Hyunsu Lee

Background/Objectives: Colorectal cancer (CRC) remains one of the leading causes of cancer-related morbidity and mortality worldwide. Despite advances in treatment, outcomes for advanced CRC remain unsatisfactory due to uncontrolled proliferation, metastasis, and recurrence. This study investigated the anti-cancer effects of KMU-11342, an indolin-2-one-based multi-protein kinase inhibitor with previously reported anti-inflammatory properties, in human colorectal cancer models. Methods: The anti-cancer effects of KMU-11342 were evaluated in colorectal cancer cells and further investigated in three-dimensional (3D) spheroid and patient-derived organoid models. Cell proliferation, migration, apoptosis, and cell cycle progression were assessed. Kinase activity profiling and molecular docking analyses were performed to identify potential targets and characterize the underlying signaling pathways. Results: KMU-11342 significantly inhibited the proliferation and migration of CRC cells. It reduced CRC cell density by 58.9% and 83.3% at 0.5 and 1 μM, respectively. These effects were accompanied by G2/M cell cycle arrest and apoptotic cell death. In 3D models, spheroid formation was markedly reduced and stemness-related characteristics were diminished. Patient-derived CRC organoids also showed decreased viability, exhibiting 38.6% and 77.4% reductions at 1 and 2 μM, respectively. These effects were observed in a dose-dependent manner in both two-dimensional (2D) and 3D colorectal cancer models. Kinase activity profiling and molecular docking analyses identified glycogen synthase kinase 3 beta (GSK3β) and cyclin-dependent kinase 1 (CDK1) as potential mediators of the anti-cancer effects of KMU-11342 through the p53/nuclear factor kappa B (NF-κB) and FoxO1 signaling axes, respectively. Conclusions: KMU-11342 exhibits potent anti-tumor activity against CRC through suppressing proliferation, migration, and stemness in both 2D and 3D models, including patient-derived organoids. Its effects may be mediated, at least in part, through modulation of GSK3β and CDK1 via the p53/NF-κB and FoxO1 signaling pathways.

Background: Liver transplant recipients are highly susceptible to infectious complications due to surgical invasiveness and immunosuppressive therapy, and post-transplant bloodstream infection is associated with substantial morbidity and mortality. Although several prediction models for bloodstream infection have been proposed, most focus on emergency department or general ward populations and rely on black-box approaches. This limits their applicability and clinical interpretability in liver transplant settings. Therefore, this study aimed to develop predictive models for post-transplant bloodstream infection using preoperative and perioperative clinical data and to derive an interpretable risk equation through symbolic regression. Methods: We conducted a retrospective observational study including 245 adult liver transplant recipients treated at a single tertiary center. Clinical and laboratory variables were extracted from electronic medical records and analyzed using standard statistical methods. For prediction tasks, multiple conventional machine learning models were developed and compared with a symbolic regression-based model. Predictive performance and model interpretability were evaluated using discrimination metrics and Shapley Additive Explanations. Results: Post-transplant bloodstream infection occurred in 82 patients (33.4%). In the test set, conventional machine learning models showed modest discriminative performance (area under the curve, 0.53–0.64). The symbolic regression model achieved comparable discrimination (area under the curve, 0.63) while providing transparent, threshold-based risk equations. While conventional models primarily relied on laboratory variables, symbolic regression additionally identified perioperative clinical factors and viral serologic markers as important predictors. Discussion: Although overall predictive performance was modest, symbolic regression highlighted viral serologic markers as potential indicators of immunologic vulnerability, extending beyond standard laboratory predictors. Conclusions: This interpretability-focused approach may inform future risk stratification models incorporating richer perioperative data.

Background Rheumatoid arthritis (RA) is characterized by chronic inflammation and joint destructioncaused by dysregulated immune signaling. Despite the development of various protein kinaseinhibitors to treat inflammatory diseases such as RA, the results achieved are still notsatisfactory. Purpose This study aimed to investigate the therapeutic potential and biological mechanisms of KMU-11361, an innovative multi-protein kinase inhibitor which modulates the inflammatorypathways involved in the pathophysiology of RA, using both in vitro and in vivo models ofarthritis. Methods We investigated the effects of KMU-11361 on lipopolysaccharide (LPS)-inducedinflammation, receptor activator of NF-κB ligand (RANKL)-induced osteoclast differentiation,and osteoblast differentiation, using various assays in human rheumatoid arthritis (RA)-fibroblast-like synoviocytes (RA-FLS), THP-1 cells, RAW264.7 cells, T cells, and MC-3T3-E1 cells. Furthermore, we evaluated the toxicity and efficacy of KMU-11361 in a zebrafishmodel and a K/BxN serum transfer-induced arthritis mouse model. In addition, we performedkinase profiling and conducted molecular docking simulations to examine whether KMU-11361 interacts with MAPK1 and TAK1 involved in TLR4 signaling. Results Kinase profiling and molecular docking analyses revealed that KMU-11361 interacts with MAPK1 and TAK1, key regulators of TLR4 signaling. KMU-11361 markedly suppressedTLR4-mediated inflammatory signaling by reducing LPS-induced pro-inflammatory cytokineproduction, phosphorylation of TAK1 and MAPKs, NF-κB activation, and NLRP3 expressionin RA-FLS and THP-1 cells. Additionally, KMU-11361 inhibited RANKL-inducedosteoclastogenesis in RAW264.7 cells and attenuated T-cell activation-associated signalingpathways. Notably, KMU-11361 promoted osteoblastic differentiation in MC3T3-E1 cells. Inthe K/BxN serum transfer-induced arthritis mouse model, KMU-11361 significantly reducedankle swelling and IL-6 levels. Conclusion KMU-11361 exhibits potent anti-inflammatory and bone-protective properties by targeting keykinases involved in TLR4-mediated signaling and osteoclast differentiation. These findingsprovide mechanistic insight into its biological activity and highlight its therapeutic potential inboth in vitro and in vivo models of arthritis.

Stress is a critical factor that increases the risk of traffic accidents and should be addressed in driver monitoring and intelligent transportation systems. Electroencephalography (EEG) offers a direct, non-invasive measure of neural activity, making it a promising modality for real-time stress recognition. However, most existing studies have primarily focused on binary classification within constrained experimental settings, which limits reproducibility and practical relevance. In this study, the DEAP dataset was repurposed for driver stress detection via a Russell’s circumplex model–based label transformation, yielding three stress levels (no, low, high). Four state-of-the-art EEG models—EEGNet, EEGNeX, FBCNet, and TSception—were benchmarked under a subject-balanced stratified split strategy, with automated hyperparameter optimization using Optuna’s TPE sampler and Focal Loss to mitigate class imbalance. Experimental results revealed clear trade-offs among accuracy, inference latency, and computational complexity. FBCNet achieved the lowest inference latency (2.1 ms, 60,523 samples/s) with the smallest model footprint, whereas EEGNet demonstrated balanced performance with 78.8% accuracy. EEGNeX attained 81.6% accuracy at the cost of increased complexity, while TSception achieved the highest accuracy (98.9%) and strong generalization, confirming its suitability for EEG-based stress detection. By combining dataset relabeling, reproducible evaluation protocols, and comprehensive benchmarking, this study suggests TSception as the most effective model for EEG-based stress recognition and provides practical guidance for integrating EEG-driven stress monitoring into driver safety and intelligent transportation systems.

The hippocampus has historically been recognized as a critical neural substrate for the formation and retrieval of episodic memories. Concurrently, reinforcement learning (RL) has emerged as a foundational framework in both psychology and artificial intelligence for understanding how organisms and computational agents optimize behavior through experience. This Research Topic synthesizes these traditionally separate domains, revealing how hippocampal processes not only support memory functions but also fundamentally influence adaptive decision-making guided by reward and punishment signals.A central theme unifying the contributions in this collection is the recognition that hippocampal circuits do not merely store static representations of past experiences. Rather, they continuously integrate reward predictions, contextual information, and future-oriented simulations into dynamic memory representations. This integrative processing suggests that hippocampal-dependent memory consolidation and reinforcement-based learning strategies are deeply interconnected systems that mutually shape cognitive processes across multiple timescales. Tesler et al. (2024) tackle the intricacies of scaling from local synaptic processes to large-scale oscillatory phenomena relevant to learning. Their multiscale model of the CA1 region underscores why bridging cellular mechanisms with broader network activity is critical for simulating hippocampal contributions to brain-wide RL tasks.The notion that reinforcement signals can drive persistent adaptation is explored by Yoder et al. (2022), who demonstrate how biologically inspired RL feedback shapes central pattern generation within recurrent neural networks. By illustrating how neuromodulatory signals guide network plasticity over the lifespan, their study illuminates how hippocampal and related circuits maintain the capacity for behavioral recalibration.In considering offline replay and consolidation, Lee and Jung (2024) propose that hippocampal replay events parallel the “Dyna” RL approach. Their framework posits that CA3 simulates multiple scenarios while CA1 evaluates each for potential reward, thereby selectively strengthening connections tied to high-value outcomes. This reframes replay from a purely episodic recapitulation to a computational mechanism for optimizing future performance.Adaptation to contextual cues takes center stage in Kappel and Cheng (2024). Demonstrating that global remapping of place cell firing patterns facilitates the renewal of extinguished behaviors, they show that hippocampal circuits dynamically re-encode changing environments, allowing organisms to flexibly re-engage learned associations when contexts shift.Building on the idea of mental simulation, Kim and Lee (2025) present “Meta-Dyna,” a computational model that unites hippocampal replay with prefrontal meta-control. Hippocampal simulations rehearse potential outcomes, while prefrontal circuits arbitrate which simulations to adopt—thus enabling swift adaptation under dynamic or uncertain conditions. This synthesis of biologically inspired loops underscores how hippocampal and frontal circuits collaborate to bolster RL processes.Methodological advances in reinforcement-driven decoding are highlighted by Zhang et al. (2024), who describe an EEG-based attention model. While primarily centered on neural data analysis, the selective focus on reward-contingent signals echoes hippocampal processes, where significant stimuli and memories receive enhanced representation. This underscores the translational potential of RL in parsing real-time neural recordings for clinical or research applications.Finally, a comparative perspective is offered by Lochner et al. (2024), who apply RL principles to insect navigation in mushroom body–like circuits. The authors highlight parallels to hippocampal map-based strategies for spatial representation, suggesting a broad, cross-species principle where neural substrates fuse reward information with spatial cues to optimize navigational behavior.The studies in this Research Topic converge on several key themes that advance our understanding of hippocampal function in reinforcement-based learning. First is contextual flexibility, whereby hippocampal processing encodes environmental transitions and enables rapid retrieval of prior learning when contextual cues reemerge. Another prominent theme is offline replay and consolidation: through simulating possible scenarios, the hippocampus refines subsequent behavior, consistent with RL principles of planning and exploration. In addition, these articles highlight biologically plausible RL, revealing how neuromodulatory feedback in recurrent networks and mental simulations in prefrontal-hippocampal circuits can implement or inspire novel RL algorithms. Lastly, adaptive navigation emerges as a unifying aspect across species and neural systems, demonstrating how spatial representations informed by reinforcement signals guide goal-directed movement and decision-making.These insights carry significant implications for clinical contexts. Conditions that impair hippocampal function---such as Alzheimer’s disease, cognitive aging, or post-traumatic stress disorder---often disrupt both reward-based learning and contextual processing. Leveraging RL-based computational models to explore hippocampal plasticity may inform targeted therapies, from pharmacological interventions enhancing beneficial plasticity to cognitive rehabilitation exercises anchored in reward-driven strategies.By integrating theoretical perspectives, experimental investigations, and computational modeling, this Research Topic reframes the hippocampus as not merely a depot for episodic memories but an active driver of reinforcement-based decision-making and cognitive flexibility. Future research should continue weaving together neuroscience, machine learning, and comparative studies, forging comprehensive...

Background/Objectives: Anatomical variations in the biceps brachii muscle (BBM) are extremely frequent, leading to developmental and clinical implications. We studied the topography of the BBM and analyzed its correlations with other structures in the brachial region. Methods: A total of 103 cadaveric upper extremities were dissected. The length of the upper extremities was utilized as a reference line. The origin and insertion of the BBM, the coracobrachialis muscle (CBM), and the related neurovascular system were evaluated. Results: Each variable was calculated as a percentile and compared according to the presence of the considered variation; in particular, a supernumerary head of the BBM was found in 12/103 (11.65%) of upper extremities and was associated with a longer upper limb (506.25 ± 32.55 mm vs. 484.27 ± 30.41 mm, p = 0.022). When the variables were standardized by the length of the upper limb, the accessory head of the BBM was associated with the distal insertion point of the CBM (28.18 ± 3.54% vs. 30.59 ± 2.94%, p = 0.011) and BBM length (55.11 ± 2.17% vs. 58.18 ± 3.72%, p = 0.006). Other variables did not present significant differences with respect to the presence of the supernumerary head of the BBM. Conclusions: BBM variations may affect the topography of other structures, such as the length of the upper extremities, insertion of the CBM, and length of the BBM. Further studies are required to elucidate its clinical implications.