ChatGPT as a Bridge between Cancer Patients and AI-based Diagnostic Tools (Preprint)

Abstract

BACKGROUND Cancer diagnosis and treatment is highly dependent on conventional pathology techniques such as histopathology, cytopathology and ancillary approaches. Artificial intelligence (AI)-powered data processing and learning has emerged as a key driver for medical progress, and pathology and laboratory medicine are no exception. OBJECTIVE This study highlights the significance of cancer diagnosis within the framework of ChatGPT, a language model driven by "artificial intelligence" (AI). Through data analysis and useful answer generation, ChatGPT offers invaluable insights and support for cancer diagnosis, treatment, and patient education. METHODS The purpose of this paper is to create 21 cancer-related questions. We used Google Bard and the most recent version of ChatGPT 4.0 to get responses. We performed an intermodal evaluation and double-checked the answers to these inquiries. RESULTS We were able to identify 12 prevalent cancer types and provide information on their symptoms, recommended course of treatment, and available medications. The precision and dependability of the data provided by ChatGPT were confirmed by the use of publications from the National Cancer Institute (NCI) and the PubMed online databases. In addition, we compared the output produced by the ChatGPT application using Bard, a large language model. The results demonstrated that the ChatGPT application can identify the most serious types of cancer, including pancreatic, lung, and brain malignancies. CONCLUSIONS This research indicates that future developments in AI technology and real-time data integration can further enhance the usefulness of ChatGPT and other similar applications in cancer management and treatment.