Citations

Artificial intelligence (AI) is rapidly transforming healthcare through applications in clinical decision support, diagnostic imaging, population health management, and workflow optimization. Despite these advances, real-world deployment continues to expose critical challenges related to safety, bias, transparency, and integration into clinical workflows. Algorithmic bias can exacerbate health disparities, limited explainability may undermine clinician trust, and insufficient validation and post-deployment monitoring can compromise patient safety. Although the World Health Organization (WHO) has established six ethical principles for AI in health, including autonomy, well-being and safety, transparency, accountability, equity, and sustainability, translating these high-level principles into practical and enforceable governance mechanisms remains a persistent challenge. This narrative review synthesizes insights from bioethics, health policy, computer science, and clinical medicine to identify gaps in current AI governance approaches and proposes a lifecycle-aligned governance-by-design framework that operationalizes WHO ethical principles across key stages of the healthcare AI lifecycle, including data collection, model development, validation, deployment, and post-deployment monitoring. The framework integrates concrete governance mechanisms such as consent governance, fairness evaluation, external validation, explainability, clinician oversight, and continuous performance monitoring. Overall, this work advances a practical, lifecycle-integrated approach to AI governance and provides a structured foundation for developing safe, equitable, and trustworthy AI systems in healthcare.

Abstract Background Geometric uncertainties in high‐dose‐rate brachytherapy, including applicator displacement, catheter deflection, and interstitial needle spread, produce dose deviations that are patient and plan specific. Unlike external beam radiotherapy, where robustness evaluation frameworks are increasingly standardized, high dose rate brachytherapy lacks a normalized aggregate metric for quantifying plan sensitivity to geometric perturbations. Purpose To define a bounded, dimensionless robustness index for per‐fraction evaluation of high dose rate brachytherapy plans that is normalized to clinically meaningful dose‐limit constraints, incorporates dose‐dependent penalty sensitivity, and provides both a full‐scenario severity measure and a dose‐threshold‐filtered variant. Methods Geometric perturbation scenarios (translations, rotations, radial expansion/contraction, catheter deflection, and combined modes) are applied to source dwell positions. For each organ at risk, a convex‐transformed, constraint‐normalized penalty quantifies the fraction of the remaining margin to the dose‐limit constraint consumed per scenario. For target volumes, a prescription‐normalized piecewise penalty captures under‐coverage and excessive dose escalation. The Robustness Index is defined as one minus the mean scenario penalty, yielding a value in [0,1]. Results The framework produces structure‐independent, interpretable indices where a Robustness Index indicates that scenarios consume on average only 10% of the normalized remaining margin to the dose‐limit constraint. A threshold‐filtered variant evaluates robustness among only those scenarios that exceed a clinically specified dose level. Conclusions The proposed robustness index provides a standardized, clinically anchored metric for high dose rate brachytherapy plan evaluation under geometric uncertainty, complementing existing dose volume histogram based reporting with a single interpretable number per structure.

Abstract Purpose The temporarily implanted nitinol device (TIND) is a minimally invasive surgical technique (MIST) for benign prostatic hyperplasia (BPH) that promises fewer unwanted side effects compared with traditional bladder outlet procedures for small glands with high-riding bladder necks, including transurethral incision of the prostate (TUIP). This study sought to compare the effect of TIND vs. TUIP on ejaculatory function in sexually active males using the Male Sexual Health Questionnaire Ejaculatory Function Short Form (MSHQ-EjD-SF). Methods A retrospective review of a prospectively maintained database of all patients who underwent TIND at our institution between 1/1/2022 and 10/1/2024. Patients were excluded if they reported no postoperative ejaculations. Patients who underwent TUIP during the same period were compared. Age, prostate volume, PSA, and urinary parameters were collected. MSHQ-EjD-SF scores were collected for all patients. Continuous variables were compared using Wilcoxon Rank sum test and categorical variables using Fisher’s exact test. Results A total of 27 patients underwent TIND, of which 17 reported postoperative ejaculations and were compared with 17 patients who underwent TUIP during the same period with similar baseline age, prostate volume, and urinary parameters. Median MSHQ-EjD-SF score was higher after TIND compared with TUIP (15.0 (14.0, 15.0) vs. 13.0 (11.0, 13.0), p < 0.01), while bother score was lower (0.0 (0.0, 0.0) vs. 1.0 (0.0, 2.0), p = 0.02). Conclusions Patients experienced negligible effect on ejaculatory function after TIND compared with TUIP. This may be an effective bladder outlet procedure for patients who prioritize their postoperative ejaculatory function.

Abstract Hepatic spheroids are widely used in bioartificial liver (BAL) systems, hepatotoxicology, and pharmacology due to improved functional stability compared with monolayer cultures. However, many existing methods rely on specialized equipment or static platforms that limit scalability and dynamic culture applications. Primary rat hepatocytes were cultured as three‐dimensional spheroids using a rocker‐based suspension system. Spheroid formation was evaluated across a range of rocking speeds (7–18 cycles per minute [cpm]) and inoculation densities. Spheroid morphology was quantified over time using two‐dimensional imaging. Functional performance was assessed by urea and albumin production on days 1, 4, 7, and 10. Spheroid formation efficiency was evaluated using volumetric modeling. Mean spheroid diameter decreased from 199 ± 41 μm at 7 cpm to 80.2 ± 17.6 μm at 18 cpm. The spheroid area was significantly influenced by both culture day ( p < 0.0001) and rocker speed, with significant differences observed across speeds at all time points ( p < 0.001, days 1–7). Mean spheroid area at 10 cpm was 13,750 ± 9051 μm 2 (day 1), 26,730 ± 8658 μm 2 (day 4), and decreased to 22,526 ± 8705 μm 2 (day 7). Urea production differed across speeds (ANOVA, p = 0.043), with peak performance at 12 cpm (median 3.47 μg/μg DNA) compared to 7 cpm (median 0.11 μg/μg DNA). Albumin production was 1086.8 ng/μg DNA on day 1 and 1534.4 ng/μg DNA on day 10. Urea production was 0.62 μg/μg DNA on day 1 and 1.25 μg/μg DNA on day 10. Median hepatocyte incorporation was 86% (range 73%–116%) by day 4. This rocker‐based method provides a simple, low‐cost, and scalable approach for generating hepatic spheroids with tunable size distributions under dynamic conditions. The system supports progressive functional maturation and is compatible with established scale‐up strategies, supporting its potential for translational applications.

Abstract Pain assessment plays a critical role in clinical decision-making, yet conventional methods rely heavily on subjective self-reports such as the Visual Analog Scale (VAS) or the Numeric Rating Scale (NRS). These approaches fail in scenarios involving infants, sedated patients, unconscious individuals, or those with cognitive or communication impairments. To address this gap, this research proposes PainNet, an AI-driven, real-time facial pain recognition system that integrates deep learning-based object detection with Action Unit (AU)–based pain intensity scoring. The system uses YOLOv8 for rapid facial expression detection and incorporates Py-Feat to extract pain-relevant AUs, enabling computation of the Prkachin and Solomon Pain Intensity (PSPI) score. A balanced dataset comprising 500 images sourced from Delaware, BioVid, and RaFD was curated, annotated, and preprocessed for training. The YOLOv8-based binary pain vs. no-pain classifier achieved a mean Average Precision (mAP) of 0.89 and an F1-score of 0.80 on the test dataset, demonstrating strong performance in diverse environments and facial variations. The design emphasizes computational efficiency by activating AU analysis only when YOLOv8 detects probable pain, making PainNet suitable for continuous monitoring in ICUs, elder-care facilities, and telemedicine systems. This study demonstrates that the fusion of real-time object detection and clinically validated AU scoring can significantly advance automated, objective, and scalable pain assessment.

Abstract Interstitial lung disease (ILD) is a chronic, irreversible condition that is challenging to diagnose due to subtle symptoms and limitations of conventional tools. Noninvasive lung sound analysis using phonopulmography offers a promising alternative. This study evaluated the feasibility of a deep learning model for distinguishing normal and pathological lung sounds in ILD. Lung sound recordings were collected from 12 subjects (6 healthy, 6 ILD) and segmented beat-to-beat using simultaneously acquired ECG and phonopulmography, yielding 2,280 segments. Mel-frequency cepstral coefficients (MFCCs; 13 per segment) were extracted, and a one-dimensional CNN-based You Only Listen Once (YOLO) model was trained for segment classification. Evaluation used Leave-One-Out Cross-Validation, with patient-level classification by aggregating segment predictions via majority vote. On the test set (190 segments per class), the model achieved an average segment-level accuracy of 0.738 ± 0.087 and mean AUC of 0.850 ± 0.062. Precision, recall, and F1-score were 0.775 ± 0.148, 0.755 ± 0.183, and 0.738 ± 0.089, respectively. These preliminary results suggest that beat-level MFCCs with a 1D CNN-based YOLO model enable accurate segment- and patient-level classification. This feasibility study highlights the potential of AI-assisted auscultation for noninvasive monitoring and clinical decision support in ILD management, pending further validation in larger cohorts.