To compare the accuracy, scientific adequacy, and clarity of responses provided by ChatGPT-4o and Gemini to frequently asked patients' questions about warfarin use. Forty patients' questions were posed to ChatGPT and Gemini using the zero-shot method. Four cardiovascular surgeons evaluated the responses for adequacy, scientific accuracy, and clarity on a 5-point Likert scale. The first and second set of data evaluations were separated by seven days to minimize any memory effect. The experts were blinded to the type of LLM that created the response. ChatGPT responses were significantly shorter (187.3±47.6 vs 291.4±98.1 words; P<0.001) but scientifically more adequate (4.38±0.30 vs 4.17±0.35; P=0.004). Gemini scored higher in terms of clarity (4.69±0.24 vs 4.48±0.33; P<0.001). The two models did not significantly differ in terms of accuracy (P = 0.606). Both LLMs provide reliable and understandable information for warfarin counseling. While ChatGPT provides dense and scientifically detailed responses, Gemini excels in clarity and user-friendly communication. However, expert supervision and appropriate clinical guidance are critical for safe and comprehensive guidance.

OBJECTIVE: The aim of the present study was to determine the effect of systemic inflammatory response index in predicting the development of postoperative atrial fibrillation after isolated coronary artery bypass grafting (CABG).

Polytetrafluoroethylene (PTFE) grafts are of great importance for vascular surgery and many methods have been developed to improve their biocompatibility. The most important of these methods is the coating of the inner surfaces of the grafts. In this study, the effects of surface coatings used in vascular grafts on endothelialization and bacterial biofilm formation were investigated. Three different PTFE graft types, heparin coated, carbon coated and uncoated, were compared. HUVEC cell culture was used for endothelialization experiments and Staphylococcus aureus strain was used for biofilm formation. Endothelialization was evaluated by inverted microscopy and scanning electron microscopy (SEM). Heparin-coated grafts showed more biofilm formation than other graft types ( p < 0.01). Moderate biofilm formation was observed in carbon-coated grafts ( p < 0.05). When evaluating endothelialization, heparin-coated grafts showed more cell adhesion in the first days, but lagged behind the other graft types in the following days. Carbon-coated grafts showed more endothelial cell proliferation in the long term. While biofilm formation was high in heparin-coated grafts, carbon-coated grafts provided better endothelialization. Our study showed that the coating of PTFE grafts significantly affects biocompatibility and infection risk.

Objective: Vascular access is a key component of effective hemodialysis treatment. Patient education regarding selection, complications, and daily care remains challenging. Recently, large language models (LLMs) have been explored as supportive tools in this field. This study compared three widely used LLMs (ChatGPT-5, Gemini-2.5, and DeepSeek-V3) in addressing patient-centered questions on hemodialysis vascular access. Methods: Twenty-five frequently asked patient questions were compiled from literature, educational materials, and expert input. Each question was submitted to the three LLMs in standardized sessions, and answers were anonymized. Four cardiovascular surgeons independently evaluated responses in a blinded manner using 5-point Likert scales for accuracy, clarity, and scientificity. Statistical analyses compared model performances. Results: DeepSeek-V3 provided significantly responses with a higher word count and achieved higher scientific depth scores compared to ChatGPT-5 and Gemini-2.5 (P<0.01). Accuracy scores showed ChatGPT-5 demonstrated significantly lower than both Gemini-2.5 and DeepSeek-V3 (P<0.001). No significant differences were observed among models regarding clarity. Overall, DeepSeek achieved the highest mean scores across all criteria. Conclusion: DeepSeek-V3 demonstrated superior scientific depth and overall reliability, whereas ChatGPT-5 showed relative weaknesses in accuracy. The comparable clarity across models highlights LLMs’ potential as supportive tools for patient education. Future studies should further validate these tools in real-world clinical settings and define appropriate safeguards for their use.