Citations

Abstract For implantable closed loop neuromodulation systems, real-time neural sensing, biomarker detection, and targeted stimulation are essential components for delivering personalized therapy. Current limitations in sampling rate and the number of available recording channels in implantable systems may pose significant challenges, especially in epilepsy research where high-frequency biomarkers and multi-site monitoring are essential. The Brain Interchange (BIC) system of CorTec is an externally powered, wireless implantable system that enables sensing and stimulation up to 32 channels at 1 kHz addressing the current limitations in the field. In this study, we present a modular closed loop neuromodulation environment developed for the BIC, enabling real-time data streaming, event detection, and flexible stimulation control. The framework integrates a Simulink-based processing pipeline, real-time communication, and a graphical user interface supporting both open and closed loop paradigms. System performance was evaluated through impedance measurement, latency assessment across three stimulation modes, and real-time closed loop stimulation using electrocardiogram (ECG) for in-lab demonstration. This modular environment provides a flexible platform for developing and validating neuromodulation algorithms and can be extended to incorporate advanced biomarker detection and multi-channel neural interfaces for future clinical research.

Abstract Asthma is a chronic inflammatory condition marked by episodic airway narrowing, with exercise-induced bronchoconstriction (EIB) a common trigger. Up to 70% of athletes experience EIB-related symptoms, yet current management relies on reactive treatment rather than early physiological monitoring. This study presents a proof-of-concept multimodal wearable platform integrating impedance pneumography (IPG), electrocardiography (ECG), and inertial measurement unit (IMU) sensing to detect early physiological signatures of EIB. Candidate biomarkers include elevated heart rate, increased respiratory rate, and reduced thoracic impedance amplitude. Real-time embedded algorithms perform on-device feature extraction and threshold-based decision logic without cloud dependence, while IMU data provide posture context for signal interpretation. The system was evaluated through subsystem validation and a resistive-breathing protocol simulating bronchoconstriction. Results demonstrate reliable detection of characteristic cardiorespiratory changes, establishing the feasibility of real-time wearable detection of early EIB markers and supporting future large-cohort validation and form-factor refinement.

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.

Abstract Brain arteriovenous malformations (AVMs) result from abnormal growth of vascular structures. AVMs pose a significant rupture risk, and its treatment is challenging due to the complex hemodynamics within the tangled nidus. Currently, investigating AVM hemodynamics remains a challenge due to the low resolution of clinical imaging modalities. In this study, a high-fidelity Computational Fluid Dynamics (CFD) simulation of a brain AVM from a KRAS-mutated mouse model is developed from micro-CT images. Brain AVMs were induced in mice through systemic delivery of AAV-BR1-CAG vector encoding KRASG12V-GFP transgene to cerebral endothelial cells. High-resolution microCT imaging at 6.551 𝜇𝑚 pixel size captured the complex vascular anatomy. The cerebral arteries were reconstructed using the 3D Slicer and Blender software. The curvilinear immersed boundary (CURVIB) method was employed with approximately 65 million grid points and spatial resolution of 0.7−2.3 𝜇𝑚. Our preliminary results reveal heterogeneous flow distribution through the nidus, with blood channeling through two primary pathways. The extremely low Reynolds number (0.038) indicates viscous-dominated flow characteristic of microcirculation. Our work presents the first high-resolution CFD simulation of brain AVM in the literature, providing a foundation for investigating flow-mediated vascular remodeling, rupture risk assessment, and therapeutic intervention planning in future.

Abstract Heart failure (HF) is a major cause of hospitalization and mortality worldwide. Ventricular assist devices (VADs) represent an alternative to cardiac transplantation for advanced HF. The main FDA-approved VAD (HeartMate III, Abbott), while effective, requires major cardiac surgery to insert the inflow cannula into the cardiac apex, and is associated with numerous co-morbidities and re-hospitalizations related to continuous flow and an external drive line and power supply. Mi-VAD, Inc. aspires to improve outcomes and quality of life with a miniature, minimally invasive, pulsatile, transcutaneously-powered VAD that is implanted above the aortic valve and/or the pulmonary valve and offering a wide range of hemodynamic support suitable for both advanced-stage and less advanced HF. Through its unique placement, there is the potential for right ventricular and biventricular cardiac assist for patients with restrictive cardiomyopathy and HFpEF for whom current VAD therapy is ineffective. Mi-VAD, Inc. has achieved proof of concept with its prototype, providing pulsatile ECG gated cardiac assist up to 10L/minute. Following initial bench top performance testing and in vivo efficacy and safety studies future work will focus on the development of wireless charging capability and conducting GLP builds for verification, regulatory clearance and commercialization. Additional technical development will focus on AI-driven hemodynamic optimization using multimodal physiologic sensing towards patient-specific digital twin of the Mi-VAD platform.

Abstract Accurate assessment of left-ventricular ejection fraction (EF) is a critical metric in diagnosing and managing heart failure; however, methods of assessment remain resource intensive, not widely available in low-acuity or remote settings and are intermittent in nature. This study explores the potential of a novel framework, cardiac electromechanical fingerprinting (cEMF) using synchronized electrocardiography (ECG) and seismocardiography (SCG) collected from a wearable sensing patch as a low-cost, non-invasive, continuous estimate of EF that can be used for rapid point of care screening. A machine learning pipeline was developed to process raw ECG and SCG data to derive EF. Signals were cleaned using band-pass filtering, NeuroKit2-based ECG cleaning and fiducial detection of critical electromechanical events. We extracted an ECG-SCG feature space consisting of heart-rate variability metrics, ECG morphological characterization, SCG timing intervals, spectral and entropy-based features. The top thirty most important features for EF were used to train and evaluate several regressor models. The best performing model was the Gaussian Process Regressor with an MAE of 5.43. Our findings demonstrate the promise of producing clinically relevant cardiac function measures, such as EF, from wearable electromechanical signals. Future work using larger datasets and complementary physiological signals such as phonocardiogram (PCG) may significantly strengthen EF% estimation and other cardiac functions for clinical applicability.

ABSTRACT Primary plasma cell leukemia (pPCL) is a rare and aggressive plasma cell disorder with inferior outcomes compared with multiple myeloma (MM). However, studies describing outcomes after first progression and subsequent lines of therapy remain limited. We retrospectively analyzed 104 patients with pPCL who progressed after first‐line therapy treated at one comprehensive cancer center between 2006 and 2024. Clinical characteristics, treatments, time to next treatment (TTNT), and overall survival (OS) were assessed using Kaplan–Meier methods. All pPCL patients received novel agent–based induction. At the time of analysis, 63% had died; median OS was 39 months. High‐risk cytogenetics were present in 71%. The median TTNT after first‐line therapy was 10 months. Symptomatic progression occurred in 29% of patients and was associated with significantly inferior subsequent OS (13 vs. 34 months) compared with those experiencing asymptomatic biochemical progression. The TTNT progressively shortened with subsequent lines of therapy (3rd line: 6 months; 4th line: 4 months; 5th line: 3 months). Notable exceptions included pPCL patients with t (11;14) treated with anti‐BCL‐2‐based regimens, who achieved a median TTNT of 13 months. Additionally, 24 patients received anti‐BCMA CAR T‐cell therapy. With a median follow‐up of 21 months, their median TTNT was 32 months. pPCL is characterized by frequent disease progression and diminishing response duration with successive therapeutic lines. Symptomatic progression portends particularly poor outcomes. Anti‐BCL‐2‐based therapy in t (11;14) disease and anti‐BCMA CAR T‐cell therapy appear to provide more durable disease control and represent important advances for the management of progressed pPCL.

Background Invasive micropapillary carcinoma (IMPC) of the breast is associated with a high propensity for lymphovascular invasion and lymph node metastasis. However, whether chemotherapy is associated with improved survival outcomes in these patients remains unclear. Methods This retrospective study analyzed IMPC patients from the SEER database (2010–2018) (n=1,150) and an external validation cohort from Nanjing Drum Tower Hospital (n=199). Propensity score matching assessed chemotherapy’s impact on overall survival (OS) and breast cancer-specific survival (BCSS) in the SEER cohort. A risk stratification model was developed and validated using the external cohort. Results In the SEER cohort, chemotherapy was associated with significantly improved OS (HR = 0.517, 95% CI 0.293-0.910, P = 0.020) but not BCSS. A risk model incorporating marital status, tumor size, subtype, and radiotherapy history identified high- and low-risk groups. Chemotherapy was associated with significantly better OS only for high-risk patients (internal training cohort: HR = 0.577, 95% CI 0.349-0.952, P = 0.029; internal validation cohort: HR = 0.213, 95% CI 0.070-0.647, P = 0.003). External validation consistently confirmed poorer prognosis for high-risk patients (HR = 2.969, 95% CI 1.028-8.571, P = 0.035) and verified the model’s ability to identify patients with differential OS associations with chemotherapy: chemotherapy was associated with better OS in high-risk patients (HR = 0.163, 95% CI 0.031-0.868, P = 0.016), but not in low-risk patients (HR = 0.793, 95% CI 0.153-4.103, P = 0.782). Conclusions Chemotherapy was associated with improved prognosis particularly in high-risk patients with IMPC. However, chemotherapy decisions should be made cautiously for low-risk patients, taking into account a comprehensive assessment of clinical and pathological factors.

Objective: To develop a predictive model for pathological complete response (pCR) after total neoadjuvant therapy (TNT) to inform selection for watch-and-wait (W/W). Summary Background Data: Patient selection for W/W after TNT for locally advanced rectal cancer remains challenging. Methods: An ensemble of tabular foundation models was fine-tuned in adults with clinical stage II or III microsatellite stable primary rectal adenocarcinoma undergoing TNT and total mesorectal excision (TNT+TME) from 2018-2023 to predict pCR, using pre-TNT, post-TNT and pre-TME variables. This model was externally validated on patients having TNT and W/W (TNT+W/W) to predict persistent clinical complete response (pcCR; the absence of local regrowth, distant metastases, or persistent near-cCR). Area under the receiver operator curve (AUROC), area under the precision-recall curve (AUPRC), and Brier score were calculated with 95% confidence intervals (CI). Results: Among 308 patients that underwent TNT+TME (median age 56; 40% female), the model predicted pCR with an AUROC 0.71 (95% CI 0.65-0.77), AUPRC 0.44 (95% CI 0.35-0.57), and was well calibrated with a Brier score of 0.17 (95% CI 0.15-0.20). At external validation in a cohort of 83 patients that are being managed with TNT+W/W (median age 57; 37% female), the model predicted pcCR with an AUROC 0.69 (95% CI 0.57-0.82), AUPRC 0.90 (95% CI 0.82-0.96), and Brier score of 0.30 (95% CI 0.26-0.33), improving to 0.17 with recalibration. Conclusions: This novel predictive model demonstrated good discrimination and calibration for pCR after TNT+TME with utility in TNT+W/W for pcCR after appropriate recalibration, supporting its application for W/W patient selection.