Biomedical Signal Processing and Control

Pneumonia is a leading cause of infectious disease mortality worldwide, accounting for approximately 2.5 million deaths annually and 15% of deaths in children under five. Chest X-ray imaging remains the primary diagnostic tool, but accurate interpretation requires radiological expertise that is disproportionately concentrated in high-income settings, creating a diagnostic gap where disease burden is highest. Automated deep learning offers a scalable complement to specialist-dependent diagnosis, yet clinical adoption requires both high accuracy and transparent, interpretable reasoning. Convolutional neural networks (CNNs) have shown strong potential for pneumonia detection from chest X-rays, but two barriers impede clinical translation: the interpretability of black-box models and the computational feasibility of large architectures in resource-constrained settings. Explainable AI (XAI) methods such as Grad-CAM, Grad-CAM++, and Score-CAM address the interpretability barrier, yet systematic quantitative comparisons across multiple CNN architectures remain scarce. Furthermore, CNN architectures widely used for medical image classification carry high parameter counts that limit feasibility in resource-constrained settings, motivating architectures that achieve competitive accuracy with substantially fewer parameters. Here we propose a parameter-efficient deep learning framework for pneumonia detection based on transfer learning, evaluated across three CNN architectures representing distinct architectural families: EfficientNet-B0 with fine-tuning (proposed method), ResNet50, and DenseNet121, trained under identical conditions on the Kaggle chest X-ray dataset (5,863 images). Our method achieved 90% classification accuracy, outperforming both baselines while requiring 4.8x fewer parameters than ResNet50. To evaluate explainability, Grad-CAM, Grad-CAM++, and Score-CAM were applied across all three architectures and compared quantitatively using Intersection over Union against manually annotated lung segmentation masks, Insertion score, and Deletion score, with pairwise statistical validation via Wilcoxon signed-rank tests and Bonferroni correction. Findings show that classification accuracy and XAI explanation quality must be evaluated independently, and that the proposed parameter-efficient architecture offers a favorable trade-off for resource-constrained clinical deployment.

Minimally invasive endovascular procedures offer reduced surgical trauma, shorter recovery times, and improved outcomes, but rely on 2D fluoroscopic X-ray imaging, which provides limited depth perception and exposes patients and clinicians to ionizing radiation. Here we present an augmented reality (AR) system that fuses intravascular ultrasound (IVUS) and electromagnetic (EM) position tracking with preoperative computed tomography (CT) to produce an anatomically accurate, deformation-corrected navigational reference. A robotic device performs ECG-gated pullback of the IVUS probe, capturing 4D aortic motion across the cardiac cycle. We introduce a deep learning architecture for extracting vascular lumen boundaries and side-branch orifices from artifact-prone IVUS streams, and a semantically driven non-rigid CT-IVUS fusion pipeline robust to false positive landmarks. We evaluate the platform with trained surgeons in benchtop phantom studies and in-vivo ovine models, and demonstrate its application to fenestrated endovascular aneurysm repair (FEVAR). Compared to fluoroscopy alone, AR guidance significantly reduces cannulation time, radiation exposure, and cognitive workload, while improving procedural efficiency and safety. Our IVUS-EM and CT aortic datasets are released open source.

The aim of this work is to validate patient-specific EEG baseline establishment using the e-norms method as a screening and retrospective-review tool for seizure detection in pediatric epilepsy. The method was applied to 247 seizure-free EEG recordings (263.92 hours) from 10 patients in the CHB-MIT Scalp EEG Database (ages 3-18). A composite stability metric combining first-derivative dynamics, spectral entropy, variance, and line length was computed per 2-second epoch across 23 channels. Patient-specific detection thresholds were derived from each patient's seizure-free baseline using a weighted statistical procedure. Performance was validated against 72 expert-annotated seizures (2,705 epochs) across 62 seizure files, with durations spanning 6 to 264 seconds (44-fold range). The results show that detection achieved 94.4% event-level sensitivity (68 of 72 seizures; 95% CI 86.6-97.8%) and 81.5% epoch-level sensitivity (2,204 of 2,705 epochs; 95% CI 80.0-82.9%). Eight of ten patients achieved 100% event-level sensitivity with epoch-level sensitivity ranging from 58.7% to 100.0%. Two patients showed partial event-level failures (CHB-15: 17 of 20; CHB-18: 5 of 6), with the four missed events attributable to two characterizable failure modes. Patient-specific thresholds ranged from 4.06 to 4.81 (mean 4.51 +/- 0.25); threshold variation did not correlate reliably with age or sex, confirming that no universal threshold could achieve comparable performance. Detection margins ranged from 0.88 to 1.24 times. Patient-specific e-norms achieves 94.4% event-level sensitivity for pediatric EEG seizure detection without requiring labeled seizure training data, exceeding published human expert inter-rater agreement (50-76%) and recent automated approaches in adult cohorts using behind-the-ear EEG and wearable ECG. Two characterizable failure modes account for the four missed events and inform appropriate clinical use. As a high-sensitivity screening tool complementary to real-time alarm systems, the method is ready for adult validation, prospective deployment, and head-to-head benchmarking.

Motor speech disorders (MSDs) are early markers of neurological disease, but expert perceptual analysis is rarely available outside specialized centers. Automated speech analysis offers a scalable alternative, yet prior studies have not systematically compared modeling approaches or assessed clinically relevant metrics in independent datasets. This study compared static acoustic features, articulatory informed Phonet features, and self-supervised pretrained models for binary and multi label MSD classification. We trained and evaluated models on 583 speech samples using speaker level splits. Baseline models included logistic regression and Gated Recurrent Units (GRUs) trained on eGeMAPS and MFCCs. We extracted three types of Phonet derived features and evaluated pretrained HuBERT and SSAST models in frozen, partially fine-tuned, and fully fine-tuned configurations. Binary classification distinguished MSDs from controls, while multi label classification identified six MSD subtypes. Models were assessed using validation AUC, and cut points were tested on two independent datasets. Pretrained and Phonet based models substantially outperformed static acoustic features. In binary classification, HuBERT achieved the highest AUC (0.95), while compact Phonet derived GRUs achieved comparable performance (up to 0.94). These models generalized well to independent datasets, maintaining high sensitivity (0.94) and specificity (0.97). In multi label classification, Phonet models achieved the highest macro average AUC (0.86), but threshold-based subtype performance declined on unseen data. Automated MSD detection is feasible and clinically promising. Binary classification generalized well, whereas multi label classification showed limited threshold stability across datasets.

BACKGROUND: Artificial intelligence (AI) medical scribes rely on speech-to-text (STT) models for transcription. Evaluations of STT models in non-English settings remain scarce. We benchmarked ten STT models on medical consultations from Latin American (LatAm) Spanish and assessed whether fine-tuning improves transcription accuracy. METHODS: Ten YouTube videos depicting medical consultations. Human transcriptions were the ground truth. Five open-source models were evaluated: Whisper Large, Whisper Large v3, Whisper Large v3 Turbo, Voxtral Mini 3B, and Canary 1B v2; and so were five close-source models: gpt-4o-transcribe, gpt-4o-mini-transcribe, gemini-2.5-pro, Eleven Labs, and Assembly AI. Whisper Large v3 was fine-tuned. One video was withheld from training. Performance assessed using Word Error Rate (WER), Character Error Rate (CER), BLEU Score, ROUGE-L, BERT Score, and Semantic Similarity on the one withheld video. RESULTS: None of the fine-tuning iterations outperformed the vanilla Whisper Large v3. With the withheld video, Gemini-2.5-pro was the close-source model with the best performance in four of six metrics. In comparison to the close-source models, the fine-tuned model never outperformed the other models (withheld video); conversely, in comparison to the close-source models, the fine-tuned model showed better performance across metrics, for instance: BLEU score (63% vs to 58% for the second-ranking model), BERT (89% vs to 86%), and semantic similarity (89% vs to 83%), CER (19% vs 20%). CONCLUSIONS: Whisper Large v3 and its fine-tuned variant are the best open-source STT models for transcribing medical conversations in LatAm Spanish. These findings provide an evidence base for developing AI medical scribes tailored to Spanish-speaking LatAm.

The apnea-hypopnea index (AHI), the conventional metric of obstructive sleep apnea (OSA) severity, is typically studied using scalar summaries of sleep architecture, such as the total time spent in each sleep stage. Although clinically interpretable, these summaries fail to capture the temporal organization of overnight sleep-stage sequences and may obscure stage-specific associations with OSA severity. Modeling the complete sleep-stage trajectory provides substantially richer temporal information; however, because total sleep duration varies across individuals, sleep-stage trajectories are observed over subject-specific domains, limiting the applicability of conventional functional regression methods that assume a common observation interval. We therefore applied Variable-Domain Functional Regression (VDFR) to overnight polysomnographic data from the APPLES study (n= 1,103), treating the epoch-by-epoch sleep-stage sequence as a continuous, variable-length functional predictor of AHI. We compared three levels of sleep-stage granularity: five stages (Wakefulness, N1, N2, N3, REM), three stages (Wakefulness, Non-REM, REM), and binary staging (Wakefulness vs. Sleep). Functional sleep-stage terms were significant across all staging granularities and model structures (all p-values [≤]0.001). Wake, N1, and N2 were positively associated with AHI, whereas N3 and REM were negatively associated, with REM exhibiting the strongest association. These effects were attenuated under coarser staging representations, highlighting the importance of preserving fine-grained sleep architecture. To our knowledge, this is the first application of VDFR to overnight polysomnographic data in OSA, showing that accommodating subject-specific sleep durations enables the identification of stage-specific temporal associations with AHI severity that are attenuated or obscured by coarser staging and conventional scalar analyses.

As machine learning and artificial intelligence systems are increasingly used in healthcare, rigorous evaluation of their classification performance has become critical. The F1 and F{beta} scores are widely adopted metrics for assessing performance in imbalanced biomedical data. Recently, we introduced psF1, a unified statistical framework for inference and study design for single and comparative F1 and F{beta} scores under the assumption of independent classifiers. In practice, however, benchmarking two classifiers on the same dataset creates a correlated paired setting. Ignoring this intrinsic dependency leads to overestimation of the standard error and a substantial loss of statistical power. To address this, we develop psF1pair, an advanced framework for statistical inference and power analysis that explicitly accounts for correlations between classifier pairs. Extensive simulation studies demonstrate the performance of psF1pair, and its utility is further illustrated through application to a real-world imaging classification system. As expected, higher correlation between classifiers yields narrower confidence intervals and enhanced statistical power. A freely available R package is provided to facilitate implementation, supporting accurate evaluation and study design for predictive and classification models in biomedical research.

We develop a mathematical model of Rift Valley Fever integrating mosquito vectors, ruminants, and humans, based on an SEIR-type structure with vertical transmission in vectors. Local data from the Sudanian and especially the Sahelian zones are used to capture the impact of climatic variations on mosquito population dynamics. The mathematical analysis establishes the models positivity, determines the basic reproduction number R0, and demonstrates the local and global stability of the disease-free equilibrium. Sensitivity analysis (PRCC) highlights the most influential parameters, while the stochastic approach using a continuous-time Markov chain confirms the major role of seasonal rainfall. Numerical simulations reveal a peak in animal and human infections around the 9th month, correlating with periods of heavy rainfall. This model provides a relevant tool for surveillance and prevention within a "One Health" approach in Chad.

Background: Globally, at least 2.6 billion people need rehabilitation services and more than 2.5 billion people need assistive technology (AT). However, reliable data are lacking on population level need for rehabilitation services and assistive products (AP) in different settings for evidence-based policy and programme planning. This first study paper describes the development of the Functional Needs Assessment Tool (FNAT), a new survey tool developed to fill this data gap between 2018 and 2023. Objective: To develop a new multidomain tool to assess population-level functional difficulties and need for service and AP utilising both self-report and clinical assessment methodologies. Development stages: FNAT was developed based upon primary and secondary data analysis, existing survey tools and expert consultation through a series of four steps: Step 1 Inform, Step 2 Build, Step 3 Draft and Step 4 Develop. FNAT uses both self-reported and clinical assessment tools to estimate the prevalence of functional difficulties/impairment and the need for services and AP in the following seven domains: vision, hearing, mobility, communication, cognition, self-care and mental health. It uses a two-stage population-based assessment with data collection through a bespoke tablet-based mobile application and web-based platform. Discussion: FNAT is a new multi-domain modular tool developed to address data gaps by estimating prevalence of functional difficulties and service/AP needs in a population. Potential advantages and disadvantages were highlighted during the development stages, and the tool needs to be pilot tested to assess the feasibility of the methodology and the functionality of the tablet-based mobile data collection application.

BACKGROUND In line with International Federation of Clinical Chemistry and Laboratory Medicine (IFCC) recommendations for high-sensitivity cardiac troponin assays, analytical validation and reference limit assessments are required to confirm that an assay meets performance criteria. This study evaluated the analytical performance and established the 99th percentile upper reference limit (URL) for the SPINCHIP High-Sensitivity Cardiac Troponin I (SPINCHIP hs-cTnI) point-of-care assay. METHODS Analytical performance characteristics, including the limit of blank (LoB), limit of detection (LoD), and limit of quantification (LoQ), were assessed. Additionally, 1,053 plasma samples and 1,055 whole-blood samples were used to determine the URL. Imprecision around the 99th percentile URL was evaluated as part of the analytical validation. High-sensitivity criteria were assessed by confirming measurable cTnI in [&ge;]50% of healthy individuals (n=432 plasma; n=431 whole blood) and achieving imprecision <10% at the 99th percentile (plasma, n=960; whole blood, n=480). RESULTS SPINCHIP hs-cTnI demonstrated a LoB of 0.3 ng/L; LoDs of 0.8 ng/L (plasma) and 0.9 ng/L (whole blood); and LoQs of 1.1 ng/L (plasma) and 1.4 ng/L (whole blood). The analytical measuring range was 1.1-9,000 ng/L. Imprecision at the common 99th percentile URL (14 ng/L) was 5.8%; for men (URL=16 ng/L) 5.6% and for women (URL=10 ng/L) 6.3%. Greater than 85.2% (94.0% and 76.1% in men and women, respectively) of healthy individuals showed measurable cTnI above the LoD. CONCLUSIONS The SPINCHIP hs-cTnI assay meets the IFCC high-sensitivity requirements, demonstrating <10% imprecision at the 99th percentile, reliable low-concentration precision and cTnI detection in more than half of healthy individuals.

Aphasia diagnosis in Indonesia remains challenging due to limited culturally and linguistically appropriate instruments. Widely used tools such as the Boston Diagnostic Aphasia Examination (BDAE) and Western Aphasia Battery (WAB) are not adapted to the Indonesian context, while Tes Afasia untuk Diagnosis, Informasi, dan Rehabilitasi (TADIR) provides screening but lacks diagnostic accuracy. To address this gap, we developed the Instrumen Diagnosis dan Evaluasi Afasia (IDEA) for native Indonesian speakers and evaluated its validity, reliability, and normative cutoff values in cognitively healthy Indonesian adults. Eighty-three cognitively normal adults (screened using MoCA-Ina) with no history of neurological disease were assessed using IDEA, which evaluates six language domains. Items were adapted from existing tools and reviewed by experts. Content validity, internal consistency (Cronbachs alpha), and construct validity (Exploratory Factor Analysis) were analyzed using SPSS v25. A total of 83 participants were included (median age = 55.81 years, 54% secondary education). IDEA demonstrated good feasibility, with an average completion time of 45-60 minutes depending on participant engagement. Content validity was established by unanimous expert consensus. Construct validity showed meritorious sampling adequacy (KMO = .872) and significant sphericity (Bartletts test {chi}^2 (15) = 278.523, p<.001), supporting factor analysis. Internal consistency showed good reliability across six domains (Cronbachs = 0.896). IDEA is a valid and reliable tool for assessing aphasia in Indonesian natives. It is a culturally appropriate assessment tool which offers structured, domain-based evaluation and supports differential diagnosis of both classical and progressive aphasia syndromes. Keywords: Aphasia, Language Assessment, Indonesian, IDEA, Validity

Purpose This study aimed to examine the association between screen time and depressive symptoms among Chinese college students, and to investigate the mediating roles of sleep quality and emotion regulation in this relationship. Furthermore, a serial mediation model was constructed to elucidate the underlying psychological mechanisms linking screen exposure to depression. Methods A stratified cluster sampling method was employed to recruit 10,999 college students for a cross-sectional questionnaire survey. Data were collected on screen time, sleep quality, emotion regulation ability, and depressive symptoms. Descriptive statistics, correlation analyses, and regression analyses were conducted using SPSS 26.0 A serial mediation model was tested using the PROCESS macro (Model 6), and bootstrapping procedures were applied to estimate the significance of indirect effects. Results Correlation analyses indicated that screen time was significantly positively associated with depressive symptoms (r = 0.16, p < 0.01) and sleep quality (r = 0.15, p < 0.01), and significantly negatively associated with emotion regulation (r = -0.13, p < 0.01). Sleep quality was positively correlated with depressive symptoms (r = 0.31, p < 0.01), whereas emotion regulation was negatively correlated with depressive symptoms (r = -0.42, p < 0.01). Regression analyses further showed that screen time significantly positively predicted depressive symptoms ({beta} = 0.712, p < 0.001), positively predicted sleep quality ({beta} = 0.217, p < 0.001), and negatively predicted emotion regulation ({beta} = -0.085, p < 0.001). In addition, both sleep quality ({beta} = 1.318, p < 0.001) and emotion regulation ({beta} = -0.424, p < 0.001) were significant predictors of depressive symptoms. Mediation analyses demonstrated that sleep quality significantly mediated the association between screen time and depressive symptoms (95% CI [0.239, 0.332]), as did emotion regulation (95% CI [0.269, 0.416]). Moreover, a significant serial mediation effect of sleep quality and emotion regulation was observed in the relationship between screen time and depressive symptoms (95% CI [0.082, 0.117]). Conclusion Screen time is significantly associated with depressive symptoms among college students, with sleep quality and emotion regulation serving as important mediating mechanisms. Extended screen exposure may be linked to higher levels of depressive symptoms by impairing sleep quality and weakening emotion regulation capacity.

Purpose: To develop an accelerated motion-compensated diffusion-weighted pseudo-continuous arterial spin labeling (MCDW-pCASL) method using a spatial subspace low-rank reconstruction method for efficient quantification of blood-brain barrier (BBB) water exchange (kw) and permeability (PSw). Methods: An accelerated multidelay MCDW-pCASL sequence was developed to simultaneously encode intravascular and extravascular diffusion-weighted ASL signals across multiple post-labeling delays (PLDs). A spatial subspace low-rank reconstruction framework was optimized to enable joint estimation of cerebral blood flow (CBF) and BBB water exchange rate and permeability. Fourteen young healthy adults underwent test-retest scans (separated by ~1 week) at 3T with both the accelerated MCDW-pCASL and a conventional diffusion-prepared (DP) pCASL sequence. Whole-brain, gray-matter, and white-matter CBF and kw values were quantified to assess test-retest repeatability and cross-method agreement. An additional cohort of 30 older adults underwent single-session MCDW and DP scans to evaluate age-related perfusion and BBB kw/PSw differences. Intraclass correlation coefficients (ICCs) were used to assess reliability and agreement. Results: Accelerated MCDW-pCASL demonstrated excellent agreement with DP-pCASL for CBF (ICC = 0.89) and fair agreement for kw (ICC = 0.56). Test-retest repeatability of MCDW-pCASL was good for CBF, BBB kw and PSw (ICC {approx} 0.6). Across both sequences, younger subjects exhibited significantly higher CBF and kw compared with older adults. Conclusion: Incorporating a spatial low-rank subspace reconstruction enables accelerated MCDW-pCASL acquisition with reliable simultaneous quantification of CBF, BBB kw and PSw. Clinical applications of this method for assessing perfusion and BBB function are warranted.

Red blood cell (RBC) disorders arose as an advantageous evolutionary response to malaria infections. In a heterozygous condition, such as in Southeast Asian Ovalocytosis (SAO), hosts are protected against severe malaria. In malaria-endemic regions, RBC disorders are presumed to be highly prevalent. Tanjung Leidong, a moderately endemic area in North Sumatra (API 1.13 in 2024), lacks comprehensive data on RBC disorder prevalence beyond G6PD deficiency. Therefore, this study aims to characterize the RBC disorders in this region as well as to characterize the anemia status in children living in Tanjung Leidong. Schoolers attending D. I. Panjaitan elementary to high school were recruited and screened for malaria by microscopy and G6PD deficiency using the STANDARD G6PD Assay. The DNA of the participants was also extracted to be genotyped for SAO, Hemoglobin E (HbE), and -thalassemia. Exclusively, G6PD-deficient DNA samples were genotyped further to determine variants. The proportion of G6PD deficiency, SAO, HbE, -thalassemia one-gene deletion, and two-gene deletion were 0.90%, 0.90%, 2.40%, 6.26%, and 0.30%, respectively. Anemia prevalence was approximately 14%, and RBC disorders were observed across children with normal to obese BMI. No malaria infections were detected by microscopy. The predominance of asymptomatic RBC disorders highlights that they are protective against malaria infection, although their protective role against malaria could not be directly assessed in this study. Both nutritional and genetic factors are found to contribute to anemia in this cohort. These findings underscore the importance of integrated screening strategies for RBC disorders and anemia in malaria-endemic settings.

Malaria remains a severe health problem in endemic regions because people lack adequate diagnostic tools, leading to delayed medical care and elevated death rates. This research introduces a dual-mode artificial intelligence system that uses two complementary models to enhance malaria pre-screening and diagnosis. The patient-centered model uses multivariate logistic regression to analyze biosignals, including heart rate, body temperature, and oxygen saturation, collected through a wearable sensor prototype and a mobile interface for symptom analysis. The system enables patients to begin self-assessment to determine their level of need before scheduling a doctor's appointment. The clinician-centered model represents a customized convolutional neural network that uses annotated microscopy images of red blood cells to achieve 94.84% accuracy, 95.71% precision, 93.87% recall, 94.78% F1 score, and 0.84 Area Under Curve (AUC). The patient model achieved 94.6% accuracy and an AUC of 0.985 using a 70/30 train-test split. These systems work together to create a layered diagnostic system that can operate independently or together to detect malaria at an early stage, especially in areas with limited resources. The findings demonstrate that wearable biosignal data integration with image-based deep learning can produce dependable, scalable, and user-friendly systems for malaria pre-screening. Keywords - malaria diagnosis, artificial intelligence (AI), convolutional neural networks (CNN), wearable biosensors, multivariate logistic regression

Respiratory diseases cause considerable morbidity in autumn and winter and are a priority in public health monitoring. In Germany, they are subject to a number of surveillance systems, including both pathogen-specific and syndromic indicators. In this paper we present a collaborative multi-target and multi-model real-time forecasting system rolled out during the 2024/25 season, and discuss differences to earlier efforts carried out during the COVID-19 pandemic. A total of nine models were run to generate forecasts of general practitioner consultations for acute respiratory infections (ARI), hospitalizations for severe acute respiratory infections (SARI) and confirmed cases of seasonal influenza and RSV. As all indicators were subject to retrospective revisions, forecasting models were combined with a nowcasting step. Whenever multiple models were available for the same indicator, we combined them into an ensemble. Nowcasts showed convincing performance, even though for some models Christmas break effects led to an upward bias in early January. Forecasts were overall well-calibrated and most models outperformed simple benchmark models. These improvements were generally more substantial for age-stratified than pooled targets, and concentrated at lead times of two to three weeks. Anticipating the peak timing and magnitude proved to be challenging, with many models predicting too flat curves with a too early turnaround (e.g. already in late January rather than mid-February for SARI). The combined ensemble forecast was among the best-performing approaches, but unlike in previous related projects did not consistently outperform individual models. We conclude by discussing learnings on the organization of collaborative forecasting projects in post-COVID-19 times and the potential of AI-supported modelling.

Purpose: To compare the cross-site generalization of radiomic features and deep learning embeddings for MRI prediction of substantial lymphovascular space invasion (LVSI) in endometrial cancer. Materials and Methods: This retrospective two-center study included 206 women (mean age, 59.8 years) with endometrial cancer who underwent preoperative 3-T MRI from March 2016 to March 2023. Hospital A (n = 130) was used for development and Hospital B (n = 76) for strict external testing. T2-weighted, reduced field-of-view diffusion-weighted, and apparent diffusion coefficient images were manually segmented. Radiomic features and seed-pooled embeddings from 3D ResNet18, DenseNet121, and U-NEXtractor were modeled with elastic-net logistic regression or XGBoost. Out-of-fold Platt calibration and sensitivity-targeted thresholds were estimated using development data only. AUCs were summarized with 95% bootstrap confidence intervals. Results: External radiomics with elastic-net achieved an AUC of 0.609 (95% CI: 0.464, 0.740) and sensitivity of 0 of 12 (0%). DenseNet121 with elastic-net had the highest external AUC (0.685; 95% CI: 0.538, 0.822) but sensitivity of 3 of 12 (25%). U-NEXtractor with elastic-net detected 10 of 12 positive cases (83.3%) with specificity of 32 of 64 (50.0%) and balanced accuracy of 0.667. XGBoost showed higher apparent development performance but weaker external operating behavior. Conclusion: Under real-world cross-site MRI acquisition shift, DenseNet121 and U-NEXtractor embeddings showed better external generalization than handcrafted radiomic features for substantial LVSI prediction.

INTRODUCTION: The biomarker-based amyloid/ tau/ neurodegeneration (A/T/N) framework has become a popular staging method for Alzheimer's disease (AD) research. Previous studies use the framework either as a rule-based or data-driven approach but typically sacrifice either adaptivity or interpretability. METHODS: We present an interpretable, hybrid method, called Neurosymodal Data Fusion, for predicting incident AD in the ADNI dataset. Specifically, we encode the A/T/N framework as a logic program, where the input biomarker features are extracted by one or more neural networks. RESULTS: Our pipeline predicted four-year incident AD with a sensitivity of up to 0.84. Additionally, our models learned scores for each A/T/N profile, denoting relative importances to model predictions. These scores also indicated that empirically-derived cut-off values for the A and T criteria might be uninformative for the ADNI data. DISCUSSION: Our pipeline provides a novel way to use the A/T/N framework that could potentially improve early AD screening years before clinical manifestations.

Background: In Japan, acute inpatient care is divided into approximately 335 secondary medical care areas, which serve as the basic units for planning healthcare delivery systems under the 8th National Health Care Plan. While comparisons between regions and facilities typically rely on a single risk-adjusted metric, this approach confuses differences in patient demographics with differences in the actual infrastructure of intensive care units (ICUs). This paper presents a framework - MedZone Embedder - for deriving data-driven indicators of regional structural vulnerability by mapping secondary medical care areas onto a learned similarity space, together with its working implementation. The paper sets out the concept, the method, a proof of concept, and an explicit staged validation program, rather than national empirical results. Methods: Each area is represented by a feature vector consisting of aggregated values of intensive care provision indicators derived directly from the Japan Intensive Care Patient Database (JIPAD) - specifically, risk-adjusted mortality rates (standardized mortality ratios and an in-hospital composite indicator), technical efficiency, length of stay, readmission rates, case severity, and case composition - with the within-area variance of these indicators also taken into account. No hierarchical processing by facility type is performed. A contrastive autoencoder (multilayer perceptron encoder 32 -> 16 -> 8, symmetric decoder) is trained by self-supervised learning, using an objective function that combines reconstruction and normalized temperature cross-entropy (NT-Xent) on noise-augmented views. The resulting 8-dimensional embedding supports area searches based on cosine similarity and anomaly scoring in the embedding space (using isolation forest, Mahalanobis distance, or k-nearest-neighbor density), which is normalized to a vulnerability score ranging from 0 to 1. If deep learning libraries are unavailable, or if the number of areas is small, an alternative method using deterministic principal component analysis is employed. Results: This method was implemented and deployed within an operational ICU decision support system on a managed cloud platform. The proof of concept (PoC) is structured around five secondary medical care areas within Kyoto Prefecture and runs entirely on synthetic facility-level aggregate data constructed to follow the JIPAD indicator schema; no registry data were accessed. It generated: an aggregate provision profile for each area; an area embedding space equipped with a similar-area search function; and a vulnerability ranking that identifies areas with low patient numbers and low diversity that exhibit overall poor outcomes. At this scale, the contrastive autoencoder falls back to principal component projection. The deep learning pathway has been implemented and unit testing has been completed; training and evaluation on actual registry data are pending data-use approval and the expansion of data integration. Validation is staged: Stage 2 will train the contrastive pathway over JIPAD-covered areas to assess construct validity against public structural indicators (ICU/HCU beds, population, accessibility), and Stage 3 will extend coverage to all areas via National Database (NDB) linkage. Conclusion: MedZone Embedder reframes regional comparison from single-indicator ranking to structural representation: which areas are alike, and which are structural outliers. The contribution of this paper is the framework - the proposal that the intensive care provision structure of Japanese secondary medical care areas can be learned from a national outcomes registry and read through the lens of what we call institutional debt - together with a deployed implementation and a pre-specified validation program. To our knowledge, this is a candidate first application of contrastive representation learning to Japanese secondary medical care areas.

We conducted a scaling evaluation of unlabeled pretraining for electrocardiogram foundation model performance. One-dimensional vision transformer masked autoencoders were pretrained across increasing ECG volumes and fine-tuned for rhythm, morphology, diagnostic, and structural heart disease tasks. Models pretrained below 400,000 ECGs failed to consistently exceed controls without self-supervised pre-training, whereas 600,000 to 800,000 ECGs improved AUROC across tasks, suggesting a minimum threshold for effective ECG representation learning.