Cognitive assessment in older adults with low educational attainment presents a diagnostic limitation. Conventional instruments may show reduced accuracy in populations with limited literacy. Therefore, we aimed to compare the diagnostic performance of four cognitive screening instruments for detecting mild cognitive impairment (MCI) in older adults with low education. In this prospective study, adults aged ≥65 years with educational attainment at or below the primary school level were recruited from a geriatric outpatient clinic. Participants underwent assessment with the RUDAS, QMCI-TR, S-MMSE, and DemTect. MCI was diagnosed according to Petersen criteria by geriatricians blinded to cognitive screening test results. Receiver operating characteristic analysis and multivariable logistic regression were performed. A total of 242 participants (mean age 74.0 ± 5.8 years; 62.0% female) were included, of whom 100 (41.3%) had MCI. RUDAS demonstrated comparatively better diagnostic accuracy (AUC 0.740, 95% CI 0.666-0.798), with a cut-off value of ≤23 yielding 83.0% sensitivity and 73.2% specificity. QMCI-TR and S-MMSE showed lower discriminative performance (AUCs 0.666 and 0.631, respectively), while DemTect showed no significant discriminative value (AUC 0.464). RUDAS remained the strongest independent predictor of MCI in multivariable analysis. Cognitive screening performance varied substantially across instruments in older adults with low educational backgrounds. While RUDAS showed the highest, albeit modest, diagnostic accuracy, other tools showed more limited or no diagnostic utility. Given that MCI represents a high-risk state for progression to dementia, these findings highlight the potential for misclassification, underscoring the need for appropriate test selection to improve early detection.
This year marks the 50th anniversary of the Ebola virus identification, but the 2026 outbreak of Bundibugyo ebolavirus disease has exposed important limitations in Ebola preparedness strategies that remain largely focused on Zaire ebolavirus. Although major advances in diagnostics, vaccines, and therapeutics have followed the 2014-2016 West African Ebola epidemic, most licensed countermeasures were developed against Zaire ebolavirus and may provide limited protection against other ebolavirus species, including Bundibugyo ebolavirus. Herein, we examine the epidemiological significance of Bundibugyo ebolavirus and review current and emerging diagnostics, vaccines, antibody therapies, and antiviral strategies, with emphasis on their species coverage and limitations. We further discuss how diagnostic blind spots and limited species-inclusive countermeasures contributed to challenges during the current outbreak. Future Ebola preparedness should adopt a broader framework encompassing multiple ebolavirus species with epidemic potential.
This study investigated the diagnostic accuracy of lung ultrasound (LUS) and serum KL-6 levels for detecting interstitial lung disease (ILD) in Sjögren's disease (SjD) patients. This retrospective study included 70 SjD patients evaluated at Shantou Central Hospital. All patients underwent chest high-resolution computed tomography (HRCT), LUS, and KL-6 measurement within one month. LUS was performed at 50 scanning sites. The presence and patterns of ILD were defined by HRCT findings. Serum KL-6 levels were measured using chemiluminescent enzyme immunoassay. Correlations between B-lines score, KL-6 level, and the HRCT Warrick score were analyzed. ROC curves with DeLong test and Spearman correlation analysis were performed to evaluate diagnostic efficiency and correlations with the Warrick HRCT fibrosis score. The concordance rate between LUS and HRCT was 82.86% (Kappa value = 0.6572). The optimal cut-off value for detecting SjD-ILD was 22 B-lines (AUC = 0.952, sensitivity 82.86%, specificity 100%). Serum KL-6 levels were also significantly higher in patients with ILD compared to those without ILD on HRCT (430.8 [274.6-698.3] vs. 240.4 [168.5-298.8]). ROC analysis showed that a serum KL-6 concentration of 329 U/mL was the optimal cut-off value for detecting ILD in SjD patients (AUC = 0.814, sensitivity 65.71%, and specificity 88.57%). B-line score, KL-6 level, and Warrick score presented significant pairwise positive correlations. LUS B-lines and serum KL-6 exhibit robust diagnostic value for SjD-ILD. The cut-offs > 22 B-lines and 329 U/mL KL-6 are recommended. Key Points • The concordance rate between LUS and HRCT was 82.86%, providing a noninvasive and reliable strategy for SjD-ILD screening, with an optimal cut-off of 22 B-lines (AUC = 0.952). • B-line score and serum KL-6 both strongly correlate with HRCT Warrick score, supporting their application as noninvasive tools to assess ILD severity in SjD.
To investigate the impact of different ROI delineation strategies on the utility of Time-dependent diffusion MRI (TDD-MRI)-derived microstructural parameters for distinguishing adenocarcinoma (AC) from squamous cell carcinoma (SCC). In this prospective study, patients with pathologically confirmed cervical cancer who underwent TDD-MRI between August 2024 and June 2025 were enrolled. Three region-of-interest (ROI) delineation strategies were used: small solid ROI (ROIs), single-slice ROI (ROIss), and whole-volume ROI (ROIwt). Microstructural parameters including intracellular volume fraction (fin), extracellular diffusion coefficient (Dex), diameter, and cellularity, along with three apparent diffusion coefficient (ADC) measures were investigated. The intraclass correlation coefficient (ICC) was used to determine inter- and intra-readers reproducibility. Logistic regression was performed to predict pathological subtypes. Diagnostic performance was quantified by area under the receiver operating characteristic curve (AUC). Pearson's correlation analysis validated the relationship between TDD-MRI parameters and pathological measurements. A total of 92 women (79 with SCC and 13 with AC) with cervical cancer (mean age, 55.5 ± 10.4 years) were included. For TDD-MRI-derived microstructural parameters and ADCs, intra- and inter-reader ICCs were 0.910-0.971 and 0.904-0.981, respectively. The ROIss strategy showed significant differences between AC and SCC in four parameters (all P < 0.05); ROIwt showed significance only for cellularity, and ROIs showed none. The ROIss-derived parameters achieved relatively favorable performance for distinguishing histologic subtypes (AUC = 0.821). The combined Dex, cellularity, and ADC40Hz model based on the ROIss approach achieved an AUC of 0.917. TDD-MRI-derived parameters showed correlations with histopathologic measurements (n = 15; r = 0.698-0.794; P < 0.01). TDD-MRI-based microstructural parameters derived from the ROIss show promise as effective imaging biomarkers to assist in differentiating histologic subtypes in cervical cancer.
To systematically evaluate the diagnostic accuracy and methodological quality of machine learning (ML) prediction models for pregnancy outcomes after assisted reproductive technology (ART). PubMed, Embase, the Cochrane Library, IEEE Xplore, MEDLINE, ClinicalTrials.gov, CNKI, Wanfang, and VIP were searched from inception to July 2026. Eligible studies developed or validated ML models to predict clinical pregnancy or live birth after ART. For studies reporting complete 2 × 2 contingency data, pooled sensitivity, specificity, diagnostic odds ratio (DOR), and summary receiver operating characteristic (SROC) curves were estimated using random-effects diagnostic meta-analysis. Risk of bias was assessed with PROBAST. Twenty studies were included in the systematic review, of which 14 contributed to the diagnostic meta-analysis. Overall risk of bias was low in 1 study (5.0%), high in 8 studies (40.0%), and unclear in 11 studies (55.0%). The pooled sensitivity was 0.737 (95% CI, 0.662-0.799) and the pooled specificity was 0.789 (95% CI, 0.709-0.851), with substantial heterogeneity (I2 = 97.7% and 99.0%, respectively). The pooled DOR was 10.49 (95% CI, 6.28-17.53), and the SROC curve indicated acceptable overall discrimination. Exploratory DOR subgroup analyses showed comparable performance for clinical pregnancy and live birth. No statistically robust subgroup difference was observed by algorithm type, center type, or validation status under a random-effects framework; study design showed a significant subgroup difference, but this estimate was driven by a single prospective study. ML models show moderate diagnostic accuracy for predicting ART pregnancy outcomes, but the evidence base is limited by substantial heterogeneity and frequent high or unclear risk of bias. Future studies should follow TRIPOD + AI and PROBAST-aligned standards, report calibration and clinical utility, and prioritize prospective multi-center external validation before clinical implementation. PROSPERO, CRD420251108846.
Breast cancer is a biologically heterogeneous disease, and reliable non-invasive biomarkers are needed to improve early detection and clinical risk stratification. MicroRNAs (miRNAs) represent promising diagnostic candidates due to their regulatory roles in tumor-associated gene expression. In this study, an integrative analysis of miRNA and mRNA expression profiles was performed using paired tumor and normal samples from the TCGA-BRCA cohort to identify diagnostically relevant miRNAs. Differential expression analysis and receiver operating characteristic (ROC) curve analysis were used to identify miRNAs with high discriminatory capacity. Priority was given to candidates showing strong inverse correlation with predicted target genes. A logistic regression model was developed for internal validation using a held-out test set, and exploratory clinical serum analysis was conducted using qRT-PCR on serum samples from breast cancer patients and healthy controls. Multiple upregulated miRNAs with strong discriminatory capacity (AUC values > 0.90) were identified. Among these, hsa-miR-200a was prioritized based on consistently high diagnostic accuracy and a strong inverse correlation with its predicted target gene, TNS1. hsa-miR-200a expression was elevated across pathological stages and molecular subtypes, whereas TNS1 expression was progressively reduced. The logistic regression model achieved an accuracy of 92.9% and a sensitivity of 96.7% in a held-out internal test set. Exploratory serum analysis suggested altered circulating hsa-miR-200a-5p patterns, with distinct perioperative dynamics according to clinical risk. These findings support hsa-miR-200a as a potential breast cancer-associated miRNA biomarker candidate and highlight the utility of integrative transcriptomic and exploratory clinical serum approaches for future precision diagnostics.
We evaluated the clinical significance of plasma autoantibodies against cancer stem cell-associated antigens in breast cancer, with diagnostic performance as the primary endpoint and clinicopathological and prognostic relevance as secondary endpoints. In this single-center observational study, autoantibodies against SOX2, survivin, and DNAJB8 were measured using ELISA. The diagnostic performance of anti-DNAJB8 autoantibodies was evaluated in 292 patients with breast cancer and 24 benign controls. Clinicopathological and postoperative analyses were exploratory in a preoperative cohort (175 patients). Only DNAJB8 showed a distinct high-titer subset in the breast cancer group. The primary endpoint was not met, as plasma anti-DNAJB8 autoantibodies showed a non-significant diagnostic performance (AUC, 0.602; 95% CI, 0.488-0.715; P = 0.080). In the preoperative cohort, with six deaths and 11 recurrences, anti-DNAJB8 autoantibody positivity was associated with a worse overall and recurrence-free survival in age-adjusted analyses using Firth's penalized likelihood Cox regression (overall survival: hazard ratio, 10.607; 95% CI, 2.280-61.840; P = 0.003; recurrence-free survival: hazard ratio, 4.329; 95% CI, 1.247-13.082; P = 0.023). The primary diagnostic endpoint was not achieved for plasma anti-DNAJB8 autoantibodies. However, anti-DNAJB8 autoantibody positivity was associated with poorer postoperative outcomes, suggesting a hypothesis-generating prognostic relevance.
The clinical utility of 18F-fluoroglutamine (18F-FGln) PET/CT for characterizing metabolic heterogeneity and improving lung cancer staging remains underexplored. In this prospective study, 31 patients with 36 primary lung lesions underwent dual-tracer (18F-FDG/18F-FGln) PET/CT. A metastatic cohort (n = 28) was analyzed for distant staging. Diagnostic performance was assessed using ROC analysis, logistic regression, and radiomic texture parameters (volume, mass, CT histogram features). For primary lesions, 18F-FDG showed marginally higher detection rates (86.1% vs. 80.6%) and significantly greater avidity (SUVmax: 9.15 ± 0.16 (FDG) vs. 3.94 ± 1.57 (FGln), P < 0.001). CT texture analysis revealed kurtosis as an independent predictor of 18F-FGln uptake (OR = 1.16, P = 0.025), correlating with metabolic-structural coupling (r = 0.445, P = 0.009). In nodal staging, 18F-FGln identified improved diagnostic performance than 18F-FDG (AUC: 0.92 vs. 0.62). Metastatic lymph nodes showed higher 18F-FGln uptake, with increased SUVmax (3.57 ± 1.23 vs. 2.03 ± 0.47, P < 0.001) and TBR (2.11 ± 0.92 vs. 0.93 ± 0.21, P < 0.001). Multivariate analysis identified 18F-FGln SUVmax (OR = 32.79, P < 0.001) and CT density (OR = 0.90, P < 0.001) were predictors of metastatic lymph nodes (LNs). For distant metastases, 18F-FGln detected more distant lesions (86 vs. 70), particularly in bone (SUVmax 7.48 ± 3.03 vs. 6.80 ± 4.48) and brain (TBR 7.04 ± 2.96 vs. 0.88 ± 0.30, P < 0.001), altering staging in 3 cases. 18F-FGln PET/CT showed promising clinical potential in lung cancer, particularly in nodal staging, with higher diagnostic performance compared with 18F-FDG. It also demonstrated improved detection of bone and cerebral metastases in selected patients. Furthermore, 18F-FGln uptake correlated with CT-derived texture features, especially kurtosis, suggesting a possible association with tumor heterogeneity. Despite several false-negative cases, these results indicate that 18F-FGln may serve as a complementary metabolic imaging biomarker in lung cancer. Multicenter validation studies are needed to confirm these findings. ChiCTR2000037834 Retrospectively Reg Date：2020-09-02.
Retinal vasculitis is a sight-threatening condition associated with diverse ocular and systemic diseases. Variability in terminology and definitions across clinical practice and research has limited diagnostic consistency, communication among specialists, and comparability of studies. To develop standardized, consensus-based definitions for retinal vasculitis and related terms to improve diagnostic clarity and harmonize communication among ophthalmologists and other medical specialties. This was an international modified Delphi consensus study conducted using a structured consensus process guided by a comprehensive literature review, including systematic reviews and meta-analyses. Two rounds of Delphi surveys were performed, and consensus was predefined as at least 75% agreement. Included was an international expert panel of 27 specialists in ophthalmology, rheumatology, pathology, imaging, and research methodology. The specialists had recognized expertise in retinal and systemic vasculitis, retinal imaging, or consensus methodology. Study data were analyzed from March to September 2025. Participation in a structured Delphi process evaluating proposed definitions and terminology related to retinal vasculitis and associated vascular inflammatory entities. Consensus definitions for retinal vasculitis and related entities, with agreement using predefined consensus thresholds. A total of 27 international experts participated in the Delphi process. After 5 executive committee members involved in developing preliminary definitions were excluded from voting to minimize bias, a total of 22 independent experts completed both Delphi rounds. All candidate statements exceeded the predefined consensus threshold in the first round (agreement range, 83.3%-95.5%), although some demonstrated variability in the strength of agreement. After refinement of definitions, consensus strengthened in the second round, with agreement levels ranging from 95.2% to 100%. The panel established consensus definitions for 8 terms: retinal vasculitis, vascular leakage, infectious retinal vasculitis, noninfectious retinal vasculitis, retinal perivasculitis, retinal vasculopathy, primary retinal vasculitis, and far-peripheral vascular leakage. These definitions were developed to distinguish vascular leakage from true vasculitis, clarify inflammatory and noninflammatory vascular disorders, and promote consistent terminology across clinical practice and research. This international Delphi consensus established standardized nomenclature for retinal vasculitis and related vascular inflammatory entities. Adoption of these definitions may improve diagnostic accuracy, facilitate interpretation of imaging findings, enhance consistency across clinical studies, and support future research efforts, including the development of artificial intelligence-based classification systems for retinal vascular disease.
Monochromator phototesting is a specialist investigation to assess abnormal skin response to defined ultraviolet and visible wavebands in patients with suspected photodermatoses. Monochromator phototesting spans ultraviolet B (UVB), ultraviolet A (UVA) and visible light (VL) and is used in specialist UK photodiagnostic centres. Abnormal-response yield varies between individuals and diagnostic groups. However, the procedure is time- and resource-intensive and limited to specialist centres. To identify the wavebands and waveband combinations that captured the greatest proportion of abnormal monochromator responses, and to explore whether these findings could inform future evaluation of limited-waveband approaches within specialist phototesting pathways. This retrospective single-centre analysis included 668 phototesting results from 552 individuals collected between 2020 and 2025 at the Scottish Photobiology Service, NHS Tayside, Dundee, UK. Records prior to 2020 were reviewed in some selected patients to distinguish persistently negative monochromator tests from previously documented abnormal responses that had resolved before the study period. Assessed wavebands were 305 ± 5 nm (UVB), 335 ± 27 nm (UVB + UVA), 365 ± 27 nm (UVA), 400 ± 27 nm (UVA + VL) and 430 ± 27 nm (VL). Of 552 individuals, 353 received a final clinical diagnosis of photodermatosis and 199 did not. Across individuals with photodermatoses, 335 nm showed the highest single-wavelength abnormal-response yield (203/353, 57.5%), followed by 365 nm (194/353, 55.0%), 305 nm (169/353, 47.9%), 400 nm (108/353, 30.6%) and 430 nm (46/353, 13.0%). Selected two- and three-waveband combinations increased abnormal-response yield, with 305 + 365 nm showing the highest two-waveband yield (239/353, 67.7%) and 305 + 365+400 nm showing the highest selected three-waveband yield (259/353, 73.4%). In complete-case paired analysis, abnormal-response rates differed significantly across wavelengths (Cochran's Q = 200.4, df = 4, p < 0.001). Exploratory subgroup analyses showed distinct wavelength-response profiles across CAD, PLE and SU: CAD showed high yields at 305 and 335 nm, PLE showed lower monochromator abnormal-response yields across selected wavelengths and combinations, and SU showed greater longer-wavelength involvement. In this exploratory single-centre study, around one quarter of individuals with a final photodermatosis diagnosis in our service did not demonstrate abnormal responses on narrow-waveband monochromator phototesting. Limited waveband monochromator combinations showed differing abnormal response yields across photodermatoses, capturing most abnormal responses in chronic actinic dermatitis and many in solar urticaria, but under-detecting polymorphic light eruption. These findings support waveband phototesting across the ultraviolet and visible spectrum in specialist phototesting services. Broadband and provocation-based photodiagnostic approaches remain important, particularly for conditions such as polymorphic light eruption.
To commemorate the centennial of the Japanese Orthopaedic Association, this review highlights Japan's pioneering and globally impactful contributions to spinal surgery, focusing specifically on degenerative cervical myelopathy (DCM) as a representative field of clinical excellence born from unique demographic needs. A comprehensive historical and clinical overview was conducted, tracing the evolutionary path of diagnostic paradigms and posterior decompressive techniques developed by Japanese innovators, driven by a high domestic prevalence of developmental canal stenosis and ossification of the posterior longitudinal ligament (OPLL). In the pre-MRI era, clinical necessity fostered a rigorous neurological culture, yielding diagnostic landmarks such as the Hattori classification, the ten second test, and precise level diagnosis criteria. To circumvent the catastrophic complications of conventional laminectomy, Japanese spine surgeons revolutionized global practice by inventing laminoplasty, evolving from Hattori's Z-plasty to Hirabayashi's open-door and Kurokawa's double-door techniques. Recent decades have seen these procedures meticulously refined to preserve cervical alignment and musculature, successfully adapting to an unprecedented super-aged population. As epitomized by the historical breakthroughs in DCM, Japanese spinal innovations have long established universal standards in patient care. At the forefront of a rapidly aging global population, Japanese orthopaedic surgeons remain uniquely positioned to spearhead the next frontier of clinical innovation across the broader field of spinal care.
To compare the diagnostic performance of microvascular ultrasonography-derived vascularity index (VI) and median nerve cross-sectional area (CSA) measured at multiple anatomical levels in relation to electroneuromyography (ENMG)-based severity of carpal tunnel syndrome (CTS). In this prospective study, 101 hands from 67 patients with ENMG-confirmed CTS (mild, moderate, severe) were evaluated. CSA was measured at four anatomical levels, and VI was obtained at three levels using a standardized microvascular ultrasonography protocol. Group differences were analyzed with non-parametric tests. Receiver operating characteristic (ROC) analyses were performed to assess diagnostic performance. CSA differed significantly across severity groups only at the pre-carpal tunnel level (P = 0.009). VI showed significant overall differences at the pre-carpal, proximal tunnel, and distal tunnel levels (P < 0.001, P < 0.001, and P = 0.001, respectively). For distinguishing moderate-to-severe CTS from mild CTS, pre-carpal and proximal tunnel VI measurements showed comparable discriminatory performance, with an AUC of 0.75 at both levels. For identifying severe CTS, pre-carpal tunnel VI showed an AUC of 0.76. VI derived from microvascular ultrasonography demonstrates a more consistent and discriminative association with CTS severity than CSA. These findings support its potential as a quantitative, non-invasive imaging biomarker for CTS severity stratification in clinical practice.
Decapod iridovirus 1 (DIV1) is a highly lethal pathogen that infects decapod crustaceans including Litopenaeus vannamei, causing mass mortality in cultured shrimp and severe economic losses worldwide. The ATPase gene is a highly conserved region within the DIV1 genome, plays a critical role in viral replication and represents an ideal target for molecular diagnostic development. In this study, we established a rapid, sensitive and field-adaptable detection platform for DIV1 by integrating recombinase polymerase amplification (RPA) with the CRISPR/Cas12a system. RPA enables efficient isothermal amplification of target nucleic acids, achieving exponential enrichment of the target nucleic acids and exerting the function of signal amplification. While the CRISPR/Cas12a system upon crRNA-guided specific recognition of the amplicon, triggers robust trans-cleavage activity against reporter probes for signal generation and readout. After systematic optimization, the RPA reaction was performed at 38°C for 10 min and the CRISPR-Cas12a reaction was conducted at 37°C for 20 min. The integrated two-step workflow completed detection within 40 min, with a limit of detection of 2.3 × 101 copies/μL. Specificity evaluation confirmed that the RPA-CRISPR/Cas12a assay exclusively recognised DIV1 without cross-reaction with other major shrimp pathogens. Further validation using clinical shrimp samples demonstrated stable and reliable performance, supporting its practical utility in aquaculture settings. In conclusion, the established CRISPR/Cas12a-based detection platform provides a robust technical tool for early warning and on-site rapid screening of DIV1, facilitating timely disease control and risk management in shrimp farming.
Central nervous system solitary fibrous tumors (CNS SFTs) are rare mesenchymal neoplasms. The 2021 World Health Organization (WHO) classification recognizes a single SFT entity characterized by NAB2::STAT6-associated biology. This study evaluated clinicopathologic and immunohistochemical features, with particular emphasis on STAT6, CD34, and p16 expression. We retrospectively reviewed 25 CNS SFTs identified between 2004 and 2024. Clinical, radiologic, histologic, immunohistochemical, treatment, and outcome data were collected. Histologic slides were reviewed by two neuropathologists. Immunohistochemistry for STAT6, CD34, and p16 was performed; molecular testing was unavailable. The cohort included 13 men and 12 women, with a mean age of 51 years. Among tumors with available site data, most were intracranial. WHO grades were grade 1 in 40%, grade 2 in 24%, and grade 3 in 36%. Nuclear STAT6 expression was present in all cases. CD34 expression was greatest in grade 1 tumors, whereas p16 expression was numerically highest in grade 3 tumors; neither marker showed a statistically significant association with recurrence. Tumor size increased across WHO grades. Outcome analyses were limited by few events and heterogeneous follow-up. CNS SFTs are clinicopathologically heterogeneous. STAT6 was a consistent diagnostic marker in this cohort. CD34 and p16 showed grade-related numerical patterns, but their prognostic value was not established. Larger, molecularly confirmed cohorts with standardized long-term follow-up are required.
Saliva-based point-of-care testing (POCT) is critical for the early diagnosis of periodontal disease. However, current wearable platforms, such as smart mouthguards, are predominantly limited to monitoring small metabolites (e.g., glucose) and cannot detect specific macromolecular proteins essential for disease characterization. To enable precise early diagnosis, we demonstrate a novel biosensor that integrates microfluidics with organic electrochemical transistors (OECTs). Functionally, this integration enables the simultaneous, multiplexed detection of a complementary biomarker combination: interleukin-6 (IL-6) and matrix metalloproteinase-8 (MMP-8). In terms of performance, the device leverages the high transconductance of OECTs to ensure high precision even at trace levels, achieving exceptional sensitivity with distinct dynamic ranges tailored for early-stage detection (IL-6: 1-80 pg/mL; MMP-8: 10-500 ng/mL). Validating its diagnostic utility, the sensor showed precise quantification (R2 > 0.97) correlated with gold-standard laboratory measurements in a rat model. By delivering clinical-grade precision for complex biomacromolecules, this platform overcomes the limitations of existing wearables and offers a viable path for translating advanced bioelectronics into practical periodontal healthcare.
Depression is prevalent yet frequently underdiagnosed. Although speech-based detection methods show promise, most prior studies have emphasized binary classification in nontonal languages. This study examined whether acoustic features of spontaneous Mandarin speech reflect depression as a categorical condition or a continuous, dimensional construct. A validated emotional Mandarin speech corpus paired with self-reported depression severity scores was analyzed. Acoustic features were extracted using the extended Geneva Minimalistic Acoustic Parameter Set. A multistage analytic framework was applied, including random forest classification to distinguish depressed from nondepressed clips, linear mixed-effects modeling to examine group and severity effects, and unsupervised clustering to examine latent structure in the acoustic feature space. The random forest classifier achieved an accuracy of 72.3% and an area under the receiver operating characteristic curve of 0.826 on the held-out test set. Model performance favored sensitivity over precision in identifying depressed clips. SHapley Additive exPlanations analysis identified features across multiple acoustic domains, including spectral, pitch-related, voice quality, cepstral, and intensity measures, as important contributors to model predictions. Among the Top 10 features, no acoustic features showed statistically significant group differences. In contrast, median and mean fundamental frequencies were significantly positively associated with depression severity. Clustering analysis revealed overlapping group structures and a continuous distribution of severity scores across clusters. These findings provide preliminary support for a dimensional conceptualization of depression in spontaneous Mandarin speech. While classification models can distinguish depressed from nondepressed speech with moderate accuracy, acoustic features appear to vary more consistently with symptom severity than with diagnostic group. Speech-based measures may therefore have potential for continuous monitoring of depression, although this possibility requires further longitudinal verification.
Heart rate variability (HRV) has remained a relatively finite and niche tool in cardiology despite decades of research supporting its physiological and clinical relevance. This limited adoption may resemble the early history of electrocardiography (ECG), which was initially regarded by many physicians as a laboratory instrument rather than a routine clinical tool. The delayed acceptance of ECG reflected technological limitations, cultural resistance and the need for clinicians to master unfamiliar concepts derived from physics and electrophysiology. HRV faces comparable barriers today. Although derived from ECG RR intervals, HRV requires interpretation of time-domain, frequency-domain, geometric and nonlinear indices that may appear mathematically complex and distant from conventional bedside reasoning. We argue that HRV should not be viewed as a replacement for ECG, but as an extension of ECG from electrical morphology to physiological dynamics. Lessons from ECG history were examined and compared with the current state of HRV adoption in clinical practice. The complementary diagnostic roles of ECG morphology and HRV analysis were considered, together with the potential contribution of wearable sensors, remote monitoring, artificial intelligence and large language models to facilitate HRV interpretation, education and clinical integration. Whereas conventional ECG morphology identifies arrhythmias, conduction disturbances, ischemic alterations and overt electrical abnormalities, HRV provides insight into autonomic modulation, cardiovascular adaptability and systemic physiological regulation. The emergence of wearable sensors, remote monitoring, artificial intelligence and large language models creates an opportunity to overcome barriers that have limited HRV adoption. Artificial intelligence may serve as an educational and interpretive bridge, translating complex HRV metrics into clinically meaningful concepts while supporting medical training, artefact awareness, case-based learning and workflow integration. HRV faces barriers comparable to those encountered during the early adoption of ECG, including technological limitations, educational challenges and resistance to incorporating unfamiliar physiological concepts into routine clinical practice. Lessons from ECG history suggest that HRV adoption will depend not only on evidence but also on standardization, education, clinical interpretation and cultural acceptance within cardiology.
Clinical decision-making in glaucoma is complex and requires integration of heterogeneous information, including patient history, examination findings, and risk stratification. While artificial intelligence (AI) has shown strong performance in image-based ophthalmic tasks, its capability in specialty-specific clinical reasoning remains insufficiently explored. Performance was evaluated by glaucoma specialists using a predefined rubric across three clinically oriented domains: medical accuracy (40%), key-point recall (30%), and logical completeness (30%). The weighted composite score was used as a descriptive summary of case-based reasoning quality. AI models showed structured clinical reasoning performance in this case-based dataset, with weighted mean scores overlapping with those of attending ophthalmologists and exceeding those of some lower-performing trainees. These findings should be interpreted as exploratory performance patterns rather than evidence of equivalence. Inter-individual variability was substantial among human clinicians, particularly residents. AI systems often included safety-critical diagnostic and management elements, while the best-performing human clinician achieved the highest individual score overall. In this limited 34-case evaluation, large language model-based AI systems produced structured glaucoma-related reasoning with performance that overlapped with attending ophthalmologists but did not establish clinical equivalence. These systems require specialist oversight and further validation before clinical use, but may have potential as supervised decision-support and educational tools.
This essay discusses the public discourse around “missed” diagnoses and how physicians should engage patients early in diagnostic reasoning.
Dedifferentiated liposarcoma (DDLPS) typically presents as a solitary retroperitoneal mass; diffuse granular and nodular peritoneal sarcomatosis as the inaugural manifestation is exceptional. We report such a case to highlight a critical diagnostic consideration: distinguishing primary sarcomatous peritoneal dissemination from the more common epithelial peritoneal carcinomatosis. A 72-year-old Chinese man with no prior abdominal surgery presented with incidentally discovered multifocal intra-abdominal masses. CT demonstrated multiple peritoneal and retroperitoneal masses with encasement of the jejunum and transmural infiltration of the descending colon, and a retroperitoneal plaque abutting the left kidney. CT-guided biopsy confirmed DDLPS via MDM2 and CDK4 amplification on fluorescence in situ hybridization. Multidisciplinary team consensus directed surgical exploration for impending dual-site bowel obstruction. Laparotomy revealed diffuse granular and nodular peritoneal deposits (Peritoneal Cancer Index 28/39); cytoreductive surgery achieved a completeness of cytoreduction score of 2. Histopathology confirmed FNCLCC Grade III DDLPS. Postoperative next-generation sequencing demonstrated high-level co-amplification of CDK4, MDM2, TSPAN31, CCND2, MDM4, and RAC1, with microsatellite stability and tumor mutational burden of 0 mutations/Mb. DDLPS can present with primary diffuse peritoneal sarcomatosis even with a retroperitoneal component. When diffuse peritoneal implants yield an epithelial-marker-negative spindle cell neoplasm on biopsy, sarcoma must be considered in the differential diagnosis and FISH for MDM2/CDK4 should be performed to avoid misdiagnosis as peritoneal carcinomatosis from gastric, colorectal, or ovarian cancer.