ChatGPT, a chatbot that can create human-like written responses to the most complex queries, has attracted attention for its ability to generate answers to creative questions. Is it possible to detect Alzheimer’s Disease early in its development using the algorithms of ChatGPT?
Recent research from Drexel University’s School of Biomedical Engineering Science and Health Systems has shown that OpenAI’s GPT-3 can detect clues in spontaneous speech and predict the early stages of dementia. It is 80% accurate, according to Drexel University’s School of Biomedical Engineering Science and Health Systems.
The Drexel study was published in PLOS Digital Health. It is the latest in a series to demonstrate the effectiveness of natural-language processing programs for the early detection of Alzheimer’s. This leverages current research that suggests language impairment may be an early indicator of neurodegenerative diseases.
The current method for diagnosing Alzheimer’s Disease is to review a patient’s medical history and conduct a lengthy series of neurological and physical tests. Although there is no cure, early detection can help patients find the right treatment and support.
Language impairment is a sign of dementia in between 60 and 80% of patients. Researchers have focused on programs that detect subtle clues such as hesitation, grammar mistakes, forgetting the meaning, and spelling errors. This quick test could help determine if a patient was examined.
GPT-3 is the third generation OpenAI’s General Pretrained Transformer. It uses a deep-learning algorithm that was trained by processing large amounts of data from the internet. The focus is on the use of words and the construction of language.
Thanks to this training, it can respond to simple questions and write poems or essays with a human-like style. ChatGPT excels at “zero data learning,” which means it can answer questions that would otherwise require external knowledge.
If you ask the program to write Cliff’s Notes of a text, it would normally need an explanation as to what this means. GPT-3 has had enough training to comprehend the reference and adapt to the expected answer.
Related Stories: