ChatGPT's dance with neuropsychological data: A case study in Alzheimer's disease

Abstract

Artificial intelligence continues to revolutionize the medical and scientific field, especially with the release of ChatGPT. We assessed whether it provides an accurate interpretation of neuropsychological screening. We provided ChatGPT with the neuropsychological data of a patient with mild Alzheimer's Disease and invited it and two neuropsychologists to interpret the data. While ChatGPT provided an accurate interpretation of scores on each of the neuropsychological tests, it did not use standardized scores and did not specify the cognitive domain that may be most impaired. In contrast, the neuropsychologists used standardized scores to determine that the patient was mainly suffering from memory decline. While ChatGPT may succeed in the general interpretation of neuropsychological testing, at least in patients with Alzheimer's Disease, it still cannot create a pattern of scores across different tests to better specify the nature of cognitive impairment.

Keywords: Alzheimer’s disease; Artificial intelligence; ChatGPT; Neuropsychological testing; Neuropsychology.