To conduct this review, relevant references were identified through the comprehensive searches in PubMed, Springer, ACM Digital Library and ScienceDirect databases. The search terms utilized for literature retrieval encompassed “Ocular cancer” OR “Eye cancer,” in combination with “ChatGPT application” OR “Open AI application”.

Table 1 illustrates the strategic employment of search keywords across digital libraries.

Data extraction was performed by authors in three steps as follows:

All articles covering the topic were selected for the final inclusion in the review.

Authors (SA, AR, and MH) contributed to the data extraction from the full-length articles. The search scope of this study was limited to the AI application and ocular cancer. Other articles were manually searched and added to the review to provide additional knowledge on the topic. Overall, 20 articles were selected for the inclusion in this review (Figure 1). For this purpose, we followed the PRISMA guidelines (17).

Studies’ inclusion and exclusion criteria are given as follows:

Full-text research articles published in English language were considered in this narrative review. Specifically, articles published within the last 4 years (from 2020 to 2023) were eligible for inclusion. To ensure the integrity of the review, duplicate articles were excluded. Additionally, grey literature, and book chapters were omitted from this review study. Moreover, research articles that did not relate to ocular cancer and ChatGPT were also excluded.

Titles and abstracts of studies were checked for de-duplication. Researchers of this study independently reviewed each document against the eligibility criteria. Following the initial screening, full-text records were collected to confirm the eligibility. No disagreement was observed between reviewers.

To investigate the strength of ChatGPT in the identification of ocular cancer types, we present the comparison of literature and ChatGPT-generated text on ocular cancer types. Prominent venues published articles on various types of ocular cancer. They reported Retinoblastoma and Ocular Surface Squamous Neoplasia (OSSN) in the study and found that the former type was more common than the latter one (18). In the following table, we outline ocular cancer types, their respective causes, treatments, future directions and relevant references.

The data presented in Table 2 varies due to differences in treatment oprions, location, stage and individual patient characterization. Seven types of ocular cancer have been identified from the ChatGPT responses. UV exposure is mainly pointed out as the main cause of different ocular cancer types. The table outlines future directions, offering potential areas where ongoing research projects could be efficiently implemented. Aside from the causes of ocular cancer, literature reports suggest that genetic mutation, ultraviolet radiation, age, environmental exposures, viral infections and hereditary conditions may also contribute to eye diseases (29, 30). Notably, Ocular Adnexal Malignancies has been missed by the ChatGPT application (31). Regarding treatments as listed in Table 2, the shift towards personalized medicine, especially in conjunctival melanoma and ocular melanoma, reflects a growing trend in oncology, moving away from generalized treatments towards more individualized techniques. However, the treatment for Choroidal Metastasis has not been widely explored in the literature. This gap is exemplified by one of the studies by Au (28), as listed in Table 2, which is marked with a ‘Did not reveal’ status. This indicates a significant omission of a treatment strategy in the study. Furthermore, the potential role of ChatGPT in enhancing healthcare experiences, particularly for choroidal metastasis, signals an increasing recognition of AI’s role in healthcare. However, this appears as an isolated mention, indicating an early stage of AI integration in this field.

A detailed analysis of the data in the Table 2 reveals that while ChatGPT is a valuable resource, it is not without limitations as a source of information for health data, particularly in the context of ocular cancer types. Several important aspects of ocular cancer have been overlooked by the ChatGPT application. This discrepancy may stem from the data used to train ChatGPT, which remains current only until 2021. The recent advancements in ocular cancer research have not been incorporated into the training of the OpenAI application. Consequently, we are unable to access updated and highly accurate information regarding the causes and treatment of ocular cancer.

This section presents some challenges associated with the ocular cancer.

ChatGPT is designed to converse with users in natural language, responding to their inquiries and assertions in a manner that simulates online discussion (40). Capable for addressing various topics, from general knowledge questions to more specialized inquiries spanning healthcare, travel, technology, and many other topics, the Chatbot can comprehend and react to a wide range of topics.

Here are some potential ways that GPT-based chatbots could be used to overcome ocular cancer challenges:

Education and awareness: patients can get information from chatbots about ocular cancer symptoms, risk factors, and treatment options. Cancer patients continuously require online sources to educate themselves. ChatGPT can provide them information about ocular cancer. This may aid in raising awareness and promoting early detection of ocular cancer, both of which are essential for effective treatment outcomes. Recent work evaluate the role of ChatGPT in providing recommendations about breast, lungs, and prostate cancers (41). However, this study overlooks the role of ChatGPT in giving information to patients on ocular cancer.

In developing countries, dealing with the effects of ocular cancer requires a comprehensive approach. This entails disseminating knowledge about the condition, boosting patient and family support networks, expanding access to specialized care, and promoting early identification through routine eye exams. Additionally, putting regulations in place to reduce the cost of ocular cancer treatments will aid in more effective management and treatment for affected individuals. Beyond these ChatGPT applications, existing literature also emphasizes the potential applications of ChatGPT in ocular cancer. Table 3 presents a summary of the literature on ocular cancer along with its ChatGPT applications.

The current literature on cancer and related topics regarding ChatGPT applications is rapidly emerging. Numerous published works have highlighted the role of ChatGPT applications in healthcare, with a particular focus on cancer (46, 47). However, only a limited number of articles have been published on the potential role of ChatGPT applications in enhancing the health conditions of ocular patients. These studies primarily concentrated on generating operative notes during surgical procedures (42, 43). This demonstrates a clear trend towards leveraging the ChatGPT application to enhance efficiency, as highlighted by its use in answering specialized questions and documenting operative notes. While ChatGPT is emerging as a useful tool in oncology, its effectiveness is tempered by concerns over accuracy and comprehension. For instance, ChatGPT achieves only 50% accuracy in some cases (44), and shows limitations in understanding complex medical treatments. In addition, Data privacy has not been adequately addressed in the process of generating these operative notes (44, 45). The data utilized in surgical operations is highly sensitive and must be shared with the surgical staff in a cautious and meticulous manner.

ChatGPT application demonstrates its widespread use in many areas. GPT’s role in identifying the rare eye disease has been explored in a research study (48). ChatGPT works as a consultation assistant for patients and family physicians for referral suggestions. In addition, ChatGPT application helps junior ophthalmologist in diagnosing the rare eye diseases. All ChatGPT users need to be careful and cautious to acknowledge the adequate referrals and verifications in clinical settings. ChatGPT application is being applied in the world of intelligent diagnostics, Literature reveals that ChatGPT has shown potential in radiology and interpretation of the clinical images (49). It has remarkably improved the clinical workflow and responsible utilization of radiology services. Higher accuracy in clinical decision making has been seen using the ChatGPT application (50). Trusting the ChatGPT application is crucial for its adoption in healthcare, considering that ChatGPT wasn’t primarily developed for healthcare purposes. Overreliance on this technology might potentially result in disseminating false information and health risks (51). Therefore, our efforts should focus on further enhancing the ChatGPT application to differentiate between queries that it can manage effectively and those that should be redirected to human experts.

Overall, these insights indicate a future in which AI could play a supporting and integral role in ocular cancer, necessitating continuous advancements.

Currently, AI applications have demonstrated a significant role in ophthalmology, particularly in the areas of imaging and data measurement. Conventional methods to diagnose ophthalmic disease depend on the clinical assessment and image capturing devices for the different modalities (52). These diagnostic methods are time consuming and costly, and make ophthalmology one of the areas suited to the recent deep learning models. ChatGPT application is widely used in training curricula in medicine. Although ChatGPT application is still in the research stage, it has shown promising results, particularly in ocular cancer, where the potential role of AI is extensively explored in the literature (53). In this context, ChatGPT serves as a valuable tool to assist researchers and practitioners in successfully transitioning to AI-based approaches in the field of ocular cancer. By leveraging the capabilities of ChatGPT, professionals can benefit from its insights and assistance, ultimately leading to improved outcomes for patients. By sharing insights and potential data on the AI platform, physicians may promote greater communication among other doctors and patients (54). This way, physicians can make better decisions by accessing the patients’ data and opinions from colleague doctors.

Recent work explores the use of ChatGPT in the simulated “Ophthalmic Knowledge Assessment Program” OKAP examination. ChatGPT achieved an accuracy of 59.4% on the OpthoQuestions testing set compared to a 74% accuracy by human on the Basic and Clinical Science Course (BCSC) test set (13). Another research study reported that ChatGPT attained 46% accuracy in answering OpthoQuestions for the preparation of board examinations (44). Compared to human performance on the testing set, ChatGPT exhibited slightly lower performance (55). However, ChatGPT’s performance is noteworthy and could be further enhanced with the use of updated versions of ChatGPT. The latest versions of ChatGPT can be trained with the most recent oncology data, ensuring higher accuracy and relevance in healthcare. The recent version of ChatGPT (v. 3.0) demonstrated limitations in controlling a question’s difficulty as well as its cognitive level. This limitation should be considered for future versions of ChatGPT applications.

In a most recent study, it was revealed that ChatGPT’s performance in offering recommendations for cancer treatment deviated from NCC guidelines (56). This is a major concern that needs to be addressed in the near future. Although, ChatGPT largely adhered to the NCC guidelines, instances of partial deviation should be addressed in upcoming versions of the ChatGPT application.

ChatGPT could be future of the AI technology due to its features. However, we need to evaluate and monitor it while using for the communication of the eye cancer. Therefore, potential bias must be eliminated to provide equal information to all populations. ChatGPT shows limitations to answer the queries on the updated information (57). ChatGPT is trained on the data collected prior to 2021, and emerging information on eye cancer is not accurate and its accuracy may be increased due to its training on the data collected to date. While ChatGPT version 4.0 has addressed this deficiency to some extent, there is a continued need for consistent enhancements in future versions. A recent study reports that none of the existing studies explored the proposal of AI model to diagnose the Retinoblastoma (58). Common types of Retinoblastoma diagnosis were based on the clinical examination and special signs (59). One of the leading works used AI and Machine learning ML models to analyze the images and numerical data (60). At this stage, the interpretation of results is a very sensitive task in cancer research. Therefore, ChatGPT could be incorporated to make interpretation easier in studies involving the data regarding ocular cancer. Additionally, AI models may be used in the future to detect and refer ocular oncologies, specifically focusing on the lesions where treatment is required.

The emphasis on telemedicine applications extends to ocular cancer. Recent work highlights the considerations for implementing teleophthalmology in Brazil as a future research direction (61). Incorporating ChatGPT application with other technologies may give rise to favourable approaches for physicians and patients in oncology.

A large volume of unstructured information in the form of electronic health records (EHRs) is rather difficult to analyze using traditional techniques (62). ChatGPT has the potential to revolutionize oncology by analyzing vast amounts of information, uncovering trends and patterns that might elude human detection. This innovative tool could enable physicians in oncology to make more precise decisions regarding diagnosis, treatment, and cancer prevention, leading to highly personalized treatment plans tailored to individual patients’ needs. Moreover, the strategic integration of ChatGPT with EHRs could significantly reduce the workload of frontline healthcare workers in oncology, providing crucial support and alleviating workforce shortages, particularly during times of crisis (63). Due to increased concerns about data security and privacy among AI ChatGPT users, it could be integrated with blockchain technology to ensure data consistency and protect medical data, thereby providing traceable and secure usage in oncology medicine (64). Furthermore, for the advancement of medical research in oncology, ChatGPT could become a faster and more reliable source for searching medical information and tracking the progress in the research field.

Teleconsultants benefit from ChatGPT application as they can receive timely information regarding medical conditions, symptoms, treatment and medication (49). This may help teleconsultants understand patients’ concerns, needs and expectations. In telemedicine, routine tasks include scheduling appointments, sending reminders, and refilling prescriptions. ChatGPT helps explore the virtual patient-physician interaction. A virtual assistant based on ChatGPT may be developed to assistant healthcare professionals in telemedicine to triage patients, and provide remote guidance for home care (65). However, we have observed that participants in such studies may encounter several issues while using the AI application. For example, Caruccio et al. (49) reported that during traditional consultations, patients communicate with physicians who highly prioritize the privacy and confidentiality of their interactions. However, a physician using the ChatGPT application may inadvertently create ambiguity for patients, who have a legitimate right to know whether they are interacting with an AI application or a teleconsultant. Moreover, patients’ privacy and security could potentially be compromised when physicians input sensitive information such as conditions, age, sex, and other factors into the AI system for suggestions. Consequently, significant ethical concerns regarding data privacy, security, accountability, and liability may impose limitations on the utilization of ChatGPT in telemedicine. During the COVID-19 pandemic, ocular oncology faced new challenges with a focus on preventing SARS-CoV-2 exposure. Telemedicine stands out as a major option, setting new standards for ocular treatment during unprecedented times (66). However, a major challenge is to implement the safety measures for healthcare workers and patients to enhance ocular cancer diagnosis and treatment practices.

We noted “unknown” causes for several ocular cancers (25, 27), highlighting areas for future research. Additionally, the ‘Did not reveal’ status under the treatment for choroidal metastasis (28), suggests a lack of consensus or insufficient information regarding treatments, identifying a critical area for further research and development.