Associations of early retirement and mortality risk: a population-based study in Taiwan.

BACKGROUND: Early retirement is highly prevalent in Taiwan. This study assesses the association between early retirement and all-cause and cause-specific mortality risks while exploring the modifying effect of sociodemographic factors. METHODS: Using Taiwan's National Health Insurance Research Database between 2009 and 2019, 1 762 621 early retirees aged 45-64 and an equal number of employed comparators were included. The date and cause of death were identified using the National Death Registry. Cox regression models were used to estimate HRs of early retirement for all-cause mortality and cause-specific mortality. To explore modifying effects, we conducted subgroup analyses based on age groups, sexes, occupation types and general health status (Charlson Comorbid Index score). RESULTS: The analysis revealed that early retirees, compared with their concurrently employed counterparts, had a higher mortality risk (adjusted HR (aHR) 1.69, 95% CI (1.67 to 1.71)). Specifically, younger individuals (aged 45-54) (aHR 2.74 (95% CI 2.68 to 2.80)), males (aHR 1.78 (95% CI 1.76 to 1.81)), those in farming or fishing occupations (aHR 2.13 (95% CI 2.06 to 2.21)) or the private sector (aHR 1.92 (95% CI 1.89 to 1.96)), and those with the poorest health conditions (aHR 1.79 (95% CI 1.76 to 1.83)) had higher mortality risks of early retirement. Regarding specific causes of death, the top three highest risks were associated with gastrointestinal disorders, followed by suicide and neurological disorders. CONCLUSIONS: This study underscores the substantial mortality risk increase linked to early retirement, emphasising the importance of policy considerations, particularly regarding vulnerable populations and specific causes of death potentially linked to unhealthy lifestyles.

Triplet-branch network with contrastive prior-knowledge embedding for disease grading.

Since different disease grades require different treatments from physicians, i.e., the low-grade patients may recover with follow-up observations whereas the high-grade may need immediate surgery, the accuracy of disease grading is pivotal in clinical practice. In this paper, we propose a Triplet-Branch Network with ContRastive priOr-knoWledge embeddiNg (TBN-CROWN) for the accurate disease grading, which enables physicians to accordingly take appropriate treatments. Specifically, our TBN-CROWN has three branches, which are implemented for representation learning, classifier learning and grade-related prior-knowledge learning, respectively. The former two branches deal with the issue of class-imbalanced training samples, while the latter one embeds the grade-related prior-knowledge via a novel auxiliary module, termed contrastive embedding module. The proposed auxiliary module takes the features embedded by different branches as input, and accordingly constructs positive and negative embeddings for the model to deploy grade-related prior-knowledge via contrastive learning. Extensive experiments on our private and two publicly available disease grading datasets show that our TBN-CROWN can effectively tackle the class-imbalance problem and yield a satisfactory grading accuracy for various diseases, such as fatigue fracture, ulcerative colitis, and diabetic retinopathy.

Cross-Modal Vertical Federated Learning for MRI Reconstruction.

Federated learning enables multiple hospitals to cooperatively learn a shared model without privacy disclosure. Existing methods often take a common assumption that the data from different hospitals have the same modalities. However, such a setting is difficult to fully satisfy in practical applications, since the imaging guidelines may be different between hospitals, which makes the number of individuals with the same set of modalities limited. To this end, we formulate this practical-yet-challenging cross-modal vertical federated learning task, in which data from multiple hospitals have different modalities with a small amount of multi-modality data collected from the same individuals. To tackle such a situation, we develop a novel framework, namely Federated Consistent Regularization constrained Feature Disentanglement (Fed-CRFD), for boosting MRI reconstruction by effectively exploring the overlapping samples (i.e., same patients with different modalities at different hospitals) and solving the domain shift problem caused by different modalities. Particularly, our Fed-CRFD involves an intra-client feature disentangle scheme to decouple data into modality-invariant and modality-specific features, where the modality-invariant features are leveraged to mitigate the domain shift problem. In addition, a cross-client latent representation consistency constraint is proposed specifically for the overlapping samples to further align the modality-invariant features extracted from different modalities. Hence, our method can fully exploit the multi-source data from hospitals while alleviating the domain shift problem. Extensive experiments on two typical MRI datasets demonstrate that our network clearly outperforms state-of-the-art MRI reconstruction methods. The source code is available at https://github.com/IAMJackYan/FedCRFD.

Peroxygenase-Catalysed Sulfoxidations in Non-Aqueous Media.

Chiral sulfoxides are valuable building blocks in asymmetric synthesis. However, the biocatalytic synthesis of chiral sulfoxides is still challenged by low product titres. Herein, we report the use of peroxygenase as a catalyst for asymmetric sulfoxidation under non-aqueous conditions. Upon covalent immobilisation, the peroxygenase showed stability and activity under neat reaction conditions. A large variety of sulfides was converted into chiral sulfoxides in very high product concentration with moderate to satisfactory optical purity (e. g. 626 mM of (R)-methyl phenyl sulfoxide in approx. 89 % ee in 48 h). Further polishing of the ee value via cascading methionine reductase A (MsrA) gave>99 % ee of the sulfoxide. The robustness of the enzymes and high product titer is superior to the state-of-the-art methodologies. Gram-scale synthesis has been demonstrated. Overall, we demonstrated a practical and facile catalytic method to synthesize chiral sulfoxides.

Automatic view plane prescription for cardiac magnetic resonance imaging via supervision by spatial relationship between views.

BACKGROUND: View planning for the acquisition of cardiac magnetic resonance (CMR) imaging remains a demanding task in clinical practice. PURPOSE: Existing approaches to its automation relied either on an additional volumetric image not typically acquired in clinic routine, or on laborious manual annotations of cardiac structural landmarks. This work presents a clinic-compatible, annotation-free system for automatic CMR view planning. METHODS: The system mines the spatial relationship-more specifically, locates the intersecting lines-between the target planes and source views, and trains U-Net-based deep networks to regress heatmaps defined by distances from the intersecting lines. On the one hand, the intersection lines are the prescription lines prescribed by the technologists at the time of image acquisition using cardiac landmarks, and retrospectively identified from the spatial relationship. On the other hand, as the spatial relationship is self-contained in properly stored data, for example, in the DICOM format, the need for additional manual annotation is eliminated. In addition, the interplay of the multiple target planes predicted in a source view is utilized in a stacked hourglass architecture consisting of repeated U-Net-style building blocks to gradually improve the regression. Then, a multiview planning strategy is proposed to aggregate information from the predicted heatmaps for all the source views of a target plane, for a globally optimal prescription, mimicking the similar strategy practiced by skilled human prescribers. For performance evaluation, the retrospectively identified planes prescribed by the technologists are used as the ground truth, and the plane angle differences and localization distances between the planes prescribed by our system and the ground truth are compared. RESULTS: The retrospective experiments include 181 clinical CMR exams, which are randomly split into training, validation, and test sets in the ratio of 64:16:20. Our system yields the mean angular difference and point-to-plane distance of 5.68 ∘ $^\circ$ and 3.12 mm, respectively, on the held-out test set. It not only achieves superior accuracy to existing approaches including conventional atlas-based and newer deep-learning-based in prescribing the four standard CMR planes but also demonstrates prescription of the first cardiac-anatomy-oriented plane(s) from the body-oriented scout. CONCLUSIONS: The proposed system demonstrates accurate automatic CMR view plane prescription based on deep learning on properly archived data, without the need for further manual annotation. This work opens a new direction for automatic view planning of anatomy-oriented medical imaging beyond CMR.

MRL-Seg: Overcoming Imbalance in Medical Image Segmentation With Multi-Step Reinforcement Learning.

Medical image segmentation is a critical task for clinical diagnosis and research. However, dealing with highly imbalanced data remains a significant challenge in this domain, where the region of interest (ROI) may exhibit substantial variations across different slices. This presents a significant hurdle to medical image segmentation, as conventional segmentation methods may either overlook the minority class or overly emphasize the majority class, ultimately leading to a decrease in the overall generalization ability of the segmentation results. To overcome this, we propose a novel approach based on multi-step reinforcement learning, which integrates prior knowledge of medical images and pixel-wise segmentation difficulty into the reward function. Our method treats each pixel as an individual agent, utilizing diverse actions to evaluate its relevance for segmentation. To validate the effectiveness of our approach, we conduct experiments on four imbalanced medical datasets, and the results show that our approach surpasses other state-of-the-art methods in highly imbalanced scenarios. These findings hold substantial implications for clinical diagnosis and research.

Interannual evolution of the chemical composition, sources and processes of PM2.5 in Chengdu, China: Insights from observations in four winters.

The air quality in China has improved significantly in the last decade and, correspondingly, the characteristics of PM2.5 have also changed. We studied the interannual variation of PM2.5 in Chengdu, one of the most heavily polluted megacities in southwest China, during the most polluted season (winter). Our results show that the mass concentrations of PM2.5 decreased significantly year-by-year, from 195.8 ± 91.0 µg/m3 in winter 2016 to 96.1 ± 39.3 µg/m3 in winter 2020. The mass concentrations of organic matter (OM), SO42-, NH4+ and NO3- decreased by 49.6%, 57.1%, 49.7% and 28.7%, respectively. The differential reduction in the concentrations of chemical components increased the contributions from secondary organic carbon and NO3- and there was a larger contribution from mobile sources. The contribution of OM and NO3- not only increased with increasing levels of pollution, but also increased year-by-year at the same level of pollution. Four sources of PM2.5 were identified: combustion sources, vehicular emissions, dust and secondary aerosols. Secondary aerosols made the highest contribution and increased year-by-year, from 40.6% in winter 2016 to 46.3% in winter 2020. By contrast, the contribution from combustion sources decreased from 14.4% to 8.7%. Our results show the effectiveness of earlier pollution reduction policies and emphasizes that priority should be given to key pollutants (e.g., OM and NO3-) and sources (secondary aerosols and vehicular emissions) in future policies for the reduction of pollution in Chengdu during the winter months.

Nucleation Domains in Biomineralization: Biomolecular Sequence and Conformational Features.

Biomolecules play a vital role in the regulation of biomineralization. However, the characteristics of practical nucleation domains are still sketchy. Herein, the effects of the representative biomolecular sequence and conformations on calcium phosphate (Ca-P) nucleation and mineralization are investigated. The results of computer simulations and experiments prove that the line in the arrangement of dual acidic/essential amino acids with a single interval (Bc (Basic) -N (Neutral) -Bc-N-Ac (Acidic)- NN-Ac-N) is most conducive to the nucleation. 2α-helix conformation can best induce Ca-P ion cluster formation and nucleation. "Ac- × × × -Bc" sequences with α-helix are found to be the features of efficient nucleation domains, in which process, molecular recognition plays a non-negligible role. It further indicates that the sequence determines the potential of nucleation/mineralization of biomolecules, and conformation determines the ability of that during functional execution. The findings will guide the synthesis of biomimetic mineralized materials with improved performance for bone repair.

Ultralong-Term Durable Anticorrosive Coatings by Integration of Double-Layered Transfer Self-Healing Ability, Fe Ion-Responsive Ability, and Active/Passive Functional Partitioning.

The application of self-healing polymers in corrosion protection is often limited by their slow and nonautonomous healing ability and poor long-term durability. In this paper, we propose a double-layered transfer self-healing coating constructed by soft and rigid polymer layers. The soft polymer has a fast self-healing rate of 10 min to repair, which was found to accelerate the self-healing of the upper rigid layer. The rigid polymer provided relatively high barrier ability while preserving certain self-healing ability owing to the shear-thinning effect. In this way, the double-layered coating combined rapid self-healing (∼1 h) and high impedance modulus |Z|f-0.01 Hz of 2.58 × 1010 Ω·cm2. Furthermore, the introduction of pyridine groups in B-PEA and polyacrylate-grafted-polydimethylsiloxane (PEA-g-PDMS) induced the Fe ion-responsive ability and shortened the self-healing time to 40 min (100 ppm Fe). Finally, barrier and anode sacrificed layers were introduced to produce multilayered architecture with active/passive anticorrosion performance. In the presence of scratches, the |Z|f-0.01 Hz can be preserved at 1.03 × 1010 Ω·cm2 after 200 days. The created anticorrosive coating technology combines long-term durability with room temperature autonomous rapid self-healing capability, providing a broad prospect for anticorrosive applications.

Learning from multiple annotators for medical image segmentation.

Supervised machine learning methods have been widely developed for segmentation tasks in recent years. However, the quality of labels has high impact on the predictive performance of these algorithms. This issue is particularly acute in the medical image domain, where both the cost of annotation and the inter-observer variability are high. Different human experts contribute estimates of the "actual" segmentation labels in a typical label acquisition process, influenced by their personal biases and competency levels. The performance of automatic segmentation algorithms is limited when these noisy labels are used as the expert consensus label. In this work, we use two coupled CNNs to jointly learn, from purely noisy observations alone, the reliability of individual annotators and the expert consensus label distributions. The separation of the two is achieved by maximally describing the annotator's "unreliable behavior" (we call it "maximally unreliable") while achieving high fidelity with the noisy training data. We first create a toy segmentation dataset using MNIST and investigate the properties of the proposed algorithm. We then use three public medical imaging segmentation datasets to demonstrate our method's efficacy, including both simulated (where necessary) and real-world annotations: 1) ISBI2015 (multiple-sclerosis lesions); 2) BraTS (brain tumors); 3) LIDC-IDRI (lung abnormalities). Finally, we create a real-world multiple sclerosis lesion dataset (QSMSC at UCL: Queen Square Multiple Sclerosis Center at UCL, UK) with manual segmentations from 4 different annotators (3 radiologists with different level skills and 1 expert to generate the expert consensus label). In all datasets, our method consistently outperforms competing methods and relevant baselines, especially when the number of annotations is small and the amount of disagreement is large. The studies also reveal that the system is capable of capturing the complicated spatial characteristics of annotators' mistakes.

A Simple Access to γ- and ε-Keto Arenes via Enzymatic Divergent C─H Bond Oxyfunctionalization.

Performing divergent C─H bond functionalization on molecules with multiple reaction sites is a significant challenge in organic chemistry. Biocatalytic oxyfunctionalization reactions of these compounds to the corresponding ketones/aldehydes are typically hindered by selectivity issues. To address these challenges, the catalytic performance of oxidoreductases is explored. The results show that combining the peroxygenase-catalyzed propargylic C─H bond oxidation with the Old Yellow Enzyme-catalyzed reduction of conjugated C─C triple bonds in one-pot enables the regio- and chemoselective oxyfunctionalization of sp3 C─H bonds that are distant from benzylic sites. This enzymatic approach yielded a variety of γ-keto arenes with diverse structural and electronic properties in yields of up to 99% and regioselectivity of 100%, which are difficult to achieve using other chemocatalysis and enzymes. By adjusting the C─C triple bond, the carbonyl group's position can be further tuned to yield ε-keto arenes. This enzymatic approach can be combined with other biocatalysts to establish new synthetic pathways for accessing various challenging divergent C─H bond functionalization reactions.

A deep weakly semi-supervised framework for endoscopic lesion segmentation.

In the field of medical image analysis, accurate lesion segmentation is beneficial for the subsequent clinical diagnosis and treatment planning. Currently, various deep learning-based methods have been proposed to deal with the segmentation task. Albeit achieving some promising performances, the fully-supervised learning approaches require pixel-level annotations for model training, which is tedious and time-consuming for experienced radiologists to collect. In this paper, we propose a weakly semi-supervised segmentation framework, called Point Segmentation Transformer (Point SEGTR). Particularly, the framework utilizes a small amount of fully-supervised data with pixel-level segmentation masks and a large amount of weakly-supervised data with point-level annotations (i.e., annotating a point inside each object) for network training, which largely reduces the demand of pixel-level annotations significantly. To fully exploit the pixel-level and point-level annotations, we propose two regularization terms, i.e., multi-point consistency and symmetric consistency, to boost the quality of pseudo labels, which are then adopted to train a student model for inference. Extensive experiments are conducted on three endoscopy datasets with different lesion structures and several body sites (e.g., colorectal and nasopharynx). Comprehensive experimental results finely substantiate the effectiveness and the generality of our proposed method, as well as its potential to loosen the requirements of pixel-level annotations, which is valuable for clinical applications.

Ultracompact programmable inverse-designed nanophotonic devices based on digital subwavelength structures.

Inverse design is a powerful approach to achieve ultracompact nanophotonic devices. Here, we propose an ultracompact programmable near-infrared nanophotonic device platform to dynamically implement inverse-designed near-infrared devices with different functions by programming the state of the phase-change material filled in each pixel. By tuning PCM block by block, the subwavelength condition for inverse-designed ultracompact devices is satisfied with large tuning pixel size. Based on the inverse-design device platform with a footprint of 6.4µm×8µm, we design and theoretically demonstrate four power splitters with different split ratios and one mode multiplexer working in the near-infrared band. The average excess losses for the power splitters with ratios of 0:1,1:1, 2:1, and 3:1 are less than 0.82, 0.65, 0.82, and 1.03 dB over a wavelength span of 100 nm, respectively. Meanwhile, the insertion losses of the mode multiplexer are 1.4 and 2.5 dB for T E 0 and T E 1 mode, respectively, and the average crosstalk is less than -20 and -19d B, respectively. The five different devices could be configured online in a nonvolatile way by heating phase change materials with an off-chip laser, which may significantly enhance the flexibility of on-chip optical interconnections.

Efficient Osteogenic Activity of PEEK Surfaces Achieved by Femtosecond Laser-Hydroxylation.

Poly(etheretherketone) (PEEK) is regarded as an attractive orthopedic material because of its good biocompatibility and mechanical properties similar to natural bone. The efficient activation methods for the surfaces of PEEK matrix materials have become a hot research topic. In this study, a method using a femtosecond laser (FSL) followed by hydroxylation was developed to achieve efficient bioactivity. It produces microstructures, amorphous carbon, and grafted -OH groups on the PEEK surface to enhance hydrophilicity and surface energy. Both experimental and simulation results show that our modification leads to a superior ability to induce apatite deposition on the PEEK surface. The results also demonstrate that efficient grafting of C-OH through FSL-hydroxylation can effectively enhance cell proliferation and osteogenic differentiation compared to other modifications, thus improving osteogenic activity. Overall, FSL hydroxylation treatment is proved to be a simple, efficient, and environmentally friendly modification method for PEEK activation. It could expand the applications of PEEK in orthopedics, as well as promote the surface modification and structural design of other polymeric biomaterials to enhance bioactivity.

H2O2-driven enzymatic oxyfunctionalization of tertiary C-H bonds.

Peroxygenase from Agrocybe aegerita catalyses the selective hydroxylation of tertiary C-H bonds, whereby tertiary alcohols, diols, ketols, etc., were obtained in good to high regioselectivity and turnover numbers. This method can also be expanded for late-stage functionalization of drug molecules, which represents a streamlined synthetic method to give access to useful compounds.

Defective T-cell response to COVID-19 vaccination in acute myeloid leukaemia and myelodysplastic syndromes.

Limited data exist on COVID-19 vaccination efficacy in patients with acute myeloid leukemia and myelodysplasia with excess blasts (AML/MDS-EB2). We report results from a prospective study, PACE (Patients with AML and COVID-19 Epidemiology). 93 patients provided samples post-vaccine 2 or 3 (PV2, PV3). Antibodies against SARS-COV-2 spike antigen were detectable in all samples. Neutralization of the omicron variant was poorer than ancestral variants but improved PV3. In contrast, adequate T-cell reactivity to SARS-COV-2 spike protein was seen in only 16/47 (34%) patients PV2 and 23/52 (44%) PV3. Using regression models, disease response (not in CR/Cri), and increasing age predicted poor T cell response.

Self-healable and transparent PDMS-g-poly(fluorinated acrylate) coating with ultra-low ice adhesion strength for anti-icing applications.

The high ice adhesion strength (τ) and low adhesion of lubricant-free slippery polymers have restricted their applications. We synthesized polysiloxane-g-fluorinated acrylate polymer with a branched structure, anchored groups and dynamic cross-linked network, features imparting increased chain segment slipperiness and self-healability. The coating showed a low τ (6 kPa), strong adhesion and prolonged life.

Correction: Construction of durable superhydrophobic and anti-icing coatings via incorporating boroxine cross-linked silicone elastomers with good self-healability.

Correction for 'Construction of durable superhydrophobic and anti-icing coatings via incorporating boroxine cross-linked silicone elastomers with good self-healability' by Hengfei Liang et al., Soft Matter, 2022, https://doi.org/10.1039/d2sm01106a.

Construction of durable superhydrophobic and anti-icing coatings via incorporating boroxine cross-linked silicone elastomers with good self-healability.

The fragility of the micro-nano structure makes superhydrophobic coatings highly susceptible to stress, resulting in a decrease in their superhydrophobic and anti-icing performance. In this work, we proposed a new insight to improve durability by incorporating a thin layer of self-healable elastomer with a dynamic network on the micro-nano structure. We constructed superhydrophobic coatings (EP/SiO2/BFVSE) with a three-layered structure of the epoxy resin/silica nanoparticle/silicon elastomer. The silicon elastomer (BFVES) with a B-O dynamic cross-linked network and fluorinated moieties was synthesized by graft polymerization on vinyl silicon oil. The preparation route is facile and convenient for mass production. BFVES has rapid self-healing properties for scratches at room-temperature, underwater and at -18 °C. EP/SiO2/BFVSE preserved apparently higher CAs after being immersed in pH = 1, pH = 13, and NaCl solutions for 96 h as compared with the EP/SiO2 coating. In a water striking environment, the CA of EP/SiO2/BFVSE was slightly decreased to 153°. SEM images further reveal that the recovery of superhydrophobicity and icephobicity is attributed to the self-healing behavior of the boroxine-containing silicon elastomer. The EP/SiO2/BFVSE coating also possesses additional self-healing ability under chemical oxidation. The high durability of the self-healable superhydrophobic coating enables great application potential in aircraft, marine vessels, and outdoor facilities in harsh environments.

Femtosecond laser optimization of PEEK: efficient bioactivity achieved by synergistic surface chemistry and structures.

Poly-ether-ether-ketone (PEEK) is considered a potential orthopedic material due to the excellent mechanical properties and chemical resistance, but its biological inertness hampers its further clinical application. In this study, advanced femtosecond laser microfabrication technology was utilized to induce the change of the surface characteristics of PEEK to improve its bioactivity. Meanwhile, the mechanism of surface reaction and improved bioactivity was interpreted in detail from the perspective of material science. The surface physical-chemical characterization results showed that femtosecond laser etching could increase the surface energy, and the contents of active sites including amorphous carbon and carbon-hydroxyl on PEEK surfaces. In vitro validation experiments demonstrated that the samples etched with a femtosecond laser had a better ability to induce apatite deposition and cell proliferation than those treated with popular sulfonation modification, which would lead to better bioactivity and osteointegration. The current work fully presents the mechanism of the femtosecond laser low-temperature plasma effect on PEEK and the resulting surface characteristics, which could broaden the application of PEEK in the orthopedic field. Moreover, it has great potential in the surface design and modification of other biomaterials with enhanced bioactivity.