Search Results (386)

Heart failure (HF) is a systemic syndrome in which prognosis depends on cardiac dysfunction, congestion, cardiorenal and cardiohepatic interactions, inflammation, and metabolic dysregulation. The liver is exposed to elevated systemic venous pressure and reduced forward flow, and contributes to albumin and coagulation factor synthesis, bile acid metabolism, iron homeostasis, and the acute-phase response. Cardiohepatic injury involves hemodynamic stress, sinusoidal endothelial dysfunction, oxidative stress, inflammatory signaling, fibrogenesis, and altered metabolic regulation. Congestive hepatopathy is associated with right-sided HF, tricuspid regurgitation (TR), pulmonary hypertension, and elevated central venous pressure, whereas hypoxic hepatitis develops during low-output states, shock, or acute circulatory deterioration. These mechanisms may coexist, producing congestive/cholestatic, hypoperfusive/ischemic and mixed/systemic reserve profiles. Composite liver-related scores, including Model for End-Stage Liver Disease (MELD), MELD excluding International Normalized Ratio (MELD-XI), MELD with sodium (MELD-Na), MELD-Albumin and albumin–bilirubin (ALBI) score, may reflect congestion, hepatorenal dysfunction, nutritional status and reduced systemic reserve. This review summarizes hemodynamic and molecular mechanisms of cardiohepatic injury, liver-related biomarkers and composite scores, with emphases on advanced HF, left ventricular assist device (LVAD) therapy and heart transplantation. Liver-related abnormalities remain underrecognized in HF. Their serial interpretation may support risk stratification, but composite scores should complement rather than replace comprehensive clinical assessment. Full article

Multimodal sentiment analysis aims to integrate heterogeneous textual, visual, and acoustic information for effective emotion understanding. However, existing methods often suffer from insufficient cross-modal interaction modeling, limited adaptability in multimodal fusion, and inadequate suppression of modality-specific noise under complex conversational scenarios. To address these challenges, this paper proposes a framework for learning adaptive cross-modal interactions for multimodal sentiment analysis. The proposed framework consists of three stages: modality-aware preprocessing, heterogeneous representation learning, and adaptive multimodal fusion. First, a unified preprocessing strategy is designed to improve cross-modal consistency through textual normalization, speaker-aware visual alignment, and utterance-level acoustic representation enhancement. Second, modality-specific encoders are constructed to capture complementary semantic, spatial, and utterance-level acoustic characteristics from textual, visual, and acoustic modalities, respectively. Third, an adaptive fusion framework is introduced to explicitly model cross-modal interactions, dynamically estimate the importance of different modality combinations, and further calibrate discriminative feature channels through channel attention. By jointly performing modality-level interaction learning and channel-wise feature refinement, the proposed framework effectively enhances multimodal representation capability for sentiment classification. Extensive experiments conducted on the CMU-MOSI and MELD benchmark datasets demonstrate that our framework consistently outperforms previous methods. In particular, the proposed model achieves 90.27% accuracy and 90.26% F1-score on CMU-MOSI, together with 66.57% accuracy and 66.21% F1-score on MELD. Additional ablation studies and qualitative analyses further validate the effectiveness of the proposed preprocessing strategy, modality-specific representation learning, and adaptive fusion mechanism. Full article

Background: Systemic inflammation significantly impacts graft survival and clinical outcomes following living donor liver transplantation (LDLT). The systemic immune-inflammation index (SII), which integrates peripheral neutrophil, platelet, and lymphocyte counts, has shown prognostic value in various clinical settings, but its role in LDLT has not been thoroughly investigated. Methods: This retrospective cohort study included 378 consecutive adult patients with end-stage liver disease (ESLD) who underwent primary LDLT between March 2016 and February 2025. The SII was calculated as (neutrophil × platelet)/lymphocyte. Spearman correlation analysis was used to assess the relationships between the preoperative SII, neutrophil-to-lymphocyte ratio (NLR), and platelet-to-lymphocyte ratio (PLR) and clinical outcomes, including the preoperative Model for End-Stage Liver Disease (MELD) score, duration of mechanical ventilation, and lengths of stay in the intensive care unit (ICU) and hospital. Receiver operating characteristic (ROC) curve analysis determined the optimal cut-off values for predicting 30-day mortality, and the areas under the curve (AUROCs) were compared using the DeLong test. Multivariable logistic regression was used to identify independent predictors of 30-day mortality. Results: Among 378 analyzable patients, the 30-day mortality rate was 7.9% (30/378). Both the SII and the NLR were significantly higher in non-survivors than in survivors (SII: median 368.8 vs. 169.1, p < 0.001; NLR: 6.0 vs. 2.4, p < 0.001), whereas the PLR did not differ significantly. On ROC analysis for 30-day mortality, the NLR and the SII showed comparable discrimination (NLR AUROC = 0.729, 95% CI: 0.63–0.82; SII AUROC = 0.705, 95% CI: 0.60–0.80; DeLong p = 0.43), both exceeding the PLR (AUROC = 0.541). The optimal SII cut-off was 275 × 10⁹ cells/L (sensitivity 70.0%; specificity 69.5%). Patients with an SII ≥ 275 × 10⁹ cells/L had significantly lower 30-day survival than those below the cut-off (83.5% vs. 96.4%; log-rank p < 0.001). On multivariable logistic regression adjusting for age and MELD score as continuous variables, an SII ≥ 275 × 10⁹ cells/L remained an independent predictor of 30-day mortality (aOR = 3.78; 95% CI: 1.60–8.93; p = 0.002); the MELD score was also independently predictive (aOR = 1.04 per point; 95% CI: 1.01–1.08; p = 0.018). The association persisted when the SII was modelled continuously (aOR = 1.60 per unit log SII; 95% CI: 1.08–2.38; p = 0.020). At the 275 cut-off, sensitivity for 30-day death was 70.0% and the positive predictive value 16.5%. Conclusions: The preoperative SII and NLR are simple, inexpensive, CBC-derived inflammatory indices significantly associated with 30-day mortality after LDLT for ESLD. An elevated preoperative SII independently predicts early post-transplant mortality and may aid perioperative risk stratification, although it does not outperform the NLR. Full article

Background/Objectives: In cirrhotic patients, guidelines generally recommend maintaining mean arterial pressure (MAP) ≥ 65 mmHg, but the prognostic impact of MAP in transplant candidates is unclear. This study aimed to evaluate the association between MAP and waitlist mortality and cirrhosis-related complications in patients listed for liver transplantation. Methods: We conducted a retrospective cohort study of 103 adults (age ≥ 18 years) with cirrhosis listed for liver transplantation (MELD 20–24) at a single center (2019–2023). Patients with hepatocellular carcinoma were excluded. The primary outcome was death on the transplant waitlist (n = 9 events). Logistic regression was used to assess the association of MAP (per 1 mmHg) with mortality. Continuous variables were compared using the t-test, and categorical variables were compared using the chi-square test; p < 0.05 was considered significant. Results: Mean MAP at listing was significantly higher in survivors than non-survivors (83.2 ± 9.4 vs. 76.9 ± 8.7 mmHg; p = 0.04). In logistic regression, higher MAP was associated with lower odds of waitlist death (unadjusted odds ratio [OR] per 1 mmHg increase = 0.94; 95% confidence interval [CI] 0.89–0.99; p = 0.041). Subgroup analysis showed a significant inverse association between MAP and hepatorenal syndrome (HRS) (OR per 1 mmHg = 0.94; 95% CI 0.89–0.98; p = 0.011), whereas MAP was not significantly associated with hepatic encephalopathy, ascites, or variceal bleeding (all p > 0.2). Conclusions: Among the cirrhotic patients listed for transplantation, lower MAP at baseline is associated with waitlist mortality and hepatorenal syndrome. These findings should be interpreted cautiously given the small number of events and require validation in larger cohorts. Full article

Background/Objectives: Cardiac comorbidities are increasingly common among liver transplant candidates, particularly those with metabolic dysfunction-associated steatohepatitis (MASH)-related cirrhosis. Although the Liver Transplant Comorbidity Index identifies coronary artery disease (CAD) as a predictor of post-transplant mortality, the impact of overall cardiac comorbidity burden on early outcomes after liver transplantation remains unclear. We evaluated the association between pre-transplant cardiac comorbidities and early post-transplant outcomes, with emphasis on MASH-related cirrhosis. Methods: We performed a retrospective cohort study using the TriNetX US Collaborative Research Network. Adults undergoing first-time isolated liver transplantation through May 2026 were included. Pre-transplant CAD, heart failure (HF), and atrial fibrillation (AF) documented within 12 months before transplantation were identified using ICD-10-CM codes. Patients were categorized by cardiac comorbidity burden (0–3 conditions). Recipients with any cardiac comorbidity underwent 1:1 propensity score matching to those without cardiac disease using 16 baseline demographic, clinical, and laboratory variables, including MELD-Na. The estimand was the average treatment effect in the treated patients. Primary outcomes comprised 30- and 90-day all-cause mortality. Secondary outcomes included a prespecified restricted major adverse cardiac event (MACE) composite, limited to hard endpoints (death, myocardial infarction, cardiac arrest, ischemic stroke), and a broader composite, i.e., acute kidney injury, prolonged mechanical ventilation, vasopressor requirement, renal replacement therapy, ICU and hospital length of stay, and 90-day readmission. Results: Among 5124 recipients, 986 (19.2%) exhibited at least one cardiac comorbidity. After matching, 974 patients remained in each group. Pre-transplant cardiac comorbidity was associated with higher 30- and 90-day mortality and increased risks of all secondary outcomes. The association with MACE persisted but was attenuated when restricted to hard endpoints (90-day RR 1.55; 95% CI 1.19–2.03) when compared with the broad composite (RR 1.75; 95% CI 1.40–2.18). MASH recipients with cardiac comorbidities experienced numerically higher event rates than did non-MASH recipients, but interaction estimates were imprecise and non-significant. In separate matched analyses, AF was most strongly associated with MACE, whereas CAD showed the strongest association with mortality. Conclusions: Pre-transplant cardiac comorbidity burden is associated with worse early post-transplant outcomes. Although MASH-cirrhosis recipients experienced numerically higher event rates, exploratory subgroup analyses did not demonstrate statistically significant differences from the non-MASH recipients. These findings may help refine cardiac risk prediction and perioperative planning, but they do not establish that intensified cardiac risk stratification or perioperative optimization improve outcomes. Prospective studies incorporating detailed cardiac, donor, operative, frailty, and medication data are needed to validate these associations and determine whether targeted risk-stratification and perioperative strategies can improve post-transplant outcomes. Full article

Transjugular intrahepatic portosystemic shunt (TIPS) is an established treatment for complications of portal hypertension, but hepatic encephalopathy (HE), liver dysfunction, rebleeding, and mortality remain difficult to predict in otherwise eligible candidates. However, predicting post-TIPS outcomes remains challenging using conventional risk scores such as MELD 3.0 and Child–Turcotte–Pugh. This narrative review critically evaluates how clinical, biochemical, procedural, conventional imaging, handcrafted radiomics, and deep-learning features can be integrated for personalized post-TIPS risk prediction, supplemented by backward and forward reference checking. Thirty-two original post-TIPS studies met the core inclusion criteria: 18 focused primarily on clinical, biochemical, hemodynamic, microbiome, or procedural predictors and 14 on imaging, body composition, radiomics, or multimodal models. AI, ML, and radiomics models, by the aim of logistic regression, tree-based ensembles, support vector machines, artificial neural networks, and hybrid deep-learning models, able to capture non-linear interactions, have consistently demonstrated improved predictive performance compared with conventional scores, particularly for HE, with reported incidences of approximately 20–47%, mortality, and liver dysfunction. Their advantage lies in the ability to model complex, non-linear relationships and integrate heterogeneous data sources, including laboratory parameters, ammonia levels, hemodynamic variables, and imaging-derived features. Radiomics and deep learning approaches further enhance predictive accuracy. CT is currently the principal imaging substrate, while direct post-TIPS radiomics evidence for MRI and ultrasound remains sparse. Studies of liver and spleen morphology, portal-vein geometry, muscle and adipose tissue, and radiomic texture suggest incremental information beyond conventional scores, particularly when clinical and imaging features are combined; however, negative volumetric findings show that additional image features do not automatically improve prediction. However, most studies are retrospective, single-center, and lack external validation and no validated transformer-based or other sequence model has yet been established for post-TIPS outcomes. Standardization issues in radiomics and limited model interpretability remain significant barriers. Future directions should lead to prospective multicenter cohort validation, increasing sample sizes, harmonized imaging and endpoint definitions, locked external validation with recalibration, and the development of clinically interpretable tools to make it easier to identify those patients suitable for TIPS and their post-procedural management. Full article

Background/Objectives: The inflammatory response is central to the pathogenesis of alcohol-associated liver disease (ALD), yet currently used prognostic scores, including Maddrey’s modified discriminant function (mDF), Child–Turcotte–Pugh (CTP), MELD-Na and MELD 3.0, reflect hepatic synthetic and excretory function rather than the underlying immune alterations. Since CC-motif chemokines (CCLs) regulate immune cell trafficking and hepatic stellate cell activation, they warrant deeper investigation, as reports on their role in ALD remain limited. Therefore, the primary objective of this study was to evaluate the diagnostic and prognostic performance of selected serum CCLs, and to correlate their levels with disease severity and short-term survival in patients with ALD. Methods: Serum MIP-3α/CCL20, MCP-1/CCL2 and TARC/CCL17 were measured in 63 patients with ALD and 25 healthy controls using ELISAs. Chemokine concentrations were correlated with liver function tests, inflammatory indices and short-term (30-day) survival. The diagnostic and prognostic performances of individual CC chemokines were assessed by ROC analysis and by univariate and multivariable logistic regression. Results: MIP-3α/CCL20 was increased more than 11-fold in ALD versus healthy controls and reached an AUC of 0.978 for distinguishing the two groups in a case-control comparison (a proof-of-principle, case-control result which requires validation against other chronic liver diseases). It increased across Child–Turcotte–Pugh classes and correlated with all other prognostic scales. MCP-1/CCL2 was significantly elevated in the ALD group but did not discriminate ALD severity. TARC/CCL17 was significantly lower in ALD than in controls. On multivariable regression, MIP-3α/CCL20 was the dominant predictor of both ALD status and severe liver dysfunction. Conclusions: In patients with ALD, MIP-3α/CCL20 appears to bridge hepatic inflammation and hepatocellular failure and is a candidate exploratory biomarker of disease severity that requires further confirmation against other chronic liver diseases. The short-term mortality analysis was based on only seven events. This evaluation is exploratory and underpowered, and its findings should not be interpreted as definitive prognostic evidence. MCP-1/CCL2 does not stratify ALD severity. Decreased TARC/CCL17 concentrations point to a separate immunoregulatory failure and are presented as a hypothesis requiring mechanistic testing in pre-clinical models. Full article

Speech emotion recognition (SER) is a fundamental task in affective computing; however, traditional unimodal approaches often struggle to capture the complex emotional cues present in spontaneous conversational speech. Bimodal frameworks that integrate acoustic and textual information have therefore emerged to provide complementary semantic and acoustic representations. This study proposes a bimodal SER framework based on a hybrid convolutional neural network–long short-term memory (CNN–LSTM) architecture. Using the Multimodal EmotionLines Dataset (MELD), the framework combines temporal acoustic features, statistical acoustic features, and predicted textual sentiment. Experimental results indicate that the proposed model achieves reliable recognition of majority emotion classes but exhibits limited performance on underrepresented minority classes due to severe class imbalance. To better understand the contribution of each modality, feature sufficiency and feature necessity analyses were conducted. Furthermore, an evaluation of alternative fusion strategies showed that the expressive attention networks did not provide meaningful performance improvements over simple feature concatenation. These findings suggest that class imbalance, rather than fusion complexity, remains the primary limitation in conversational SER, highlighting the importance of addressing data imbalance before pursuing more sophisticated multimodal architectures. Full article

Speech emotion recognition (SER) infers affective states from speech signals, but positional encoding for acoustic tokens remains underexplored in Transformer-based SER. Existing models often reuse encodings designed for text and do not explicitly account for the different sequential and two-dimensional structures of acoustic representations. We propose Fibonacci Position Embedding (FPE) and Fibonacci Target Shutter (FTS). FTS constructs overlapping candidate-index sets over time–frequency token grids, and FPE samples a Fibonacci index and applies its modulo-wrapped, dimension-dependent phase rotation to query and key vectors. The modules are integrated into STRE-Former, which fuses Wav2Vec, log-mel spectrogram, and MFCC representations through asymmetric cross-representation attention with representation-specific positional encodings. We also introduce an implementation-consistent conditional-entropy formulation that quantifies uncertainty in recovering a token location from its sampled positional representation; this quantity characterizes positional ambiguity rather than downstream modeling capacity. Experiments over 64 positional-encoding combinations on IEMOCAP and MELD identify dataset-dependent highest-observed configurations, reaching 74.21% weighted accuracy on IEMOCAP-4, 74.54% on IEMOCAP-6, and 49.44% on MELD. These empirical observations suggest that the relative behavior of positional-encoding strategies may depend on the acoustic representation and evaluation dataset, rather than supporting a single universally optimal scheme. Full article

Background/Objectives: Tuberculous constrictive pericarditis (TCP) is characterized by “backward failure” and systemic venous congestion, leading to secondary multi-organ dysfunction. The Model for End-Stage Liver Disease (MELD) 3.0 score assesses liver, renal, and nutritional status, yet its prognostic utility in TCP remains undefined. This study aimed to evaluate the preoperative MELD 3.0 score as a predictor of time to discharge in patients undergoing pericardiectomy for TCP. Methods: We analyzed 190 patients undergoing pericardiectomy for TCP between 2018 and 2024. The association between preoperative MELD 3.0 scores and time to discharge (ICU, postoperative, and total) was assessed using Fine–Gray competing risks regression to account for in-hospital mortality. Models were adjusted for age, albumin, and anti-tuberculosis therapy. Results: Each unit increase in MELD 3.0 was associated with higher operative mortality (OR 1.21, p = 0.013). In multivariable analysis, a high MELD 3.0 score (>median 9.6) was independently associated with a lower probability of discharge (i.e., delayed discharge) for total hospital stay (subdistribution hazard ratio [sHR] 0.606, 95% CI 0.453–0.811), postoperative stay (sHR 0.665), and ICU stay (sHR 0.658). Preoperative albumin was also a strong predictor of faster discharge. Conclusions: The preoperative MELD 3.0 score independently predicts delayed hospital discharge after pericardiectomy for TCP, even when accounting for mortality risk. It serves as a pragmatic tool for quantifying the systemic burden of congestion and stratifying perioperative risk. Full article

Background/Objectives: The HALP score is a simple immunonutritional index calculated from routine preoperative hemoglobin, albumin, lymphocyte, and platelet values. We evaluated its association with mortality after liver transplantation for non-malignant indications and its prognostic contribution when considered with the Model for End-stage Liver Disease (MELD) score. Methods: We retrospectively analyzed 194 adults who underwent liver transplantation for non-malignant disease at a single center between January 2015 and January 2025. The primary outcome was all-cause post-transplant death. The cohort-derived HALP threshold was used for exploratory Kaplan–Meier and Cox analyses, while continuous HALP was examined in sensitivity analyses. Censoring-adjusted cumulative/dynamic ROC analyses evaluated discrimination at 30 days, 90 days, 1 year, and 3 years. Results: Over a median observed follow-up of 1108 days, 43 recipients (22.2%) died. Non-survivors had lower HALP values than survivors [0.52 (0.39–0.67) vs. 0.77 (0.54–1.04); p < 0.001]. HALP had the highest observed conventional AUC among the evaluated markers for overall mortality (AUC 0.735; 95% CI 0.657–0.814); pairwise AUC differences were not formally tested. The exploratory threshold was 0.6786. Mortality occurred in 36.3% of the low-HALP group and 9.7% of the high-HALP group. After adjustment for MELD and albumin, high versus low HALP remained associated with lower mortality (HR 0.239; 95% CI 0.116–0.493; p < 0.001). HALP also showed consistent time-dependent discrimination across the four evaluated horizons (AUC 0.713–0.738). Conclusions: Lower preoperative HALP was associated with higher post-transplant mortality in this non-malignant liver transplant cohort. HALP may provide complementary risk information, but the cohort-derived threshold requires independent external validation before clinical use. Full article

Background/Objectives: Nonselective beta-blockers (NSBBs) are foundational for variceal bleeding prophylaxis in cirrhosis. After transjugular intrahepatic portosystemic shunt (TIPS) placement, portal pressure is mechanically decompressed, and practice guidance generally supports discontinuing NSBBs when shunt function is adequate and no other indication exists; however, real-world adoption and outcomes associated with post-TIPS NSBB strategies remain incompletely characterized in large, multicenter populations. Methods: We performed a real-world, target-trial emulation in the TriNetX U.S. Collaborative Network using a prespecified 90-day landmark. Adults (≥18 years) with cirrhosis undergoing first TIPS who had NSBB prescribed in the prior year and who survived to post-TIPS day 90 were included. Strategy was classified during days 0–90 (continuation: ≥1 NSBB prescription; discontinuation: none). Propensity score matching (1:1) balanced demographics, comorbidities, portal hypertension complications, and laboratory values (including the components of the Freiburg Index of Post-TIPS Survival [FIPS], Model for End-stage Liver Disease–Sodium [MELD-Na], and Child–Pugh scores). Follow-up began at day 90 and continued through day 365. Primary outcomes were overall survival and transplant-free survival; secondary outcomes were hepatic encephalopathy (HE), esophageal variceal bleeding (EVB), and ICU admission. Results: Among 5111 patients (continuation: n = 2558; discontinuation: n = 2553), 2180 matched pairs were analyzed. One-year survival was similar (89.3% vs. 89.6%; HR: 1.03; 95% CI: 0.84–1.27), as was transplant-free survival (78.8% vs. 77.9%; HR: 0.95; 95% CI: 0.82–1.10). The cause-specific hazard of HE was higher with continuation (HR; 1.29; 95% CI: 1.14–1.46), while those of EVB (HR: 1.08; 95% CI: 0.90–1.29) and ICU admission (HR: 1.02; 95% CI: 0.86–1.22) were similar. Results were consistent in both strict discontinuation and adherence-based sensitivity analyses. Conclusions: In stabilized post-TIPS patients with prior NSBB use, continuation was not associated with improved 1-year survival or transplant-free survival and was associated with a higher cause-specific hazard of HE. Full article

Despite the rapid advancement of Emotion Recognition in Conversation (ERC), prevailing systems that primarily integrate language and speech exhibit substantial performance disparities on underrepresented emotion classes (e.g., Fear, Disgust). This study investigates whether fine-grained non-verbal visual modalities (facial action units, hand gestures, and body pose) can effectively mitigate these biases. We propose a multi-stream fusion architecture combining language, speech, and engineered pose-aware visual features, trained with class-imbalance-aware objectives. Experiments on MELD demonstrate that hybrid pose augmentation improves F1 on the least frequent classes: Fear +9.19%, Disgust +6.07%, Sadness +6.46%. We achieve an overall weighted F1 of 68.79%, competitive with recent state-of-the-art systems while uniquely targeting minority-class debiasing. These results establish fine-grained body language as a critical debiasing signal, recovering accuracy on the subtle expressions that text and speech alone fail to capture. Full article

We introduce the Probabilistic Multimodal Consistency Index (PMCI), a prototype measure of probabilistic semantic agreement used to quantify how much two modalities share in terms of mutually compatible affective evidence. PMCI does not model feature fusion or emotion classification directly; instead, each modality is mapped to a different probability distribution over learnable latent affective-agreement anchors, and the agreement of those distributions is quantified via the Jensen–Shannon divergence. Consequently, we propose PMCI not as a substitute for a discriminative model of pair matching, but rather as an auxiliary diagnostic index of cross-modal affective agreement. Experiments were conducted on pose-based facial and body keypoint sequences obtained from the RAVDESS and MELD datasets. In the updated RAVDESS setup, we extended the cache of preprocessed keypoint files to include all 24 actors, and we sampled four distinct pose-derived 12-frame crops per source video. This resulted in a total of 11,520 pose-derived windows, with the actor label determining the train/validation/test splits for the RAVDESS dataset. MELD contained 402 filtered pose-derived windows and served as an auxiliary in-the-wild validation setting with additional noise. When applying the updated RAVDESS standard pair-matching, DirectCosine_K0 achieved ROC–AUC = 0.943 and PR–AUC = 0.917, demonstrating that it is indeed the best exact pair-matching baseline and that PMCI-based configurations are not as accurate as DirectCosine_K0 when performing common exact face–body pair discrimination. Under this protocol, PMCI_K8, PMCI_K16, and PMCI_K32 produced ROC–AUC scores of 0.851, 0.805, and 0.887, respectively. Running a one-window-per-source-video experiment with 100 repetitions yielded similar results in the following order: DirectCosine_K0 = 0.942 plus-minus 0.009, PMCI_K16 = 0.807 plus-minus 0.016, and PMCI_K32 = 0.893 plus-minus 0.012, which shows that the results obtained were not the result of only repeated temporal crops. In the DirectCosine-mined hard negative setting, PMCI produced modest diagnostic separation above chance, where Direct+PMCI_K32 yielded ROC–AUC = 0.582. In a suite of shuffle control, permutation control, model initialization control, temperature control, anchor usage control, control latency, and pose perturbation control experiments, the diagnostic stability of PMCI was tested, confirming that PMCI is sensitive to both pose-estimation accuracy and domain shift. Finally, PMCI must be interpreted as a probabilistic diagnostic index of cross-modal affective agreement that is based on pose, not as a general-purpose emotion-recognition system. Full article

Background: Benign biliary anastomotic strictures (BAS) remain among the most frequent complications after living donor liver transplantation (LDLT). Although multiple plastic stents (MPS) have traditionally been considered the standard endoscopic treatment, fully covered self-expandable metal stents (FCSEMS) and intraductal fully covered self-expandable metal stents (ID-FCSEMS) have emerged as alternative strategies. This study aimed to compare the efficacy and safety of MPS, FCSEMS, and ID-FCSEMS in the management of post-transplant biliary anastomotic strictures. Methods: This retrospective single-center cohort study included consecutive adult patients who developed benign biliary anastomotic strictures after LDLT and underwent ERCP-based treatment between January 2020 and January 2025. Patients were categorized according to the initial stent strategy: MPS (n = 32), FCSEMS (n = 30), and ID-FCSEMS (n = 25). Primary outcome was overall treatment success. Secondary outcomes included recurrence, number of ERCP sessions, treatment duration, post-procedural hospitalization, stent migration, and procedure-related adverse events. Results: A total of 87 patients were included. Clinical success was achieved in 81.3% of patients treated with MPS, 83.3% treated with FCSEMS, and 92.0% treated with ID-FCSEMS (p = 0.501). Recurrence rates were 18.8%, 10.0%, and 12.0%, respectively (p = 0.579). Patients treated with metal stents required significantly fewer ERCP sessions than those treated with MPS (median 5 vs. 2 vs. 2; p < 0.001). Median treatment duration was significantly shorter in the FCSEMS and ID-FCSEMS groups compared with the MPS group (12 vs. 10 vs. 9 months; p < 0.001). Post-procedural hospitalization was also reduced with metal stents (7 vs. 4 vs. 3 days; p < 0.001). Adverse event rates, including cholangitis, pancreatitis, bleeding, perforation, and mortality, were comparable among groups. Stent migration occurred in 21.9% of MPS patients, 23.3% of FCSEMS patients, and 12.0% of ID-FCSEMS patients (p = 0.525). On multivariable analysis, MELD score was the only independent predictor of treatment success (adjusted OR 0.90, 95% CI 0.83–0.98; p = 0.012). Conclusions: MPS, FCSEMS, and ID-FCSEMS achieved comparable rates of stricture resolution, recurrence, and adverse events in the treatment of biliary anastomotic strictures after LDLT. However, FCSEMS and ID-FCSEMS significantly reduced the number of ERCP procedures, treatment duration, and hospitalization compared with conventional plastic stenting. Among the evaluated strategies, ID-FCSEMS demonstrated the highest numerical success rate and lowest migration rate, suggesting a potential procedural advantage while maintaining comparable safety and efficacy. Full article