Adequate asparaginase is important to prevent central nervous system and testicular relapse of pediatric Philadelphia chromosome-negative B-cell acute lymphoblastic leukemia.

Asparaginase (Asp) is one of the most important drugs for treating acute lymphoblastic leukemia (ALL). However, off-protocol Asp administration (OPAA) or hypersensitivity may disturb its pharmacokinetic profile. In this retrospective study, we sought to determine whether OPAA and hypersensitivity to Escherichia coli asparaginase (E coli Asp) impaired extramedullary relapse prevention in a pediatric ALL cohort treated according to SCMC-ALL-2005 protocol from 2005 to 2014 at the Shanghai Children's Medical Center (SCMC). In total, 676 patients were enrolled in this study, including 369 with OPAA and 60 exhibiting hypersensitivity to E coli Asp. At the end of the most recent follow-up, 58 patients had extramedullary relapse. The 5-year cumulative extramedullary relapse incidence in patients with OPAA was 11.01%, whereas that in patients without OPAA was 5.28% (P = .0036). Moreover, the 5-year cumulative extramedullary relapse incidence in patients that exhibited hypersensitivity to E coli Asp was 16.48%, whereas that in patients without hypersensitivity was 7.59% (P = .0195). Concerning the relapse site, OPAA not only increased central nervous system (CNS) relapse but testicular relapse as well. Based on Fine and Gray multivariate analysis, OPAA and hypersensitivity to Asp were independent risk factors for extramedullary relapse. In conclusion, to prevent extramedullary relapse of ALL, adequate duration to administrate Asp was more important than the total dosage, and more attention should be paid to Asp inadequate due to hypersensitivity.

Lite It Fly: An All-Deformable-Butterfly Network.

Most deep neural networks (DNNs) consist fundamentally of convolutional and/or fully connected layers, wherein the linear transform can be cast as the product between a filter matrix and a data matrix obtained by arranging feature tensors into columns. Recently proposed deformable butterfly (DeBut) decomposes the filter matrix into generalized, butterfly-like factors, thus achieving network compression orthogonal to the traditional ways of pruning or low-rank decomposition. This work reveals an intimate link between DeBut and a systematic hierarchy of depthwise and pointwise convolutions, which explains the empirically good performance of DeBut layers. By developing an automated DeBut chain generator, we show for the first time the viability of homogenizing a DNN into all DeBut layers, thus achieving extreme sparsity and compression. Various examples and hardware benchmarks verify the advantages of All-DeBut networks. In particular, we show it is possible to compress a PointNet to 5% parameters with 5% accuracy drop, a record not achievable by other compression schemes.

Design optimization and experimental study of a novel mechanism for a hover-able bionic flapping-wing micro air vehicle.

Allomyrina dichotomahas a natural ultra-high flying ability and maneuverability. Especially its ability to fly flexibly in the air, makes it more adaptable to the harsh ecological environment. In this study, a bionic flapping-wing micro air vehicle (FMAV) is designed and fabricated by mimicking the flight mode ofA. dichotoma. Parametric design was employed for combining the airframe structure and flight characteristics analysis. To improve the transmission efficiency and compactness of the FMAV mechanisms, this study first analyses the body structure ofA. dichotoma, and then proposes a novel mechanism of FMAV based on its biological motion characteristics, the flight motion characteristics, and its musculoskeletal system. By optimizing the flapping-wing mechanism and mimicking the flying mechanism ofA. dichotoma, the large angle amplitude and the high-frequency flapping motion can be achieved to generate more aerodynamic force. Meanwhile, to improve the bionic effect and the wing performance of FMAV, the flexible deformation ofA. dichotomawings for each flapping period was observed by a high-speed camera. Furthermore, the bionic design of wings the prototype was carried out, therefore the wings can generate a high lift force in the flapping process. The experiment demonstrated that the aircraft can achieve a flapping angle of 160 degrees and 30 Hz flapping frequency. The attitude change of FMAV is realized by mimicking the movement for the change of attitude of theA. dichotoma, by changing the angle of attack of the wing, and executing the flight action of multiple degrees of freedom including pitch, roll and yaw. Finally, the aerodynamic experiment demonstrated that the prototype can offer 27.8 g lift and enough torque for altitude adjustment.