GPT-3 Alzheimer

A recently concluded study from Drexel University’s School of Biomedical Engineering, Science and Health Systems suggests that OpenAI’s GPT-3 program may be able to spot clues from conversations which could be used to identify the early stages of dementia. This accuracy rate of prediction was reported to be 80%.

Alzheimer’s disease is a degenerative brain disorder which currently has no cure. As speech is a human ability, it can be used to measure the onset of neurodegenerative diseases like Alzheimer’s. Is it possible for AI to analyse conversations and detect early signs of dementia? The study suggests that GPT-3, a language model created by OpenAI, could help with this. Specifically, the study showed that text embedding, powered by GPT-3, could be used to accurately distinguish individuals with Alzheimer’s from healthy controls, and even predict the subject’s cognitive testing score, based solely on speech data. It was further demonstrated that text embedding outperforms conventional feature-based approaches, and even performs as well as fine-tuned models. It is suggested that AI-driven speech analysis has the potential to be used to diagnose dementia at an early stage, providing tailored interventions to individuals to improve their quality of life.

Source: https://journals.plos.org/digitalhealth/article?id=10.1371/journal.pdig.0000168