ChatGPT-4 Stuns Researchers with its 40% CPC Diagnosis Match Rate

The potential of Artificial Intelligence (AI) to revolutionize various industries, including healthcare, is widely recognized. Among the multiple types of AI, Generative AI stands out as one with immense promise, as it can create new content based on patterns it has learned.

Recently, a team of physician-researchers at Beth Israel Deaconess Medical Center (BIDMC) studied Generative Pretrained Transformer 4 (GPT-4), a powerful generative AI chatbot. The study showed that GPT-4 has impressive diagnostic capabilities, offering exciting possibilities for the future of healthcare.

Understanding Generative AI

Unlike traditional AI, which primarily processes and analyzes existing data, generative AI learns patterns and generates new content. This technology powers the current generation of AI chatbots, which leverage natural language processing (NLP) – a branch of AI that enables machines to understand, interpret, and generate human-like language.

These chatbots are already revolutionizing various domains, including education, customer service, and creative industries. However, their effectiveness in a clinical setting remained largely unexplored until now.

Purpose of the Study

“Recent advances in artificial intelligence have led to generative AI models that are capable of detailed text-based responses that score highly in standardized medical examinations,” stated Adam Rodman, MD, MPH, co-director at BIDMC. Dr. Rodman and his team aimed to ascertain whether generative AI could bring this proficiency to real-world, complex diagnostic scenarios.

The team assessed GPT-4‘s diagnostic ability through clinicopathological case conferences (CPCs), which presented various complex patient cases and relevant clinical and laboratory data, imaging studies, and histopathological findings. The study evaluated GPT-4’s performance in diagnosing 70 of these complex cases.


The AI system showed impressive performance, matching the final CPC diagnosis in almost 40% of cases. In addition, GPT-4 accurately provided the correct diagnosis in 66% of cases and presented a list of potential diagnoses based on the patient’s symptoms, medical history, and clinical findings.

Zahir Kanjee, MD, MPH, a hospitalist at BIDMC and first author of the study, stated, “While Chatbots cannot replace the expertise and knowledge of a trained medical professional, generative AI is a promising potential adjunct to human cognition in diagnosis.” Byron Crowe, MD, an internal medicine physician at BIDMC and co-author of the study, added, “Our study adds to a growing body of literature demonstrating the promising capabilities of AI technology.”

Despite the encouraging results, the study underlines the need for more research to maximize the benefits and understand the limitations of AI in clinical settings. It also brings attention to privacy issues, urging the scientific community to address them as AI becomes more prevalent in the healthcare industry.

The future of AI in healthcare is promising, and this study is a significant step in that direction. As generative AI continues to evolve and enhance its diagnostic capabilities, it could become an invaluable tool for physicians.

About The Author

Leave a Reply