Objective To evaluate the cognitive abilities of the leading large language models and identify their susceptibility to cognitive impairment, using the Montreal Cognitive Assessment (MoCA) and additional tests. Design Cross sectional analysis. Setting Online interaction with large language models via text based prompts. Participants Publicly available large language models, or “chatbots”: ChatGPT versions 4 and 4o (developed by OpenAI), Claude 3.5 “Sonnet” (developed by Anthropic), and Gemini versions 1 and 1.5 (developed by Alphabet). Assessments The MoCA test (version 8.1) was administered to the leading large language models with instructions identical to those given to human patients. Scoring followed official guidelines and was evaluated by a practising neurologist. Additional assessments included the Navon figure, cookie theft picture, Poppelreuter figure, and Stroop test. Main outcome measures MoCA scores, performance in visuospatial/executive tasks, and Stroop test results. Results ChatGPT 4o achieved the highest score on the MoCA test (26/30), followed by ChatGPT 4 and Claude (25/30), with Gemini 1.0 scoring lowest (16/30). All large language models showed poor performance in visuospatial/executive tasks. Gemini models failed at the delayed recall task. Only ChatGPT 4o succeeded in the incongruent stage of the Stroop test. Conclusions With the exception of ChatGPT 4o, almost all large language models subjected to the MoCA test showed signs of mild cognitive impairment. Moreover, as in humans, age is a key determinant of cognitive decline: “older” chatbots, like older patients, tend to perform worse on the MoCA test. These findings challenge the assumption that artificial intelligence will soon replace human doctors, as the cognitive impairment evident in leading chatbots may affect their reliability in medical diagnostics and undermine patients’ confidence.
Background: Compensatory mechanisms in Parkinson's disease (PD) are thought to explain the temporal delay between the beginning of the neurodegenerative process and the appearance of clinical signs. The enhanced structural integrity of the corticospinal tract was previously suggested as one of these mechanisms. Objective: To understand the relations between corticospinal tract integrity and the anatomical, clinical, electrophysiological, and genetic PD characteristics. Methods: We analyzed diffusion tensor imaging (DTI) fractional anisotropy (FA) data from 40 genotyped patients with PD (18 without known genetic cause, 11 with LRRK2 -PD and 11 with GBA -PD) who were candidates for subthalamic deep brain stimulation (STN-DBS) and from 25 healthy, age-matched, controls. Results: PD is associated with higher corticospinal FA values (p < 0.001) that are negatively correlated with the disease duration (p = 0.032), confirming previous results. Higher FA values are negatively correlated with cerebral grey matter volumes (p < 0.001) but not with the motor or cognitive PD characteristics, or with the subthalamic beta-oscillatory activity measured intra-operatively. Increased corticospinal FA values are strongly affected by the genetic etiology of PD, with higher values in the monogenic forms of the disease (p < 0.001). The compensatory index, calculated by dividing the corticostriatal FA value by the cerebral grey matter volume, is highest in GBA -PD (p < 0.001) at the time of evaluation for STN-DBS. Conclusions: The genetic etiology of PD strongly shapes corticospinal tract changes along with disease-duration and cerebral grey matter atrophy. The changes may serve as compensatory mechanism.
Aim: Neurological manifestations are common in patients with chronic liver diseases. This study aimed to depict the association between liver cirrhosis and Parkinson’s disease (PD) and propose a clinically relevant diagnostic scheme. Methods: We examined patients’ medical records with PD and chronic liver impairment secondary to cirrhosis or liver metastases for temporal correlations between liver insult and Parkinsonian signs. Results: Thirty-five individuals with PD and chronic liver impairment were included due to either cirrhosis or liver metastases. In all 22 patients with PD and liver metastases, the diagnosis of PD preceded the diagnosis of cancer. Conversely, patients with cirrhosis were often diagnosed with liver impairment before diagnosing PD. Age at diagnosis did not account for this difference. Conclusions: This study reinforces the potential clinical association between cirrhosis and PD. We also provide a diagnostic scheme that may guide therapeutic interventions and prognostic assessments.
Blood transcriptomic signatures for Parkinson’s disease (PD) diagnosis have failed to integrate into clinical practice despite decades of translational efforts. We evaluated the classification performance of 13 published coding RNA-based signatures using both a large public dataset and data we collected prospectively in a controlled clinical study of levodopa-naïve patients and healthy controls. Our results show that gene overlap between signatures is low but significant (2.7%, p < 0.001) and enriched for lipid metabolism genes. Most signatures (10/13) remained significant when tested on the Parkinson’s Progression Markers Initiative (PPMI) dataset, though with lower classification performance than previously reported (median AUC: 59.7%). Performance improved for GBA1-associated PD. Rigorous standardization of clinical and environmental parameters in our prospective study (30 participants) failed to improve transcriptome-based classification. We conclude that while the search for a universal blood-based PD transcriptome may elucidate disease pathophysiology, its clinical utility is inherently limited.