Large-language models (LLMs) show promise for automating evidence synthesis, yet head-to-head evaluations remain scarce. We benchmarked five state-of-the-art LLMs—openai/o1-mini, x-ai/grok-2-1212, meta-llama/Llama-3.3-70B-Instruct, google/Gemini-Flash-1.5-8B, and deepseek/DeepSeek-R1-70B-Distill—on extracting protocol details from transcranial direct-current stimulation (tDCS) trials enrolling older adults. A multi-LLM ensemble pipeline ingested ClinicalTrials.gov records, applied a structured JSON schema, and generated comparable outputs from unstructured text. The pipeline retrieved 83 aging-related tDCS trials—roughly double the yield of a conventional keyword search. Across models, agreement was almost perfect for the binary field brain stimulation used (Fleiss κ ≈ 0.92) and substantial for the categorical primary target (κ ≈ 0.71). Numeric parameters such as stimulation intensity and session duration showed excellent consistency when explicitly reported (ICC 0.95–0.96); secondary targets and free-text duration phrases remained challenging (κ ≈ 0.61; ICC ≈ 0.35). An ensemble consensus (majority vote or averaging) resolved most disagreements and delivered near-perfect reliability on core stimulation attributes (κ = 0.94). These results demonstrate that multi-LLM ensembles can markedly expand trial coverage and reach expert-level accuracy on well-defined fields while still requiring human oversight for nuanced or sparsely reported details. The benchmark and open-source workflow set a solid baseline for future advances in prompt engineering, model specialization, and ensemble strategies aimed at fully automated evidence synthesis in neurostimulation research involving aging populations. Overall, the five-model multi-LLM ensemble doubled the number of eligible aging-related tDCS trials retrieved versus keyword searching and achieved near-perfect agreement on core stimulation parameters (κ ≈ 0.94), demonstrating expert-level extraction accuracy.

Various factors may alter the risk for cardiovascular disease in adults with Down syndrome (Ds), yet few studies have examined differences in cardiac physiology in this population. Previous research suggested lower systolic and diastolic function, but inconsistent methodologies and younger samples warrant research in adults with Ds. Our aim is to compare the cardiac structure and function of adults with Ds to age- and sex-matched adults without Ds. Echocardiography was used to assess systolic function, diastolic function, and cardiac structure in n = 19 adults (Ds n = 9, control n = 10). Regarding cardiac structure, adults with Ds had increased left ventricular posterior wall thickness at end-systole compared to adults without Ds (p = 0.007). Regarding systolic and diastolic function, adults with Ds were found to have lower septal peak systolic annular velocity (S’) (p = 0.026), lower lateral and septal mitral annular early diastolic velocity (E’) (p = 0.007 and p = 0.025, respectively), lower lateral peak mitral annular late diastolic velocity (A’) (p = 0.027), and higher lateral and septal mitral annular early systolic velocity to diastolic velocity ratios (E/e’) (p = 0.001 and p = 0.001, respectively). Differences in both cardiac structure and function were found when comparing adults with Ds to matched adults without Ds. Most of the differences were indicative of worse diastolic function.

This study reveals, for the first time, that adults with Down syndrome exhibit significantly lower ventricular-arterial coupling ratios, reflecting distinct cardiac timing and functional adaptations. These findings could indicate a compensatory mechanism to maintain cardiac output despite reduced diastolic pressure and smaller cardiac chambers, providing novel insight into the integrated cardiovascular physiology of adults with Down syndrome and its potential implications for exercise tolerance and long-term cardiovascular health.