Feb 2027· Neurology Clinical Practice· Vol 17 1, pp.
e200687
· 0 citations· 43 references
Medicine
Abstract
Purpose
OF REVIEW
Large language models (LLMs) are increasingly being evaluated for clinical neurology applications, yet their real-world utility, safety, and deployment readiness remain unclear. This review maps the literature on LLMs applied to text-based and text-derived neurology tasks, characterizing current use cases, evaluation strategies, performance metrics, and reported safety concerns. We identify areas with relatively robust evidence, highlight important evidence gaps, and outline key methodological and clinical barriers that must be addressed before responsible clinical adoption.
RECENT
Findings
Following Preferred Reporting Items for Systematic Reviews and Meta-Analyses Extension for Scoping Reviews (PRISMA-ScR), we searched 4 databases (January 2022 to February 2026) and included 32 studies (2023-2026) spanning 7 neurology subspecialties; 9 were simulation or feasibility studies, 17 analyzed retrospective clinical data, and 6 reported prospective evaluation. Proprietary models predominated; 5 studies used retrieval-augmented generation and 1 used a multiagent framework. Performance was highest for constrained tasks, including binary diagnostic classification (area under the curve 0.75-0.94), information extraction (F1 0.89-0.90), patient-education question answering (accuracy 68%-97%), and ischemic stroke thrombectomy decision support (area under the curve [AUC] 0.92) and lowest for open-ended case-based reasoning (accuracy 42%-54%). Recurrent safety signals included hallucinated and fabricated citations, poor calibration with overconfidence (for example, 100% sensitivity with 27% specificity for epilepsy diagnosis), and a systematic bias toward intervention. On adapted Quality Assessment of Diagnostic Accuracy Studies Using Artificial Intelligence (QUADAS-AI) charting, overall risk of bias was low in 1 study (3%), moderate in 16 (50%), and high in 15 (47%), driven chiefly by absent external validation (31/32, 97%). Benchmarked against high-reliability and 6 Sigma standards, reported error rates remain orders of magnitude too high for autonomous use.
SUMMARY
Evidence for LLMs in neurology is early, heterogeneous, and limited by scarce external and prospective validation. Current data support only narrow, clinician-supervised uses such as documentation and education, while cautioning against autonomous diagnostic or therapeutic use. Rigorous, prospective, externally validated studies with standardized reporting and explicit safety and ethical evaluation are needed before broader adoption.
REGISTRATION
PROSPERO CRD420251082465.
It is found that Llama 3 delivers comparable quality to leading language models such as GPT-4 on a plethora of tasks, and performs competitively with the state-of-the-art on image, video, and speech recognition tasks.
This publication proposes a definition and a classification of agile software development approaches and analyses ten software development methods that can be characterized as being "agile" against the defined criterion.
P. Abrahamsson, O. Salo, Jussi Ronkainen et al.· arXiv.org· 727 citations· ⚡54
The study shows that agile practices improve both informal and formal communication, but indicates that, in larger development situations involving multiple external stakeholders, a mismatch of adequate communication mechanisms can sometimes even hinder the communication.
M. Pikkarainen, Jukka Haikara, O. Salo et al.· Empirical Software Engineeri...· 401 citations· ⚡48
The results indicate that software engineering work practices are chosen opportunistically, adapted and configured to provide value under the constrains imposed by the startup context.
Nicolò Paternoster, Carmine Giardino, M. Unterkalmsteiner et al.· Information and Software Tec...· 394 citations· ⚡54
It is proved that any global minimizer of the trajectory balance objective can define a policy that samples exactly from the target distribution, and empirically demonstrate the benefits of the trajectories balance objective for GFlowNet convergence, diversity of generated samples, and robustness to long action sequenc...
Esmeralda S. Whitammer, Moksh Jain, Emmanuel Bengio et al.· Neural Information Processin...· 302 citations· ⚡60