AI and Neurological Rare & Neuroimmune Diseases ── Parkinson's ── 2026-11-18

Parkinson's disease and parkinsonism: treatment, trials and AI
- Abstract
Large language model (LLM)-based chatbots are increasingly used for health information, but their performance in Parkinson’s disease (PD) patient education remains unclear. This study evaluated five widely used LLM-based chatbots in answering public questions about levodopa therapy. In this cross-sectional comparative study, ChatGPT, Gemini, Copilot, DeepSeek, and Doubao were assessed using a structured set of levodopa-related patient education questions. In the primary analysis, 265 responses were evaluated by five blinded raters for safety, accuracy, empathy, and information quality using predefined criteria, including DISCERN, EQIP, JAMA benchmark criteria, and the Global Quality Score. Readability was assessed using six established indices. An additional 265 responses were generated in an independent repeat-generation analysis using the same question set to assess response stability. Inter-rater agreement was high. Safe-response rates ranged from 86.8% to 94.3%. Accuracy differed significantly, with Copilot and DeepSeek showing the highest median scores, whereas empathy differences were modest and closely clustered. Doubao had the highest DISCERN and EQIP scores, while ChatGPT had the lowest information-quality scores. Gemini showed the best readability profile and Doubao the poorest. Repeat-generation analysis showed generally high stability, with 97.0% exact safety agreement and largely preserved model rankings. Current LLM-based chatbots generated generally informative English-language responses about levodopa therapy, but safety gaps, variable information quality, and poor readability remained. These findings reflect expert-rated output quality rather than demonstrated patient education effectiveness, and patient-centered validation is warranted.
Journal IF-equivalent: 6.3 (OpenAlex 2-year mean citedness, value as of 2026-11-09, retrieved 2026-11-10; not the official Clarivate IF)Reference: Zuo H, Tang Q, Wang C, Zuo X, Cheng C, Wang L, et al. Evaluating large language models for public patient education on levodopa therapy in Parkinson’s disease: accuracy, empathy, reliability, and readability. Sci Rep. 2026 Nov 16;16(1):38113. doi:10.1038/s41598-026-74929-y.Checked: Abstract only - Abstract
It remains challenging to diagnose incipient Parkinson’s disease (PD) with its ominous, insidious onset symptoms. This paper introduces a new machine learning methodology of PD diagnosis from multimodal audio and accelerometer features. We introduce the optimization of RF classifier for feature selection by Bacterial Forging Optimization (BFO) and hyperparameter optimization by Jellyfish Search (JS). We propose a new data augmentation technique, Dynamic Signal Perturbation (DSP), to enhance model robustness by simulating actual signal perturbation. Using a test set of 250 subjects (125 PD, 125 controls), our technique is 94.7% accurate, 93.0% sensitive, 96.5% specific, and 0.96 AUC, vastly better than baselines like SVM with RFE (85.6% accurate) and RF with PSO (89.2% accurate). The higher stability and efficiency of the BFO-JS-RF model make it the most suitable candidate for scalable and non-invasive PD detection in clinics. The technique will be validated in the future from actual data.
Journal IF: unknown (could not be matched)Reference: Prameela Rani B, Gnana Priya B, Abinaya R. Novel multimodal Parkinson’s disease diagnosis with BFO-JS enhanced random forest. Drug Discovery, Development and Delivery (CTD4-2025). 2026 Nov 16 [Epub ahead of print]. doi:10.1039/9781837071555-00037.Checked: Abstract only