Background: The efficacy of large language models (LLMs) in domain-specific medicine, particularly for managing complex diseases such as osteoarthritis (OA), remains largely unexplored. Objective: This study focused on evaluating and enhancing the clinical capabilities and explainability of LLMs in specific domains, using OA management as a case study. Methods: A domain-specific benchmark framework was developed to evaluate LLMs across a spectrum from domain-specific knowledge to clinical applications in real-world clinical scenarios. DocOA, a specialized LLM designed for OA management integrating retrieval-augmented generation and instructional prompts, was developed. It can identify the clinical evidence upon which its answers are based through retrieval-augmented generation, thereby demonstrating the explainability of those answers. The study compared the performance of GPT-3.5, GPT-4, and a specialized assistant, DocOA, using objective and human evaluations. Results: Results showed that general LLMs such as GPT-3.5 and GPT-4 were less effective in the specialized domain of OA management, particularly in providing personalized treatment recommendations. However, DocOA showed significant improvements. Conclusions: This study introduces a novel benchmark framework that assesses the domain-specific abilities of LLMs in multiple aspects, highlights the limitations of generalized LLMs in clinical contexts, and demonstrates the potential of tailored approaches for developing domain-specific medical LLMs.

Discoid lateral meniscus (DLM) is relatively common in East Asia..Symptomatic discoid lateral meniscus (SDLM) is an important indication for knee arthroscopic surgery. However, studies investigating SDLM are rare. The purpose of this study was to evaluate the clinical characteristics and intra-articular variants of SDLM in a Chinese population. We retrospectively reviewed all patients with SDLM from January 2005 to December 2014 in our hospital. Clinical variables included gender, age, duration, age of onset, affected side, symptoms and trauma history as well as arthroscopic findings: DLM types, tear patterns and concomitant medial meniscus tear, which were evaluated and compared statistically. Of the 496 consecutive participants with SDLM, females outnumbered males (69.6 % vs. 30.4 %). The age of onset ranged from 3 to 80 years (median, 31 years), and was significantly higher in females than in males (p < 0.0001). Trauma history in males was significantly higher than in females (45 % vs. 35.1 %, p = 0.0356). Males showed a higher incidence of popping and snapping, while females manifested higher range-of-motion (ROM) limitations (p = 0.0179, and p = 0.0392, respectively). No significant difference in intra-articular variants was observed between genders. The complete type was the most frequent (344, 69.4 %), followed by the incomplete type (149, 30 %), and the Wrisberg type (3, 0.6 %). Significant difference in tear patterns was found between complete and incomplete types (p < 0.0001). Few patients showed medial meniscus tear (11, 2.2 %), at a significantly higher age compared with patients without tear (median, 57 years vs. 33 years, p < 0.0001). The majority of Chinese patients with SDLM are young and middle-aged females. Female patients had an older age of onset, higher incidence of ROM limitation and limited trauma history. The complete type is the most common, with tear patterns varying between complete and incomplete types. The SDLM does not significantly affect the medial meniscus.

Discoid lateral meniscus (DLM) is the most common congenital variant of the lateral meniscus, which is prone to degeneration and lesions, and often leads to knee osteoarthritis. At present, there is no consensus on the clinical practice of DLM, and this expert consensus and practice guidelines on DLM was developed and approved by Chinese Society of Sports Medicine according to the Delphi method. Among 32 statements drafted, 14 statements were excluded for redundant information, and 18 statements achieved consensus. This expert consensus focused on the definition, epidemiology, etiology, classification, clinical manifestations, diagnosis, treatment, prognosis, and rehabilitation of DLM. Restoring the normal shape, retaining appropriate width and thickness, and ensuring the stability of the remnant meniscus is critical to sustaining the physiological function of the meniscus and preserving the knee. The partial meniscectomy with or without repair should be the first‐line treatment when possible, given that the clinical and radiological long‐term outcomes of total or subtotal meniscectomy are worse.

Background: So far, the diagnosis of acute artery of percheron (AOP) infarction is uncommon. In this study, patients with acute AOP infarction were studied to explore the relationship of imaging findings, clinical manifestations and prognosis of acute AOP infarction. Materials: A total of 23 patients with acute AOP infarction in our institution from 2014 to 2019 were reviewed retrospectively. All cases were evaluated by computed tomography (CT) and magnetic resonance imaging (MRI). The modified Rankin scale (MRS), blood examination, electrocardiogram and transthoracic echocardiography were used for detailed clinical and prognostic evaluation. All standard risk factors for these patients were recorded. The MRS scores were performed 90 days after discharge. Results: Four different types of acute AOP infarction were identified: (a) bilateral paramedian thalamic infarction (BPTI, 52%); (b) bilateral paramedian thalamic with rostral midbrain infarction (BPTRMI, 30%), (c) bilateral paramedian and anterior thalamic infarction (BPATI, 13%), and (d) bilateral paramedian thalamic with red nuclei infarction (BPTRNI, 4%). These patients had consciousness disorder, memory dysfunctions, vertical gaze paresis and mesencephalothalamic syndrome. The 65% of patients with BPTI and BPATI experienced relatively good functional recovery and could carry out daily life activities (MRS score ≤ 2). However, patients with BPTRMI may have an unfavorable outcome. Conclusions: Although the clinical features are variable, DWI or ADC map can improve the diagnosis of acute AOP infarction patterns. Acute AOP occlusion requires immediate diagnosis and treatment to obtain more favorable outcome and avoid additional unnecessary procedures.

Purpose: To explore the feasibility and validity of ultrasound in the diagnosis of discoid lateral meniscus (DLM) by comparing quantitatively the morphologic difference between DLM and normal lateral meniscus. Methods: This study was designed to develop and validate the ultrasound diagnostic criterion for DLM. In the development stage (July 2018 to June 2019), data from 180 subjects were used to derive the ultrasound diagnostic criterion, including 90 patients diagnosed as DLM by magnetic resonance imaging (DLM group) and 90 matched controls diagnosed as normal lateral meniscus diagnosed by magnetic resonance imaging (control group). Twelve distinct parameters of meniscus thickness, width, 0.5∗thickness/width, and angle were obtained through anterior, lateral, and posterior views with the probe oriented perpendicular to the lateral tibiofemoral joint line. In the validation stage (July 2019 to December 2019), data from 324 additional participants were used to validate the criterion derived from the development stage. Differences of the continuous variables and categorical variables between the 2 groups were analyzed by an independent t test and χ ² test, respectively. The diagnostic value of parameters was analyzed by the receiver operating characteristic curve. Results: In the development stage, significant differences were found in the above 12 parameters between the 2 groups ( P < .05 for all). Cut‐off values of anterior meniscus angle, meniscus body angle, and posterior meniscus angle were 28.45°, 27.85°, and 29.15°, respectively. The area under the curve, sensitivity, and specificity of anterior meniscus angle (0.953, 95.6%, 91.1%), meniscus body angle (0.980, 95.6%, 95.6%), and posterior meniscus angle (0.942, 80.0%, 97.8%) were greater than other parameters. In the validation stage, the sensitivity, specificity, and accuracy of anterior meniscus angle, meniscus body angle, and posterior meniscus angle in diagnosing DLM were as high as 91.3%, 88.6%, and 89.2%; 94.2%, 93.3%, and 93.5%; and 76.8%, 95.7%, and 91.7%. Conclusions: DLM can be reliably diagnosed by ultrasound measurements of anterior meniscus angle, meniscal body angle, and posterior meniscus angle. Level of Evidence: Level III, case‐control study.