Japanese Cardiologist Questions Accuracy of Generative AI Diagnostics

JAPAN: Generative AI is steadily making its way into the healthcare sector, promising to enhance patient care and streamline medical practices. Tokyo-based neurosurgeon Dr. Ryutaro Nomura has embraced this technology, utilizing AI-generated summaries to aid in diagnosing headache patients. “The AI instantly summarizes the necessary information. It enables me to ask targeted questions and conduct smooth medical examinations,” Dr. Nomura remarks, acknowledging the convenience offered by this innovative tool.

Breaking Language Barriers

AI’s potential extends beyond summarization. At HITO Medical Center in Ehime Prefecture, a translation system powered by Microsoft’s AI eliminates language barriers, enabling Japanese nurses to communicate seamlessly with foreign nursing assistants. “Thanks to the adoption of the AI-assisted chat system last summer, misunderstandings have been eliminated,” a spokesperson for the medical center affirms, highlighting the technology’s role in ensuring accurate communication and patient care.

Aoikai develops AI diagnostics system

The drive to implement generative AI in healthcare stems from a desire to optimize limited human resources. With the “workstyle reform of doctors” regulating physicians’ overtime, AI-assisted systems offer a promising solution for alleviating workloads and improving efficiency.

Overcoming Accuracy Concerns

However, despite the promising applications, concerns persist regarding the accuracy of generative AI in medical settings. Akihiro Nomura, an associate professor at Kanazawa University and a clinical cardiologist, cautions, “A medical error carries a life-threatening risk. Unless their accuracy is improved, [such AI-assisted systems] cannot be used for diagnosis and treatment.”

Evaluations of ChatGPT, a prominent generative AI system, have revealed inconsistencies and potentially dangerous inaccuracies. In one instance, ChatGPT recommended an outdated treatment for hyperventilation, a method that could lead to suffocation. Furthermore, a study by Tokyo Medical and Dental University found that ChatGPT’s responses varied significantly, with only 4% of answers correctly identifying cervical myelopathy based on provided symptoms.

AI-Driven Dentistry: Catalyst for Enhanced Patient Care and Confidence

Addressing Privacy and Responsibility

Beyond accuracy concerns, the use of generative AI in healthcare raises questions about data privacy and responsibility. “It’s necessary to discuss responsibility when [mistakes] happen,” asserts Ryozo Nagai, president of Jichi Medical University and an expert in medical AI. He emphasizes the need to train AI on more data related to Japanese patients to improve quality and clearly define the technology’s role in supporting patient care.

As the integration of generative AI in healthcare continues, it is crucial to address these challenges and ensure the safe and responsible deployment of this powerful technology, striking a balance between its potential benefits and the imperative of patient safety.