Assessing the Role of the Generative Pretrained Transformer (GPT) in Alzheimer’s Disease Management: Comparative Study of Neurologist- and Artificial Intelligence–Generated Responses
2024

Using AI to Help Manage Alzheimer's Disease

Sample size: 14 publication 10 minutes Evidence: high

Author Information

Author(s): Eysenbach Gunther, Leung Tiffany, Zhu Lingxuan, Giannakopoulos Kostis, Kavadella Argyro, Zeng Jiaqi, Zou Xiaoyi, Li Shirong, Tang Yao, Teng Sisi, Li Huanhuan, Wang Changyu, Wu Yuxuan, Zhang Luyao, Zhong Yunheng, Liu Jialin, Liu Siru

Primary Institution: West China Medical School, Sichuan University

Hypothesis

Can the Generative Pretrained Transformer (GPT) provide accurate and useful responses for managing Alzheimer's disease compared to neurologists?

Conclusion

GPT can provide valuable educational materials for managing Alzheimer's disease, improving outcomes for patients and caregivers.

Supporting Evidence

  • GPT-generated responses scored significantly higher than neurologist-generated responses in accuracy and comprehensibility.
  • English responses from GPT outperformed Chinese responses in terms of clarity and participant satisfaction.
  • Evaluator neurologists rated GPT responses as more satisfactory compared to those from neurologists.

Takeaway

This study shows that an AI called GPT can give helpful answers about Alzheimer's disease, which can help patients and their families understand the disease better.

Methodology

The study involved posing 14 questions about Alzheimer's disease to both GPT models and neurologists, with responses rated by evaluators.

Potential Biases

Subjective judgments from evaluators may introduce bias in the assessment of responses.

Limitations

The sample size was small, and GPT's training data only goes up to 2021, which may affect the accuracy of its responses.

Participant Demographics

Participants included 9 neurologists and 5 family members of patients with Alzheimer's disease, with varying ages and backgrounds.

Statistical Information

P-Value

p<0.001

Statistical Significance

p<0.001

Digital Object Identifier (DOI)

10.2196/51095

Want to read the original?

Access the complete publication on the publisher's website

View Original Publication