Citations

Artificial intelligence (AI) and machine learning have emerged as transformative analytic methods capable of identifying patterns and proposing treatment strategies from complex, heterogeneous data generated in patient care settings. However, translating reports of AI tools into clinical decision-making requires careful interpretation and contextualization. Here, we present a structured framework (ABCDEFG framework) for clinicians to critically appraise studies reporting AI tools using a high-profile reinforcement learning analysis of vasopressin initiation in septic shock as a case example. We highlight key methodological strengths including adherence to reporting standards, alignment with clinical outcomes, and external validation. We also identify limitations related to causal inference, reward structure design, assumptions about state representation, and implementation readiness. Our framework emphasizes the importance of rigorous, clinically grounded critique before integrating AI tools into practice, and provides a roadmap for clinicians to assess the reliability and relevance of emerging AI-based decision support.

Abstract Introduction and Objective This study aims to evaluate the learning curve and perioperative outcomes of single‐port extraperitoneal robotic radical prostatectomy (SP‐EP‐RARP) performed by a single surgeon at a high‐volume academic institution. Methods A retrospective review of a prospectively maintained database was conducted for patients who underwent SP‐EP‐RARP from September 2023 to August 2025. Key metrics included operative time, estimated blood loss, conversion rate, nerve‐sparing status, pathological margin status and continence (0–1 pad for safety), and 30‐day complication rate. Descriptive statistics were used, and outcomes were analysed chronologically to assess for trends suggesting a learning curve. Results The cohort included 53 patients who underwent extraperitoneal SP‐EP‐RARP. Median operative time was 213 min (IQR: 145–281). Median estimated blood loss was 100 mL (IQR: 75–125). Nerve‐sparing was attempted in 75.5% of cases. No assist or plus one ports were utilized. Final pathology revealed pT2 disease in 64.1% (34 cases), pT3a in 32.1% (17 cases) and pT3b in 3.7% (2 cases). Overall positive margin rate was 26.4%. Thirty‐day Clavien–Dindo Grades I–II complication rates were 11.3%, and no major complications or 90‐day complications were reported. Continence rates at 6 weeks, 3 months and 6 months were 72%, 72% and 75%, respectively. Stabilization of operative times and intraoperative outcomes occurred after approximately 25–30 cases. Conclusion SP‐EP‐RARP can be safely implemented by a newly graduated fellowship‐trained robotic surgeon with acceptable oncologic and functional outcomes. Initial learning can be estimated at approximately 25–30 cases.

Background Suicidal behaviour shows notable sex differences, and understanding whether genetic factors contribute to these differences is critical for identifying at-risk individuals and prevention. Objective We aim to investigate the genetic contribution to suicide attempts and examine whether genetics account for sex differences in incidence. Methods This population-based cohort study includes 3.1 million individuals born 1963–1998 and followed through Swedish National Registers, including hospitals and specialist outpatient diagnoses and cause of death data. Suicide attempts were identified using ICD codes, indicating intentional self-harm, self-harm using lethal methods or leading to hospitalisation, or resulting in death. Familial aggregation, coaggregation, pedigree heritability and genetic correlations were estimated using genealogical data. For sex-specific analyses, we examined mother–daughter, female sibling, father–son and male sibling pairs, separately. Findings Suicide attempts were more common among females than males (3.3% vs 2.6%). In both sexes, risk aggregated within families (ORs ranged 1.6–3.4 across relative types) and was higher in first-degree than second-degree relatives. Familial aggregation was stronger in females than in males, and in same-sex first degree relatives compared with cross-sex pairs. Pedigree heritability was 41.9% (95% CI 36.0 to 48.4%) and did not differ significantly by sex (female 51.4% (95% CI 40.1% to 58.6%), male 45.1% (95% CI 32.3% to 52.5%), Bootstrap p value 0.40). Suicide attempt showed moderate to high pedigree genetic correlations with psychiatric disorders, strongest with substance use disorders (SUD, r g =0.85 (95% CI 0.83 to 0.96)), with no significant sex differences. The genetic correlation between female and male suicide attempts was high (0.85 (95% CI 0.80 to 0.99)), suggesting a substantial genetic overlap. Conclusions Suicide attempt has a moderate heritable component that largely overlaps between females and males and with other psychiatric disorders, particularly SUD. Stronger familial aggregation in females and in same-sex pairs highlights the potential role of sex-specific environmental or social factors. Future research should focus on non-genetic contributors and their potential interaction with genetic factors to better understand and address sex disparities in suicidal behaviour Clinical implications Genetic risk for suicide attempt is substantial but does not fully explain sex differences in incidence. Clinicians should, therefore, consider non-genetic, including sex-specific environmental and social factors, alongside family history and psychiatric comorbidity when assessing suicidal risk.

Summary Family history of haematological malignancy (FHHM) reflects inherited susceptibility to non‐Hodgkin lymphoma (NHL), but its prognostic significance remains poorly understood. We used a cohort study of 1994 NHL patients to evaluate associations between FHHM and clinical outcomes across lymphoma subtypes. We estimated associations, adjusted for sex and lymphoma‐specific prognostic index, of self‐reported FHHM in a first‐degree relative with overall survival (OS), event‐free survival (EFS) and lymphoma‐specific survival (LSS) using multivariable Cox regression models and with failure to achieve EFS at 24 months (EFS24) using logistic regression. FHHM prevalence was 12.8%, and the median follow‐up among survivors was 12.0 years. FHHM was not associated with OS. In subtype‐specific analyses, FHHM was associated with inferior LSS (hazard ratio [HR] = 1.98, 95% confidence interval [CI] 1.11–3.53) in follicular lymphoma (FL); this association and the association with EFS24 failure strengthened among those treated with front‐line immunochemotherapy (EFS24 failure odds ratio = 2.57, 95% CI 1.08–5.99; LSS HR = 2.68, 95% CI 1.09–6.58). In mantle cell lymphoma, FHHM was associated with inferior EFS (HR = 1.70, 95% CI 1.02–2.85). Interaction testing did not demonstrate significant heterogeneity of FHHM effects by lymphoma subtype. FHHM may be associated with adverse outcomes in select NHL groups, particularly FL. These findings warrant further investigation of germline and environmental factors potentially contributing to more aggressive FLs.

Abstract INTRODUCTION Determining apolipoprotein E ( APOE ) ε4 allele status, a key genetic risk factor for Alzheimer's disease (AD), requires molecular genotyping infrastructure not widely accessible beyond specialized centers. METHODS A fully automated high‐throughput apoE E4 proteotyping immunoassay was evaluated for clinical performance (460 participants across three cohorts) and analytical validity. Concordance with polymerase chain reaction (PCR)‐based genotyping and measures of analytical validity were reported. RESULTS The apoE E4 immunoassay demonstrated 99.6% (95% confidence interval [CI]: 98.4% to 99.9%) concordance with PCR‐based APOE ε4 genotype results across the pooled clinical cohort; 100.0% (95% CI: 97.1% to 100.0%) in those with AD ( N = 127) and 99.4% (95% CI: 97.8% to 99.8%) in those without AD (333). The assay met analytical validity criteria for E4 isoform specificity, interference, precision, and stability. DISCUSSION The apoE E4 immunoassay demonstrated high concordance with PCR‐based genotyping and robust analytical validity, offering an accessible alternative for APOE ε4 zygosity assessment. Highlights A novel high‐throughput plasma‐based proteotyping immunoassay for APOE ε4 zygosity classification was developed and evaluated for clinical performance and analytical validity. The apoE E4 immunoassay demonstrated high concordance (99.6%) with PCR‐based APOE ε4 genotyping across a diverse international cohort, and a robust analytical profile. An apoE E4 immunoassay may offer a more cost‐effective and accessible alternative to DNA genotyping approaches currently used for AD risk evaluation and anti‐amyloid treatment decisions.

Abstract Objectives Traditional laboratory regulatory frameworks provide robust guidance for conventional clinical testing but lack requirements for data documentation for artificial intelligence and machine learning (AI/ML) solutions. This gap creates significant risks since AI solutions directly inherit patterns, biases, and limitations from their training data. Here, we describe the development and implementation of a comprehensive data documentation checklist with accompanying data lineage for use within an AI lifecycle framework for clinical laboratories. Methods Building on the previously established Clinical AI Readiness Evaluator (CARE) framework, we developed a comprehensive data checklist and data lineage methodology through a multiphase process, including (1) comprehensive review of existing AI/ML data documentation frameworks, (2) focused meetings with 3 institutional AI operations teams, and (3) 3 rounds of iterative refinement by our multidisciplinary team. The checklist’s effectiveness was then assessed using 3 diverse AI/ML projects moving through the CARE framework. Results The CARE Data Checklist and Data Lineage provide a structured approach to documenting critical aspects of datasets used in AI/ML projects, including data composition, demographics, collection methods, labeling processes, usage constraints, maintenance requirements, and a data readiness assessment. The checklist addresses unique data-centric challenges of AI/ML applications, facilitating transparency, reproducibility, and regulatory compliance. Conclusions The CARE Data Checklist and Data Lineage serve as both a technical guide and a communication tool bridging gaps between technical and clinical stakeholders. By working on these documents early in the AI lifecycle, laboratories can anticipate and address data-related challenges, ultimately saving time, optimizing resources, and improving the reliability of AI-augmented laboratory solutions.

OBJECTIVES: To evaluate and compare the ability of three popular open-source artificial intelligence platforms to diagnose common trauma-related fractures using radiologic imaging. METHODS: Design: Retrospective diagnostic performance comparison study. Setting: Publicly accessible online radiologic imaging databases. Patient Selection Criteria: Five common orthopedic trauma fractures were assessed: ankle, tibial plateau, intertrochanteric, femoral neck, and humerus. Radiographs and computed tomography (CT) images were collected. Images were randomly selected from confirmed diagnoses on Radiopaedia.org. Outcome Measures and Comparisons: ChatGPT 5, Grok 3, and Claude 4.5 Sonnet were queried with each image. Diagnostic accuracy, sensitivity, specificity, positive and negative predictive values, and performance by modality (X-ray vs. CT) were assessed. The reference standard was the expert-verified diagnosis provided by Radiopaedia.org, limited to cases labeled with a “diagnosis certain” tag. RESULTS: Each model was provided with 30 radiographs and 20 CT images whenever possible. ChatGPT 5, Grok 3, and Claude 4.5 Sonnet accurately diagnosed diseased images in 26.8%, 18.8%, and 22.4% of cases, respectively. By fracture type, ChatGPT 5 demonstrated the highest correct classification rates for ankle (10%), femoral neck (38%), humerus (40%), and tibial plateau (44%) fractures. Grok 3 demonstrated the highest correct classification rate for intertrochanteric fractures (6%). Overall sensitivities were 0.267, 0.187, and 0.223 for ChatGPT 5, Grok 3, and Claude 4.5 Sonnet, respectively. ChatGPT 5 and Grok 3 outperformed Claude 4.5 Sonnet (both P < 0.001). No modality-based performance differences were observed for any model. CONCLUSIONS: Among the publicly available large language models evaluated for radiologic interpretation of orthopedic trauma imaging, ChatGPT 5 demonstrated the highest overall diagnostic accuracy, followed by Claude 4.5 Sonnet and Grok 3. Despite relative variation between the models, overall diagnostic accuracy for fracture detection was low across all platforms (<27%). In their baseline forms, these publicly accessible large language models are not recommended for radiologic imaging interpretation. LEVEL OF EVIDENCE: Diagnostic Level III. See Instructions for Authors for a complete description of levels of evidence.

Abstract Chromosome 5p15.33 harbors several independent association signals which demonstrate antagonistic pleiotropy across cancer types, with causal mechanisms largely unresolved. To identify functional variants and enhancer elements at this locus, we performed statistical fine-mapping followed by massively parallel reporter assays (MPRA) and proliferation based CRISPRi screens. This approach identified eight multi-cancer functional variants (MCFVs) across three GWAS signals. Targeting rs421629 (part of the CLPTM1L signal marked by rs465498) with CRISPRi revealed opposing effects on TERT expression in pancreatic versus lung cancer cells, consistent with the antagonistic pleiotropy observed for this signal. Furthermore, CRISPRi nominated an intronic CLPTM1L variable number tandem repeat (VNTR) as a potent enhancer. Long-read sequencing established VNTR polymorphisms as potential causal variants for the rs465498 signal. We showed that Hippo-pathway transcription factors mediate VNTR enhancer activity in lung and pancreatic cancer cells. Together, these findings indicate that cancer susceptibility at 5p15.33 may be mediated by both SNPs and VNTRs and provide an integrated framework for resolving complex pleiotropic loci.

Abstract The therapeutic landscape for systemic immunoglobulin light chain (AL) amyloidosis has been revolutionized by daratumumab-based regimens, achieving 76% 5-year overall survival in the landmark ANDROMEDA trial. However, the current prognostic models were developed using patient populations treated with now-suboptimal therapies, creating a critical gap between risk stratification models and contemporary outcomes. This comprehensive review analyzes prognostic factors and progression biomarkers in AL, categorizing them into disease-specific (clone-related and organ-related) and patient-specific factors. Notably, traditional baseline biomarkers including difference between involved and uninvolved free light chains and bone marrow plasma cell burden are losing prognostic significance with effective clone-directed therapies. Emerging approaches show promise, including dynamic markers such as minimal residual disease by free light chain mass spectrometry, cardiac imaging parameters such as global longitudinal strain, and functional measures. There is an urgent need for validation studies and prognostic model refinement to identify patients at high risk who may benefit from interventions beyond anti–plasma cell therapy.