大型语言模型（large language models, LLMs）在眼科的应用为医疗领域带来了巨大的潜力，尤其是在提升诊断效率、优化医患沟通和促进个性化医疗方面。通过自然语言处理技术，LLMs可以协助医生进行临床数据的归纳和分析，可以结合患者的病史、影像资料和症状描述，提供精准的辅助诊断，并在复杂病例中提供参考。LLMs还可以帮助医生快速撰写病历报告，改善医疗记录管理效率。在医患沟通中，LLMs能够通过生成通俗易懂的解释，帮助患者理解疾病状况及治疗方案，缩短医生与患者之间的沟通障碍。在远程医疗场景下，LLMs可通过实时分析患者上传的图像文本信息，提供初步诊断建议，助力医生远程诊疗。个性化医疗也是LLMs的重要应用方向，借助患者的遗传数据和生活习惯，可以帮助医生制定更为精准的个性化治疗方案，并预测手术后的恢复情况。此外，LLMs可以通过与临床数据的不断交互进行自我优化，提升其在眼科诊疗中的智能化程度。尽管LLMs在眼科领域的应用前景广阔，但仍面临数据隐私、模型解释性、语言理解等方面的挑战。未来LLMs将继续作为医生的辅助工具，形成“人机协同”的诊疗新模式，为患者提供更好、更精准的医疗服务。

The application of Large Language Models (LLMs) in ophthalmology presents tremendous potential for the healthcare field, particularly in enhancing diagnostic efficiency, optimizing doctor-patient communication, and promoting personalized medicine. Through natural language processing technology, LLMs can assist doctors in summarizing and analyzing clinical data. They can integrate a patient's medical history, imaging data, and symptom descriptions to provide precise diagnostic support and reference for complex cases. LLMs can also help physicians quickly draft case reports, improving the management efficiency of medical records. In doctor-patient communication, LLMs can generate easy-to-understand explanations that help patients comprehend their conditions and treatment plans, thereby reducing communication barriers between doctors and patients. In telemedicine scenarios, LLMs can provide preliminary diagnostic suggestions by real-time analyzing images and textual information uploaded by patients, aiding doctors in remote diagnosis and treatment.

Personalized medicine is another significant application direction for LLMs. By utilizing patients' genetic data and lifestyle habits, LLMs can assist physicians in formulating more precise personalized treatment plans and predicting postoperative recovery outcomes. Additionally, LLMs can self-optimize through continuous interaction with clinical data, enhancing their intelligence in ophthalmic diagnosis and treatment. Despite the broad application prospects of LLMs in the field of ophthalmology, challenges remain, including data privacy, model interpretability, and language understanding. In the future,  LLMs will continue to serve as auxiliary tools for physicians, forming a new model of "human-machine collaboration" in diagnosis and treatment, ultimately providing better and more precise medical services to patients.

青光眼是全球第二大致盲眼病，第一大不可逆性致盲眼病，其中原发性闭角型青光眼（primary angle closure glaucoma，PACG）占 25%。激光周边虹膜切除术（laser peripheral iridotomy，LPI）已成为 PACG 和原发性房角关闭的一线治疗。LPI 机制为利用激光在周边虹膜上打孔，解除 PACG 的瞳孔阻滞，加深前房，扩大房角，恢复生理性房水排出途径，从而降低眼压。研究表明 LPI 在原发性房角关闭各个疾病进程中均能比较好的控制眼压，是相对安全的治疗方法。

Glaucoma is the second leading cause of blindness and the most common cause of irreversible blindness worldwide. Primary angle closure glaucoma (PACG) accounts for 25% of glaucoma. Laser peripheral iridotomy (LPI) has become the first line treatment for PACG and primary angle closure (PAC). The mechanism of LPI is to use laser to create a hole in peripheral iris to relieve pupil block, deepen anterior chamber, expand chamber angle, restore pathway of physiological aqueous discharge and reduce intraocular pressure. Studies have shown that LPI can control intraocular pressure well in all stages of PAC, which is safe for PAC.

青光眼是全球第二大致盲眼病，第一大不可逆性致盲眼病，其中原发性闭角型青光眼(primary angle closure glaucoma，PACG)占25%。激光周边虹膜切除术(laser peripheral iridotomy，LPI)已成为PACG和原发性房角关闭的一线治疗。LPI机制为利用激光在周边虹膜上打孔，解除PACG的瞳孔阻滞，加深前房，扩大房角，恢复生理性房水排出途径，从而降低眼压。研究表明LPI在原发性房角关闭各个疾病进程中均能比较好的控制眼压，是相对安全的治疗方法。

Glaucoma is the second leading cause of blindness and the most common cause of irreversible blindness worldwide. Primary angle closure glaucoma (PACG) accounts for 25% of glaucoma. Laser peripheral iridotomy(LPI) has become the first line treatment for PACG and primary angle closure (PAC). The mechanism of LPI is to use laser to create a hole in peripheral iris to relieve pupil block, deepen anterior chamber, expand chamber angle,restore pathway of physiological aqueous discharge and reduce intraocular pressure. Studies have shown that LPI can control intraocular pressure well in all stages of PAC, which is safe for PAC.