning, focusing on the reliability, quality, and clinical validity of its responses. Sixty clinical scenarios for dental implant treatment were designed and presented to ChatGPT. Scenarios were divided into two categories: (1) patients with deficient alveolar bone and (2) patients with systemic conditions. AI-generated responses were independently evaluated by three board-certified oral and maxillofacial surgeons. Information reliability was assessed using the Modified DISCERN instrument, overall content quality using the Global Quality Score (GQS), and clinical accuracy and safety using a Likert scale. Statistical analyses were performed with IBM SPSS Statistics 26.0. Data normality was evaluated with the Shapiro–Wilk test, and non-parametric comparisons were conducted using the Mann–Whitney U test and Spearman rank correlation. Statistical significance was set at p < 0.05. GQS was significantly higher for systemic disease scenarios (3.83 ± 0.69) compared to bone deficiency scenarios (3.20 ± 0.40) (U = 675, p < 0.05). Correlation analyses revealed significant positive relationships among the evaluation scales. Specifically mDISCERN score showed a weak positive correlation with GQS (r = 0.325; p = 0.011) and a moderate positive correlation with clinical Accuracy & Safety (r = 0.535; p < 0.001). In bone deficiency scenarios, mDISCERN correlated moderately with GQS (r = 0.512; p = 0.004) and strongly with Accuracy & Safety (r = 0.651; p < 0.001), while GQS also demonstrated a strong correlation with Accuracy & Safety (r = 0.682; p < 0.001). For systemic disease scenarios, mDISCERN showed a moderate positive correlation with Accuracy & Safety (r = 0.473; p = 0.008). ChatGPT can be used as an adjunctive tool in dental implant planning but should not replace professional clinical judgment. Safe and effective implant management requires adherence to evidence-based guidelines and consultation with experienced specialists.