Insertion of a synthetic switch into insulin provides metabolite-dependent regulation of hormone-receptor activation.

Insulin-signaling requires conformational change: whereas the free hormone and its receptor each adopt autoinhibited conformations, their binding leads to structural reorganization. To test the functional coupling between insulin's "hinge opening" and receptor activation, we inserted an artificial ligand-dependent switch into the insulin molecule. Ligand-binding disrupts an internal tether designed to stabilize the hormone's native closed and inactive conformation, thereby enabling productive receptor engagement. This scheme exploited a diol sensor (meta-fluoro-phenylboronic acid at GlyA1) and internal diol (3,4-dihydroxybenzoate at LysB28). The sensor recognizes monosaccharides (fructose > glucose). Studies of insulin-signaling in human hepatoma-derived cells (HepG2) demonstrated fructose-dependent receptor autophosphorylation leading to appropriate downstream signaling events, including a specific kinase cascade and metabolic gene regulation (gluconeogenesis and lipogenesis). Addition of glucose (an isomeric ligand with negligible sensor affinity) did not activate the hormone. Similarly, metabolite-regulated signaling was not observed in control studies of 1) an unmodified insulin analog or 2) an analog containing a diol sensor without internal tethering. Although secondary structure (as probed by circular dichroism) was unaffected by ligand-binding, heteronuclear NMR studies revealed subtle local and nonlocal monosaccharide-dependent changes in structure. Insertion of a synthetic switch into insulin has thus demonstrated coupling between hinge-opening and allosteric holoreceptor signaling. In addition to this foundational finding, our results provide proof of principle for design of a mechanism-based metabolite-responsive insulin. In particular, replacement of the present fructose sensor by an analogous glucose sensor may enable translational development of a "smart" insulin analog to mitigate hypoglycemic risk in diabetes therapy.

Evolution of insulin at the edge of foldability and its medical implications.

Proteins have evolved to be foldable, and yet determinants of foldability may be inapparent once the native state is reached. Insight has emerged from studies of diseases of protein misfolding, exemplified by monogenic diabetes mellitus due to mutations in proinsulin leading to endoplasmic reticulum stress and ß-cell death. Cellular foldability of human proinsulin requires an invariant Phe within a conserved crevice at the receptor-binding surface (position B24). Any substitution, even related aromatic residue TyrB24, impairs insulin biosynthesis and secretion. As a seeming paradox, a monomeric TyrB24 insulin analog exhibits a native-like structure in solution with only a modest decrement in stability. Packing of TyrB24 is similar to that of PheB24, adjoining core cystine B19-A20 to seal the core; the analog also exhibits native self-assembly. Although affinity for the insulin receptor is decreased ∼20-fold, biological activities in cells and rats were within the range of natural variation. Together, our findings suggest that the invariance of PheB24 among vertebrate insulins and insulin-like growth factors reflects an essential role in enabling efficient protein folding, trafficking, and secretion, a function that is inapparent in native structures. In particular, we envision that the para-hydroxyl group of TyrB24 hinders pairing of cystine B19-A20 in an obligatory on-pathway folding intermediate. The absence of genetic variation at B24 and other conserved sites near this disulfide bridge-excluded due to ß-cell dysfunction-suggests that insulin has evolved to the edge of foldability. Nonrobustness of a protein's fitness landscape underlies both a rare monogenic syndrome and "diabesity" as a pandemic disease of civilization.

Brain age vector: A measure of brain aging with enhanced neurodegenerative disorder specificity.

Neuroimaging-driven brain age estimation has become popular in measuring brain aging and identifying neurodegenerations. However, the single estimated brain age (gap) compromises regional variations of brain aging, losing spatial specificity across diseases which is valuable for early screening. In this study, we combined brain age modeling with Shapley Additive Explanations to measure brain aging as a feature contribution vector underlying spatial pathological aging mechanism. Specifically, we regressed age with volumetric brain features using machine learning to construct the brain age model, and model-agnostic Shapley values were calculated to attribute regional brain aging for each subject's age estimation, forming the brain age vector. Spatial specificity of the brain age vector was evaluated among groups of normal aging, prodromal Parkinson disease (PD), stable mild cognitive impairment (sMCI), and progressive mild cognitive impairment (pMCI). Machine learning methods were adopted to examine the discriminability of the brain age vector in early disease screening, compared with the other two brain aging metrics (single brain age gap, regional brain age gaps) and brain volumes. Results showed that the proposed brain age vector accurately reflected disorder-specific abnormal aging patterns related to the medial temporal and the striatum for prodromal AD (sMCI vs. pMCI) and PD (healthy controls [HC] vs. prodromal PD), respectively, and demonstrated outstanding performance in early disease screening, with area under the curves of 83.39% and 72.28% in detecting pMCI and prodromal PD, respectively. In conclusion, the proposed brain age vector effectively improves spatial specificity of brain aging measurement and enables individual screening of neurodegenerative diseases.

"Register-shift" insulin analogs uncover constraints of proteotoxicity in protein evolution.

Globular protein sequences encode not only functional structures (the native state) but also protein foldability, i.e. a conformational search that is both efficient and robustly minimizes misfolding. Studies of mutations associated with toxic misfolding have yielded insights into molecular determinants of protein foldability. Of particular interest are residues that are conserved yet dispensable in the native state. Here, we exploited the mutant proinsulin syndrome (a major cause of permanent neonatal-onset diabetes mellitus) to investigate whether toxic misfolding poses an evolutionary constraint. Our experiments focused on an invariant aromatic motif (PheB24-PheB25-TyrB26) with complementary roles in native self-assembly and receptor binding. A novel class of mutations provided evidence that insulin can bind to the insulin receptor (IR) in two different modes, distinguished by a "register shift" in this motif, as visualized by molecular dynamics (MD) simulations. Register-shift variants are active but defective in cellular foldability and exquisitely susceptible to fibrillation in vitro Indeed, expression of the corresponding proinsulin variant induced endoplasmic reticulum stress, a general feature of the mutant proinsulin syndrome. Although not present among vertebrate insulin and insulin-like sequences, a prototypical variant ([GlyB24]insulin) was as potent as WT insulin in a rat model of diabetes. Although in MD simulations the shifted register of receptor engagement is compatible with the structure and allosteric reorganization of the IR-signaling complex, our results suggest that this binding mode is associated with toxic misfolding and so is disallowed in evolution. The implicit threat of proteotoxicity limits sequence variation among vertebrate insulins and insulin-like growth factors.

Solution structure of an ultra-stable single-chain insulin analog connects protein dynamics to a novel mechanism of receptor binding.

Domain-minimized insulin receptors (IRs) have enabled crystallographic analysis of insulin-bound "micro-receptors." In such structures, the C-terminal segment of the insulin B chain inserts between conserved IR domains, unmasking an invariant receptor-binding surface that spans both insulin A and B chains. This "open" conformation not only rationalizes the inactivity of single-chain insulin (SCI) analogs (in which the A and B chains are directly linked), but also suggests that connecting (C) domains of sufficient length will bind the IR. Here, we report the high-resolution solution structure and dynamics of such an active SCI. The hormone's closed-to-open transition is foreshadowed by segmental flexibility in the native state as probed by heteronuclear NMR spectroscopy and multiple conformer simulations of crystallographic protomers as described in the companion article. We propose a model of the SCI's IR-bound state based on molecular-dynamics simulations of a micro-receptor complex. In this model, a loop defined by the SCI's B and C domains encircles the C-terminal segment of the IR α-subunit. This binding mode predicts a conformational transition between an ultra-stable closed state (in the free hormone) and an active open state (on receptor binding). Optimization of this switch within an ultra-stable SCI promises to circumvent insulin's complex global cold chain. The analog's biphasic activity, which serendipitously resembles current premixed formulations of soluble insulin and microcrystalline suspension, may be of particular utility in the developing world.

An ultra-stable single-chain insulin analog resists thermal inactivation and exhibits biological signaling duration equivalent to the native protein.

Thermal degradation of insulin complicates its delivery and use. Previous efforts to engineer ultra-stable analogs were confounded by prolonged cellular signaling in vivo, of unclear safety and complicating mealtime therapy. We therefore sought an ultra-stable analog whose potency and duration of action on intravenous bolus injection in diabetic rats are indistinguishable from wild-type (WT) insulin. Here, we describe the structure, function, and stability of such an analog, a 57-residue single-chain insulin (SCI) with multiple acidic substitutions. Cell-based studies revealed native-like signaling properties with negligible mitogenic activity. Its crystal structure, determined as a novel zinc-free hexamer at 2.8 Å, revealed a native insulin fold with incomplete or absent electron density in the C domain; complementary NMR studies are described in the accompanying article. The stability of the analog (ΔGU 5.0(±0.1) kcal/mol at 25 °C) was greater than that of WT insulin (3.3(±0.1) kcal/mol). On gentle agitation, the SCI retained full activity for >140 days at 45 °C and >48 h at 75 °C. These findings indicate that marked resistance to thermal inactivation in vitro is compatible with native duration of activity in vivo Further, whereas WT insulin forms large and heterogeneous aggregates above the standard 0.6 mm pharmaceutical strength, perturbing the pharmacokinetic properties of concentrated formulations, dynamic light scattering, and size-exclusion chromatography revealed only limited SCI self-assembly and aggregation in the concentration range 1-7 mm Such a combination of favorable biophysical and biological properties suggests that SCIs could provide a global therapeutic platform without a cold chain.

Human Sex Determination at the Edge of Ambiguity: INHERITED XY SEX REVERSAL DUE TO ENHANCED UBIQUITINATION AND PROTEASOMAL DEGRADATION OF A MASTER TRANSCRIPTION FACTOR.

A general problem is posed by analysis of transcriptional thresholds governing cell fate decisions in metazoan development. A model is provided by testis determination in therian mammals. Its key step, Sertoli cell differentiation in the embryonic gonadal ridge, is initiated by SRY, a Y-encoded architectural transcription factor. Mutations in human SRY cause gonadal dysgenesis leading to XY female development (Swyer syndrome). Here, we have characterized an inherited mutation compatible with either male or female somatic phenotypes as observed in an XY father and XY daughter, respectively. The mutation (a crevice-forming substitution at a conserved back surface of the SRY high mobility group box) markedly destabilizes the domain but preserves specific DNA affinity and induced DNA bend angle. On transient transfection of diverse human and rodent cell lines, the variant SRY exhibited accelerated proteasomal degradation (relative to wild type) associated with increased ubiquitination; in vitro susceptibility to ubiquitin-independent ("default") cleavage by the 20S core proteasome was unchanged. The variant's gene regulatory activity (as assessed in a cellular model of the rat embryonic XY gonadal ridge) was reduced by 2-fold relative to wild-type SRY at similar levels of mRNA expression. Chemical proteasome inhibition restored native-like SRY expression and transcriptional activity in association with restored occupancy of a sex-specific enhancer element in principal downstream gene Sox9, demonstrating that the variant SRY exhibits essentially native activity on a per molecule basis. Our findings define a novel mechanism of impaired organogenesis, accelerated ubiquitin-directed proteasomal degradation of a master transcription factor leading to a developmental decision poised at the edge of ambiguity.

Protective hinge in insulin opens to enable its receptor engagement.

Insulin provides a classical model of a globular protein, yet how the hormone changes conformation to engage its receptor has long been enigmatic. Interest has focused on the C-terminal B-chain segment, critical for protective self-assembly in ß cells and receptor binding at target tissues. Insight may be obtained from truncated "microreceptors" that reconstitute the primary hormone-binding site (α-subunit domains L1 and αCT). We demonstrate that, on microreceptor binding, this segment undergoes concerted hinge-like rotation at its B20-B23 ß-turn, coupling reorientation of Phe(B24) to a 60° rotation of the B25-B28 ß-strand away from the hormone core to lie antiparallel to the receptor's L1-ß2 sheet. Opening of this hinge enables conserved nonpolar side chains (Ile(A2), Val(A3), Val(B12), Phe(B24), and Phe(B25)) to engage the receptor. Restraining the hinge by nonstandard mutagenesis preserves native folding but blocks receptor binding, whereas its engineered opening maintains activity at the price of protein instability and nonnative aggregation. Our findings rationalize properties of clinical mutations in the insulin family and provide a previously unidentified foundation for designing therapeutic analogs. We envisage that a switch between free and receptor-bound conformations of insulin evolved as a solution to conflicting structural determinants of biosynthesis and function.

Biophysical optimization of a therapeutic protein by nonstandard mutagenesis: studies of an iodo-insulin derivative.

Insulin provides a model for the therapeutic application of protein engineering. A paradigm in molecular pharmacology was defined by design of rapid-acting insulin analogs for the prandial control of glycemia. Such analogs, a cornerstone of current diabetes regimens, exhibit accelerated subcutaneous absorption due to more rapid disassembly of oligomeric species relative to wild-type insulin. This strategy is limited by a molecular trade-off between accelerated disassembly and enhanced susceptibility to degradation. Here, we demonstrate that this trade-off may be circumvented by nonstandard mutagenesis. Our studies employed Lys(B28), Pro(B29)-insulin ("lispro") as a model prandial analog that is less thermodynamically stable and more susceptible to fibrillation than is wild-type insulin. We have discovered that substitution of an invariant tyrosine adjoining the engineered sites in lispro (Tyr(B26)) by 3-iodo-Tyr (i) augments its thermodynamic stability (ΔΔGu 0.5 ± 0.2 kcal/mol), (ii) delays onset of fibrillation (lag time on gentle agitation at 37 °C was prolonged by 4-fold), (iii) enhances affinity for the insulin receptor (1.5 ± 0.1-fold), and (iv) preserves biological activity in a rat model of diabetes mellitus. (1)H NMR studies suggest that the bulky iodo-substituent packs within a nonpolar interchain crevice. Remarkably, the 3-iodo-Tyr(B26) modification stabilizes an oligomeric form of insulin pertinent to pharmaceutical formulation (the R6 zinc hexamer) but preserves rapid disassembly of the oligomeric form pertinent to subcutaneous absorption (T6 hexamer). By exploiting this allosteric switch, 3-iodo-Tyr(B26)-lispro thus illustrates how a nonstandard amino acid substitution can mitigate the unfavorable biophysical properties of an engineered protein while retaining its advantages.

Mapping Multi-Modal Brain Connectome for Brain Disorder Diagnosis via Cross-Modal Mutual Learning.

Recently, the study of multi-modal brain connectome has recorded a tremendous increase and facilitated the diagnosis of brain disorders. In this paradigm, functional and structural networks, e.g., functional and structural connectivity derived from fMRI and DTI, are in some manner interacted but are not necessarily linearly related. Accordingly, there remains a great challenge to leverage complementary information for brain connectome analysis. Recently, Graph Convolutional Networks (GNN) have been widely applied to the fusion of multi-modal brain connectome. However, most existing GNN methods fail to couple inter-modal relationships. In this regard, we propose a Cross-modal Graph Neural Network (Cross-GNN) that captures inter-modal dependencies through dynamic graph learning and mutual learning. Specifically, the inter-modal representations are attentively coupled into a compositional space for reasoning inter-modal dependencies. Additionally, we investigate mutual learning in explicit and implicit ways: (1) Cross-modal representations are obtained by cross-embedding explicitly based on the inter-modal correspondence matrix. (2) We propose a cross-modal distillation method to implicitly regularize latent representations with cross-modal semantic contexts. We carry out statistical analysis on the attentively learned correspondence matrices to evaluate inter-modal relationships for associating disease biomarkers. Our extensive experiments on three datasets demonstrate the superiority of our proposed method for disease diagnosis with promising prediction performance and multi-modal connectome biomarker location.

A Chebyshev Confidence Guided Source-Free Domain Adaptation Framework for Medical Image Segmentation.

Source-free domain adaptation (SFDA) aims to adapt models trained on a labeled source domain to an unlabeled target domain without access to source data. In medical imaging scenarios, the practical significance of SFDA methods has been emphasized due to data heterogeneity and privacy concerns. Recent state-of-the-art SFDA methods primarily rely on self-training based on pseudo-labels (PLs). Unfortunately, the accuracy of PLs may deteriorate due to domain shift, thus limiting the effectiveness of the adaptation process. To address this issue, we propose a Chebyshev confidence guided SFDA framework to accurately assess the reliability of PLs and generate self-improving PLs for self-training. The Chebyshev confidence is estimated by calculating the probability lower bound of PL confidence, given the prediction and the corresponding uncertainty. Leveraging the Chebyshev confidence, we introduce two confidence-guided denoising methods: direct denoising and prototypical denoising. Additionally, we propose a novel teacher-student joint training scheme (TJTS) that incorporates a confidence weighting module to iteratively improve PLs' accuracy. The TJTS, in collaboration with the denoising methods, effectively prevents the propagation of noise and enhances the accuracy of PLs. Extensive experiments in diverse domain scenarios validate the effectiveness of our proposed framework and establish its superiority over state-of-the-art SFDA methods. Our paper contributes to the field of SFDA by providing a novel approach for precisely estimating the reliability of PLs and a framework for obtaining high-quality PLs, resulting in improved adaptation performance.

BrainMass: Advancing Brain Network Analysis for Diagnosis with Large-scale Self-Supervised Learning.

Foundation models pretrained on large-scale datasets via self-supervised learning demonstrate exceptional versatility across various tasks. Due to the heterogeneity and hard-to-collect medical data, this approach is especially beneficial for medical image analysis and neuroscience research, as it streamlines broad downstream tasks without the need for numerous costly annotations. However, there has been limited investigation into brain network foundation models, limiting their adaptability and generalizability for broad neuroscience studies. In this study, we aim to bridge this gap. In particular, (1) we curated a comprehensive dataset by collating images from 30 datasets, which comprises 70,781 samples of 46,686 participants. Moreover, we introduce pseudo-functional connectivity (pFC) to further generates millions of augmented brain networks by randomly dropping certain timepoints of the BOLD signal. (2) We propose the BrainMass framework for brain network self-supervised learning via mask modeling and feature alignment. BrainMass employs Mask-ROI Modeling (MRM) to bolster intra-network dependencies and regional specificity. Furthermore, Latent Representation Alignment (LRA) module is utilized to regularize augmented brain networks of the same participant with similar topological properties to yield similar latent representations by aligning their latent embeddings. Extensive experiments on eight internal tasks and seven external brain disorder diagnosis tasks show BrainMass's superior performance, highlighting its significant generalizability and adaptability. Nonetheless, BrainMass demonstrates powerful few/zero-shot learning abilities and exhibits meaningful interpretation to various diseases, showcasing its potential use for clinical applications.

Ultra-stable insulin-glucagon fusion protein exploits an endogenous hepatic switch to mitigate hypoglycemic risk.

The risk of hypoglycemia and its serious medical sequelae restrict insulin replacement therapy for diabetes mellitus. Such adverse clinical impact has motivated development of diverse glucose-responsive technologies, including algorithm-controlled insulin pumps linked to continuous glucose monitors ("closed-loop systems") and glucose-sensing ("smart") insulins. These technologies seek to optimize glycemic control while minimizing hypoglycemic risk. Here, we describe an alternative approach that exploits an endogenous glucose-dependent switch in hepatic physiology: preferential insulin signaling (under hyperglycemic conditions) versus preferential counter-regulatory glucagon signaling (during hypoglycemia). Motivated by prior reports of glucagon-insulin co-infusion, we designed and tested an ultra-stable glucagon-insulin fusion protein whose relative hormonal activities were calibrated by respective modifications; physical stability was concurrently augmented to facilitate formulation, enhance shelf life and expand access. An N-terminal glucagon moiety was stabilized by an α-helix-compatible Lys 13 -Glu 17 lactam bridge; A C-terminal insulin moiety was stabilized as a single chain with foreshortened C domain. Studies in vitro demonstrated (a) resistance to fibrillation on prolonged agitation at 37 °C and (b) dual hormonal signaling activities with appropriate balance. Glucodynamic responses were monitored in rats relative to control fusion proteins lacking one or the other hormonal activity, and continuous intravenous infusion emulated basal subcutaneous therapy. Whereas efficacy in mitigating hyperglycemia was unaffected by the glucagon moiety, the fusion protein enhanced endogenous glucose production under hypoglycemic conditions. Together, these findings provide proof of principle toward a basal glucose-responsive insulin biotechnology of striking simplicity. The fusion protein's augmented stability promises to circumvent the costly cold chain presently constraining global insulin access. Significance Statement: The therapeutic goal of insulin replacement therapy in diabetes is normalization of blood-glucose concentration, which prevents or delays long-term complications. A critical barrier is posed by recurrent hypoglycemic events that results in short- and long-term morbidities. An innovative approach envisions co-injection of glucagon (a counter-regulatory hormone) to exploit a glycemia-dependent hepatic switch in relative hormone responsiveness. To provide an enabling technology, we describe an ultra-stable fusion protein containing insulin- and glucagon moieties. Proof of principle was obtained in rats. A single-chain insulin moiety provides glycemic control whereas a lactam-stabilized glucagon extension mitigates hypoglycemia. This dual-hormone fusion protein promises to provide a basal formulation with reduced risk of hypoglycemia. Resistance to fibrillation may circumvent the cold chain required for global access.

A deep connectome learning network using graph convolution for connectome-disease association study.

Multivariate analysis approaches provide insights into the identification of phenotype associations in brain connectome data. In recent years, deep learning methods including convolutional neural network (CNN) and graph neural network (GNN), have shifted the development of connectome-wide association studies (CWAS) and made breakthroughs for connectome representation learning by leveraging deep embedded features. However, most existing studies remain limited by potentially ignoring the exploration of region-specific features, which play a key role in distinguishing brain disorders with high intra-class variations, such as autism spectrum disorder (ASD), and attention deficit hyperactivity disorder (ADHD). Here, we propose a multivariate distance-based connectome network (MDCN) that addresses the local specificity problem by efficient parcellation-wise learning, as well as associating population and parcellation dependencies to map individual differences. The approach incorporating an explainable method, parcellation-wise gradient and class activation map (p-GradCAM), is feasible for identifying individual patterns of interest and pinpointing connectome associations with diseases. We demonstrate the utility of our method on two largely aggregated multicenter public datasets by distinguishing ASD and ADHD from healthy controls and assessing their associations with underlying diseases. Extensive experiments have demonstrated the superiority of MDCN in classification and interpretation, where MDCN outperformed competitive state-of-the-art methods and achieved a high proportion of overlap with previous findings. As a CWAS-guided deep learning method, our proposed MDCN framework may narrow the bridge between deep learning and CWAS approaches, and provide new insights for connectome-wide association studies.

Evaluating the resource curse hypothesis and the interplay of financial development, human development, and political stability in seven emerging economies.

The present study empirically confabulates the authenticity of the "resource curse hypothesis" in selected emerging nations. Furthermore, we also assessed the interconnections of three essential economic indicators with financial development, i.e., human development, political stability, and gross domestic product. To effectuate these objectives, we used annual data for the time frame 1990 to 2020 and advanced panel estimation techniques for getting the empirical outcomes. The study's empirical outcomes illustrate the existence of the "resource curse hypothesis" in sample nations. In addition, human development index and gross domestic product play an essential part in the furtherance of financial development in the long-run. The human development index is upsurging the financial development. Furthermore, political stability is also exerting a favorable influence on financial development. A similar interconnection is observed in the short-time period; nonetheless, the amplitude of the short-run impacts is smaller if we have a look at the long-run impacts. The empirical analysis offers a few pertinent policy insights for policymakers to improve the situation in the selected sample. Note: Financial development positively interconnected with human development, GDP and political stability while negatively associated with natural resources, respectively.

The effect of externally applied traditional Chinese medicine in diabetic foot: A systematic review and meta-analysis of 34 RCTs.

AIM: This study aimed to investigate the clinical efficacy of externally applied Traditional Chinese Medicine (TCM) on diabetic foot. METHODS: We searched the China Knowledge Network (CNKI), Wanfang Database, PubMed and Web of Science from inception to July 31, 2022, to find all randomized control trials (RCTs) related to externally applied TCMs in diabetic foot treatment. Information about the total effective rate, healing rate, and healing time were extracted. In addition, the relative risk (RR)/odds ratio (OR) or standardized mean difference (SMD) and 95 % confidence interval (CI) were calculated. RESULTS: Finally, a total of 34 RCTs including 3758 patients were included in this meta-analysis. There were 5 articles that reported hydropathic compress with astrogalin, 14 articles that reported MEBO burn cream, 9 articles that reported compound cortex phellodendri liquid and 6 articles that reported Shengji Yuhong ointment. Compared with the basic treatment, the externally applied TCM (astrogalin, MEBO burn cream, compound cortex phellodendri liquid and Shengji Yuhong ointment) combined with basic treatment improved the total effective rate (RR = 1.31 [1.20, 1.42], P < 0.0001) and healing rate (RR = 1.84 [1.56, 2.17], P < 0.0001) and shortened the healing time (SMD = - 2.51 [- 3.39, - 1.63], P < 0.0001). CONCLUSION: Our systematic review and meta-analysis revealed that common TCM applied externally could significantly improve the clinical efficacy comparing to the basic treatment.

CReg-KD: Model refinement via confidence regularized knowledge distillation for brain imaging.

One of the core challenges of deep learning in medical image analysis is data insufficiency, especially for 3D brain imaging, which may lead to model over-fitting and poor generalization. Regularization strategies such as knowledge distillation are powerful tools to mitigate the issue by penalizing predictive distributions and introducing additional knowledge to reinforce the training process. In this paper, we revisit knowledge distillation as a regularization paradigm by penalizing attentive output distributions and intermediate representations. In particular, we propose a Confidence Regularized Knowledge Distillation (CReg-KD) framework, which adaptively transfers knowledge for distillation in light of knowledge confidence. Two strategies are advocated to regularize the global and local dependencies between teacher and student knowledge. In detail, a gated distillation mechanism is proposed to soften the transferred knowledge globally by utilizing the teacher loss as a confidence score. Moreover, the intermediate representations are attentively and locally refined with key semantic context to mimic meaningful features. To demonstrate the superiority of our proposed framework, we evaluated the framework on two brain imaging analysis tasks (i.e. Alzheimer's Disease classification and brain age estimation based on T1-weighted MRI) on the Alzheimer's Disease Neuroimaging Initiative dataset including 902 subjects and a cohort of 3655 subjects from 4 public datasets. Extensive experimental results show that CReg-KD achieves consistent improvements over the baseline teacher model and outperforms other state-of-the-art knowledge distillation approaches, manifesting that CReg-KD as a powerful medical image analysis tool in terms of both promising prediction performance and generalizability.

Prognostic value of sarcopenia in older adults with transcatheter aortic valve implantation: A systematic review and meta-analysis.

OBJECTIVE: We conducted this systematic review and meta-analysis to summarize the prevalence of sarcopenia and its impact on mortality in patients undergoing TAVI. METHOD: Medline, EMBASE, and PubMed were searched from inception to October 14, 2022 to retrieve eligible studies that assessed sarcopenia in patients undergoing TAVI. Pooled sarcopenia prevalence was calculated with 95% confidence interval (CI), and heterogeneity was estimated using the I2 test. Associations of sarcopenia with mortality of post-TAVI were expressed as hazard ratio (HR) or odds ratios (OR) and 95% CI. RESULTS: 13 studies involving 5248 patients (mean age from 78.1 to 84.9 years) undergoing TAVI were included. There were eleven studies defined sarcopenia based on loss of skeletal muscle mass index (SMI), while only two studies used low muscle mass plus low muscle strength and/or low physical performance. Overall, the pooled prevalence of sarcopenia in patients undergoing TAVI was 49% (95% CI 41%-58%). Sarcopenia was associated with an increased risk of long-term (≥1 year) mortality in patients after TAVI (HR 1.57, 95% CI 1.33-1.85, P < 0.001), with similar findings in the subgroups stratified by follow-up time, definition of sarcopenia, study location, and study design. Furthermore, the 1-, 2-, and 3-year cumulative probabilities of survival in patients with sarcopenia were significantly lower than non-sarcopenia (74.0% vs 91.0%, 68.3% vs 78.0%, and 72.6% vs 79.8%, all P < 0.05). CONCLUSIONS: Although there are substantial differences in diagnostic criteria, sarcopenia is highly prevalent in patients undergoing TAVI and its linked to increased long-term mortality after TAVI.

Prognostic value of temporal muscle thickness, a novel radiographic marker of sarcopenia, in patients with brain tumor: A systematic review and meta-analysis.

Sarcopenia has been identified as a prognostic factor among certain types of cancer. However, it is unclear whether there is prognostic value of temporalis muscle thickness (TMT), a potential surrogate for sarcopenia, in adults patients with brain tumors. Therefore, we searched the Medline, Embase, and PubMed to systematically review and meta-analyze the relationship between TMT and overall survival, progression-free survival, and complications in patients with brain tumors and the hazard ratio (HR) or odds ratios (OR), and 95% confidence interval (CI) were evaluated. The quality in prognostic studies (QUIPS) instrument was employed to evaluate study quality. Nineteen studies involving 4570 patients with brain tumors were included for qualitative and quantitative analysis. Meta-analysis revealed thinner TMT was associated with poor overall survival (HR, 1.72; 95% CI, 1.45-2.04; P < 0.01) in patients with brain tumors. Sub-analyses showed that the association existed for both primary brain tumors (HR, 2.02; 95% CI, 1.55-2.63) and brain metastases (HR, 1.39; 95% CI, 1.30-1.49). Moreover, thinner TMT also was the independent predictor of progression-free survival in patients with primary brain tumors (HR, 2.88; 95% CI, 1.85-4.46; P < 0.01). Therefore, to improve clinical decision making it is important to integrate TMT assessment into routine clinical settings in patients with brain tumors.

MRF-Net: A multi-branch residual fusion network for fast and accurate whole-brain MRI segmentation.

Whole-brain segmentation from T1-weighted magnetic resonance imaging (MRI) is an essential prerequisite for brain structural analysis, e.g., locating morphometric changes for brain aging analysis. Traditional neuroimaging analysis pipelines are implemented based on registration methods, which involve time-consuming optimization steps. Recent related deep learning methods speed up the segmentation pipeline but are limited to distinguishing fuzzy boundaries, especially encountering the multi-grained whole-brain segmentation task, where there exists high variability in size and shape among various anatomical regions. In this article, we propose a deep learning-based network, termed Multi-branch Residual Fusion Network, for the whole brain segmentation, which is capable of segmenting the whole brain into 136 parcels in seconds, outperforming the existing state-of-the-art networks. To tackle the multi-grained regions, the multi-branch cross-attention module (MCAM) is proposed to relate and aggregate the dependencies among multi-grained contextual information. Moreover, we propose a residual error fusion module (REFM) to improve the network's representations fuzzy boundaries. Evaluations of two datasets demonstrate the reliability and generalization ability of our method for the whole brain segmentation, indicating that our method represents a rapid and efficient segmentation tool for neuroimage analysis.