BACKGROUND: Chimeric antigen receptor (CAR)-T cell therapy has shown remarkable success in treating hematological malignancies. However, several challenges remain, including limited efficacy against solid tumors, T cell exhaustion, and lack of T cell persistence, which have restricted its clinical efficacy across various indications. Sequence optimization of CAR constructs offers a promising strategy for enhancing the therapeutic efficacy of CAR-T cells. Recent advances in machine learning, particularly in protein language models (PLMs), have enabled the prediction of mutational effects based on sequence representations. However, applying PLMs to CARs is challenging because of the artificial nature of CARs and the absence of comprehensive CAR sequence databases. RESULTS: We developed a computational framework for predicting CAR-T cell activity by fine-tuning ESM-2 with CAR sequences generated using sequence augmentation. The CAR sequences were constructed through the in silico recombination of the homologous domains of the CARs, enabling a task-specific adaptation of the model. To evaluate the prediction performance, we experimentally assessed the cytotoxicity of CAR-T cells expressing mutated CAR variants and compared the results with model predictions. Our results demonstrated that fine-tuning ESM-2 significantly improved the prediction performance of CAR-T cell activity. Furthermore, we showed that training parameters such as sequence diversity, number of training steps, and model size substantially influenced prediction performance. CONCLUSIONS: Our findings highlight the potential of combining sequence augmentation with fine-tuning of PLMs to advance data-driven CAR-T cell design.

This paper aims to enhance underwater visible light communication using a propeller LED transmitter (P-Tx) and receiver camera. Our goal is to improve the operational efficiency of remotely operated vehicles (ROVs) that survey ocean resources. P-Tx is a rotating device with blinking LEDs. By rotating P-Tx, we can generate a pseudo surface with successive light trails (light-trail surface). This technology allows the ROV to track the transmitting ROV and receive data simultaneously without mechanically adjusting the optical axis. However, with increasing distance between P-Tx and the receiver, the light gradually decreases in intensity from the edge of the light-trail surface, causing the receiver to fail to track accurately. This study proposes a robust transmitter detection method that utilizes the characteristics of the light-trail surface and improves the detecting accuracy of P-Tx on the receiving side. We show that the proposed method achieves higher transmitter detection accuracy than the conventional one.

This paper focuses on underwater visible light communication using a camera as a receiver to improve the operational efficiency of remotely operated vehicles that investigate ocean resources. This technology can realize wireless communication underwater, where wireless communication using radio waves is difficult. However, the optical axes between the transmitter and receiver must be correctly aligned before sending data in the underwater environment. We have developed a propeller LED transmitter (P-Tx) to solve this problem. The P-Tx is a rotating device with blinking LEDs. We can generate a pseudo-surface with successive light trails by rotating the P-Tx. This study calls this pseudo surface a "light-trail surface." The receiver can achieve robust transmitter detection by finding the light-trail surface using image processing without mechanical optical-axis alignment. This study also proposes a modulation method using light-trail surfaces. This study develops a prototype P-Tx and implements the proposed schemes. We then perform ISC experiments using the P-Tx and a camera to verify their communication performance.

BACKGROUND: Extracellular vesicle-derived (EV)-miRNAs have potential to serve as biomarkers for the diagnosis of various diseases. miRNA microarrays are widely used to quantify circulating EV-miRNA levels, and the preprocessing of miRNA microarray data is critical for analytical accuracy and reliability. Thus, although microarray data have been used in various studies, the effects of preprocessing have not been studied for Toray's 3D-Gene chip, a widely used measurement method. We aimed to evaluate batch effect, missing value imputation accuracy, and the influence of preprocessing on measured values in 18 different preprocessing pipelines for EV-miRNA microarray data from two cohorts with amyotrophic lateral sclerosis using 3D-Gene technology. RESULTS: Eighteen different pipelines with different types and orders of missing value completion and normalization were used to preprocess the 3D-Gene microarray EV-miRNA data. Notable results were suppressed in the batch effects in all pipelines using the batch effect correction method ComBat. Furthermore, pipelines utilizing missForest for missing value imputation showed high agreement with measured values. In contrast, imputation using constant values for missing data exhibited low agreement. CONCLUSIONS: This study highlights the importance of selecting the appropriate preprocessing strategy for EV-miRNA microarray data when using 3D-Gene technology. These findings emphasize the importance of validating preprocessing approaches, particularly in the context of batch effect correction and missing value imputation, for reliably analyzing data in biomarker discovery and disease research.

An energy saving control of unstable periodic orbits embedded in a hybrid chaotic system is proposed. The conventional controlling chaos methods utilize small perturbations of states or parameters as control input, however, quick time responses cannot be expected since the corresponding basins of attractions for higher periodic solutions become tiny. While If one allows a large perturbation to improve the time response, rather the total controlling energy which is proposed to the distance between the target orbit and the current orbit may increases. In this paper, when we consider the chaotic hybrid system, we noticed that we could utilize the perturbation of the referenced value for controlling, i.e., only a threshold value (Poincaré mapping surface) is updated in control. No control input as a perturbation of the state or parameter value is applied to the system. In fact, the threshold value is used instantly when the feedback system determines the next updated threshold value. The variation of the threshold value can be obtained numerically by computing variational equations, and the control matrix is designed with the linear control theory. Since no affection to the state and parameters, it is emphasized that the total behavior of the controlled system is different from the conventional methods, especially it is unlike the impulsive control methods. We demonstrate this control method in a simple hybrid system and show that a large basin of attraction for the control is realized.

We design oscillators controlled by optical input, which we call "LED fireflies," and study their synchronization phenomena. The most distinctive feature of LED fireflies is their collective behavior. The LED fireflies produce a huge variety of synchronous patterns. We demonstrate an optical art work that behaves like living entities by making use of their property.

In this paper, we present a kNN search method called the dual-heap kNN method, which is used in an embedded database management system (DBMS) for in-vehicle information systems. The dual-heap kNN method is based on two conventional kNN methods: (1) the RKV method and (2) the HS method. The RKV method and the HS method based on depth-first traversal and best-first traversal, respectively, shorten the search time. Our method not only shortens the search time but also reduces the capacity of the memory usage. Our simulation experiments suggest that our method results in the same number of disk accesses as that of the HS method, which is up to 12% smaller than the RKV method. Our method results in a memory usage capacity that is up to 24% larger than that of the RKV method and up to 68% smaller than that of the HS method. In addition, our prototype evaluation using actual data indicates that our method is applicable to in-vehicle information systems.

The authors have been developing humanoid robots in order to develop new mechanisms and functions for a humanoid robot that has the ability to communicate naturally with a human by expressing human-like emotion. We considered that human arms play an important role. And, in 2003, we developed the 9-DOFs emotion expression humanoid arm. Then, we developed the emotion expression humanoid robot WE-4R (Waseda Eye No.4 Refined) by integrating the new arms into the human-like head robot WE-4. Moreover, we introduced the new control algorithm for robot arm with redundant DOF into WE-4R. The new control algorithm was described by the junction of geometric constraint and robot emotion. We describe the mechanical design and the control algorithm of 9-DOFs emotion expression humanoid arm.

The authors have been developing human-like head robots in order to develop new head mechanisms and functions for a humanoid robot that has the ability to communicate naturally with a human by expressing human-like emotion. We believe that in the future, it will be necessary for personal robots to generate active behavior to interact bilaterally between human and robot. Therefore, we developed an emotion expression humanoid robot WE-4R (Waseda eye no.4 refined) in 2003. The need model consisting of the "appetite," the "need for security" and the "need for exploration" was introduced to the mental model for humanoid robots. We also defined the "need matrix" and introduced the "equations of need" to describe the robot needs. Robots with the need model can generate and express active behavior according to their need. This paper describes the need model that is implemented in the emotion expression humanoid robot WE-4R.

Recently, storage-centric configurations, such as NAS and SAN architectures, have attracted attention in advanced data processing. For these configurations, scalability, flexibility, and availability are key features, and central control is unsuitable. We propose autonomous disks to enable distributed control in the storage-centric configurations. Autonomous disks configure a cluster in a network, and implement the above key features using ECA rules with a distributed directory. The combination of the rules and directory also manages load distribution, and has a capability of reconfiguring the cluster automatically. We focus on the cluster reconfiguration, utilizing the skew handling mechanism. Autonomous disks adopt a multi-phase synchronization protocol with a dynamic coordinator for reconfiguration, apart from data migration. They allow the usual operations to be accepted during reconfiguration. We also report preliminary results of our experimental system. They indicate that synchronization cost is adequately acceptable.