Efficient and Flexible Method for Reducing Moderate-Size Deep Neural Networks with Condensation.

Neural networks have been extensively applied to a variety of tasks, achieving astounding results. Applying neural networks in the scientific field is an important research direction that is gaining increasing attention. In scientific applications, the scale of neural networks is generally moderate size, mainly to ensure the speed of inference during application. Additionally, comparing neural networks to traditional algorithms in scientific applications is inevitable. These applications often require rapid computations, making the reduction in neural network sizes increasingly important. Existing work has found that the powerful capabilities of neural networks are primarily due to their nonlinearity. Theoretical work has discovered that under strong nonlinearity, neurons in the same layer tend to behave similarly, a phenomenon known as condensation. Condensation offers an opportunity to reduce the scale of neural networks to a smaller subnetwork with a similar performance. In this article, we propose a condensation reduction method to verify the feasibility of this idea in practical problems, thereby validating existing theories. Our reduction method can currently be applied to both fully connected networks and convolutional networks, achieving positive results. In complex combustion acceleration tasks, we reduced the size of the neural network to 41.7% of its original scale while maintaining prediction accuracy. In the CIFAR10 image classification task, we reduced the network size to 11.5% of the original scale, still maintaining a satisfactory validation accuracy. Our method can be applied to most trained neural networks, reducing computational pressure and improving inference speed.

New small-molecule alcohol synthesis by breaking the space limitation of the "aromatic cage" in Pseudomonas sp. AK1 BBOX.

Fe(II)/2OG-dependent oxygenase γ-butyrobetaine hydroxylase (BBOX) stereoselectively hydroxylates inactive C-H bonds and produces L-carnitine. It has potential applications in the biosynthesis of L-carnitine and the synthesis of other small molecule alcohols. In this paper, we systematically explore the substrate range of Pseudomonas sp. AK1 BBOX (psBBOX), with emphasis on the quaternary ammonium portion of γ-butyrobetaine (γ-BB). The space limitation of the "aromatic cage" in psBBOX in the hydroxylation of large quaternary ammonium analogues was studied, and the role of four aromatic amino acid residues in the substrate binding mode was analyzed. Consequently, the F188A mutant was developed with the ability to hydroxylate cyclic quaternary ammonium analogues and generate new alcohol compounds by breaking the limitation of the "aromatic cage".

Fluorescent dyes with multiple quaternary ammonium centers for specific image discrimination and Gram-positive antibacterial activity.

Three quaternary ammonium compounds (QACs), TPQA, T2PQA, and T3PQA, were synthesized and employed in antimicrobial tests against E. coli and S. aureus. It was confirmed that they exhibit selective bacteriostasis against S. aureus. The antibacterial activities of the compounds were evaluated via determining their minimum inhibitory concentrations (MICs) and minimal bactericidal concentrations (MBCs) against S. aureus using the 2,3,5-triphenyltetrazolium chloride (TTC) coloration method. Notably, T2PQA exhibited far better properties than TPQA and T3PQA, with the activity found to be dependent on the structure of the QA and the exposed hydrophobic groups. All three compounds showed promising potential for killing Gram-positive bacteria, efficiently guided by fluorescence imaging.

Asprosin in the Paraventricular Nucleus Induces Sympathetic Activation and Pressor Responses via cAMP-Dependent ROS Production.

Asprosin is a newly discovered adipokine that is involved in regulating metabolism. Sympathetic overactivity contributes to the pathogenesis of several cardiovascular diseases. The paraventricular nucleus (PVN) of the hypothalamus plays a crucial role in the regulation of sympathetic outflow and blood pressure. This study was designed to determine the roles and underlying mechanisms of asprosin in the PVN in regulating sympathetic outflow and blood pressure. Experiments were carried out in male adult SD rats under anesthesia. Renal sympathetic nerve activity (RSNA), mean arterial pressure (MAP), and heart rate (HR) were recorded, and PVN microinjections were performed bilaterally. Asprosin mRNA and protein expressions were high in the PVN. The high asprosin expression in the PVN was involved in both the parvocellular and magnocellular regions according to immunohistochemical analysis. Microinjection of asprosin into the PVN produced dose-related increases in RSNA, MAP, and HR, which were abolished by superoxide scavenger tempol, antioxidant N-acetylcysteine (NAC), and NADPH oxidase inhibitor apocynin. The asprosin promoted superoxide production and increased NADPH oxidase activity in the PVN. Furthermore, it increased the cAMP level, adenylyl cyclase (AC) activity, and protein kinase A (PKA) activity in the PVN. The roles of asprosin in increasing RSNA, MAP, and HR were prevented by pretreatment with AC inhibitor SQ22536 or PKA inhibitor H89 in the PVN. Microinjection of cAMP analog db-cAMP into the PVN played similar roles with asprosin in increasing the RSNA, MAP, and HR, but failed to further augment the effects of asprosin. Pretreatment with PVN microinjection of SQ22536 or H89 abolished the roles of asprosin in increasing superoxide production and NADPH oxidase activity in the PVN. These results indicated that asprosin in the PVN increased the sympathetic outflow, blood pressure, and heart rate via cAMP-PKA signaling-mediated NADPH oxidase activation and the subsequent superoxide production.

Top-Down Inhibitory Mechanisms Underlying Auditory-Motor Integration for Voice Control: Evidence by TMS.

The dorsolateral prefrontal cortex (DLPFC) has been implicated in auditory-motor integration for accurate control of vocal production, but its precise role in this feedback-based process remains largely unknown. To this end, the present event-related potential study applied a transcranial magnetic stimulation (TMS) protocol, continuous theta-burst stimulation (c-TBS), to disrupt cortical activity in the left DLPFC as young adults vocalized vowel sounds while hearing their voice unexpectedly shifted upwards in pitch. The results showed that, as compared to the sham condition, c-TBS over left DLPFC led to significantly larger vocal compensations for pitch perturbations that were accompanied by significantly smaller cortical P2 responses. Source localization analyses revealed that this brain activity pattern was the result of reduced activation in the left superior frontal gyrus and right inferior parietal lobule (supramarginal gyrus). These findings demonstrate c-TBS-induced modulatory effects of DLPFC on the neurobehavioral processing of vocal pitch regulation, suggesting that disrupting prefrontal function may impair top-down inhibitory control mechanisms that prevent speech production from being excessively influenced by auditory feedback, resulting in enhanced vocal compensations for feedback perturbations. This is the first study that provides direct evidence for a causal role of the left DLPFC in auditory feedback control of vocal production.

Optimization of Extraction of Bioactive Compounds from Baphicacanthus cusia Leaves by Hydrophobic Deep Eutectic Solvents.

Deep eutectic solvents (DESs) are considered as efficient and green solvents for the extraction of bioactive compounds from medicinal plants. In this work, a novel method of DES-based ultrasound-assisted extraction of bioactive compounds from Baphicacanthus cusia leaves (BCL) was established. Systematic screening and the morphology of the original and treated BCL were observed with scanning electron microscopy to determine the extraction efficiency of different solvents. The extraction conditions were optimized by Box-Behnken design (BBD) tests and the optimal extraction conditions were as follows: lactic acid/L-menthol ratio of 5: 2 (mol/mol), solid-liquid ratio of 80.0 mL/g and temperature of 60.5 °C. The extraction yields of tryptanthrin, indigo and indirubin reached 0.356, 1.744 and 0.562 mg/g, respectively. The results of a 2,2-diphenyl-1-picrylhydrazy (DPPH) radical scavenging activity test indicated the feasibility of DESs in the extraction of bioactive compounds. This study indicated that L-menthol/lactic acid was a green and efficient solvent for the extraction of bioactive compounds from BCL, and DES-based ultrasound-assisted extraction could be used as an effective application strategy for the extraction of bioactive compounds from medicinal plants.

Neural networks of different species, brain areas and states can be characterized by the probability polling state.

Cortical networks are complex systems of a great many interconnected neurons that operate from collective dynamical states. To understand how cortical neural networks function, it is important to identify their common dynamical operating states from the probabilistic viewpoint. Probabilistic characteristics of these operating states often underlie network functions. Here, using multi-electrode data from three separate experiments, we identify and characterize a cortical operating state (the "probability polling" or "p-polling" state), common across mouse and monkey with different behaviors. If the interaction among neurons is weak, the p-polling state provides a quantitative understanding of how the high dimensional probability distribution of firing patterns can be obtained by the low-order maximum entropy formulation, effectively utilizing a low dimensional stimulus-coding structure. These results show evidence for generality of the p-polling state and in certain situations its advantage of providing a mathematical validation for the low-order maximum entropy principle as a coding strategy.

Cerebellar contribution to auditory feedback control of speech production: Evidence from patients with spinocerebellar ataxia.

The cerebellum has been implicated in the feedforward control of speech production. However, the role of the cerebellum in the feedback control of speech production remains unclear. To address this question, the present event-related potential study examined the behavioral and neural correlates of auditory feedback control of vocal production in patients with spinocerebellar ataxia (SCA) and healthy controls. All participants were instructed to produce sustained vowels while hearing their voice unexpectedly pitch-shifted -200 or -500 cents. The behavioral results revealed significantly larger vocal compensations for pitch perturbations in patients with SCA relative to healthy controls. At the cortical level, patients with SCA exhibited significantly smaller cortical P2 responses that were source localized in the right superior temporal gyrus, primary auditory cortex, and supramarginal gyrus than healthy controls. These findings indicate that reduced brain activity in the right temporal and parietal regions are significant neural contributors to abnormal auditory-motor processing of vocal pitch regulation as a consequence of cerebellar degeneration, which may be related to disrupted reciprocal interactions between the cerebellum and cortical regions that support the top-down modulation of auditory-vocal integration. These differences in behavior and cortical activity between healthy controls and patients with SCA demonstrate that the cerebellum is not only essential for feedforward control but also plays a crucial role in the feedback-based control of speech production.

Dynamical and Coupling Structure of Pulse-Coupled Networks in Maximum Entropy Analysis.

Maximum entropy principle (MEP) analysis with few non-zero effective interactions successfully characterizes the distribution of dynamical states of pulse-coupled networks in many fields, e.g., in neuroscience. To better understand the underlying mechanism, we found a relation between the dynamical structure, i.e., effective interactions in MEP analysis, and the anatomical coupling structure of pulse-coupled networks and it helps to understand how a sparse coupling structure could lead to a sparse coding by effective interactions. This relation quantitatively displays how the dynamical structure is closely related to the anatomical coupling structure.

Association between ERCC1 C8092A and ERCC2 K751Q polymorphisms and risk of adult glioma: a meta-analysis.

While the ERCC1 C8092A and ERCC2 K751Q polymorphisms have received much attention for their potential associations with adult glioma risk, inferences from such studies are hindered by their limited statistical power and conflicting results. The aim of this meta-analysis is to provide a relatively comprehensive account of the association between these two polymorphisms and adult glioma risk. A literature search for eligible studies published before September 1, 2013 was conducted in PubMed, Embase, Web of Science, Cochrane Library, and CNKI databases. Pooled odds ratios (ORs) with their corresponding 95% confidence intervals (95% CIs) were used to evaluate the strength of the association under a fixed or random effect model according to heterogeneity test results. All analyses were performed using STATA software, version 12.0. Ten case-control studies were included in this meta-analysis, with a total of 5,843 adult glioma patients and 8,139 healthy controls. For ERCC1 C8092A (dbSNP: rs3212986, C>A), the combined results show that carriers of the AA genotype may be associated with a higher risk of adult glioma than carriers of the CA and CC genotypes. Stratified analyses show that the magnitude of the effect was especially significant among Asians, indicating ethnicity differences in adult glioma susceptibility. For ERCC2 K751Q (dbSNP: rs13181, A>C), the pooled ORs were not significant in the overall population, although all of the ORs were greater than 1. However, Asians seem to be significantly more susceptible to adult glioma than Caucasians. The results of this meta-analysis indicate that the AA genotype of ERCC1 C8092A may be associated with a higher risk of adult glioma than the CA and CC genotypes and that the risk allele of ERCC2 K751Q confers a significant susceptibility to adult glioma, especially in Asian populations. These polymorphisms may be used along with other genetic markers to identify individuals at high risk for adult glioma.

Implicit Regularization of Dropout.

It is important to understand how dropout, a popular regularization method, aids in achieving a good generalization solution during neural network training. In this work, we present a theoretical derivation of an implicit regularization of dropout, which is validated by a series of experiments. Additionally, we numerically study two implications of the implicit regularization, which intuitively rationalizes why dropout helps generalization. First, we find that input weights of hidden neurons tend to condense on isolated orientations trained with dropout. Condensation is a feature in the non-linear learning process, which makes the network less complex. Second, we find that the training with dropout leads to the neural network with a flatter minimum compared with standard gradient descent training, and the implicit regularization is the key to finding flat solutions. Although our theory mainly focuses on dropout used in the last hidden layer, our experiments apply to general dropout in training neural networks. This work points out a distinct characteristic of dropout compared with stochastic gradient descent and serves as an important basis for fully understanding dropout.

Effect of Rhythmically Cued Exercise Interventions on Functions in Patients With Parkinson Disease: A Meta-Analysis.

OBJECTIVE: The purpose of this review was to investigate the efficacy of rhythmically cued exercise interventions on motor function, cognition, and mental state in patients with Parkinson disease. METHODS: PubMed, Cochrane Database, Web of Science, Embase, and CINAHL were searched June 15, 2023. Original studies investigating the efficacy of rhythmically cued exercise interventions on the functions of patients with Parkinson disease were included. The Cochrane risk-of-bias assessment tool was used to evaluate the risk of bias. The protocol was registered in PROSPERO (CRD42022371203). RESULTS: A total of 38 original studies involving 1486 participants were included. Rhythmically cued exercise interventions demonstrated superior effects on motor function compared to exercise therapy without rhythm (standardized mean difference [SMD] = -0.31). However, no significant improvements were observed in cognition and mental state. Overall, significant improvements were observed in motor examination (SMD = -0.61), Timed "Up & Go" Test (mean difference [MD] = -0.91), activities of daily living (SMD = -0.49), balance (SMD = 0.59), walking velocity (MD = 0.06), step length (MD = 2.65), and stride length (MD = 0.04) following rhythmically cued exercise interventions. No significant improvements were observed in freezing of gait and cadence. Assessment of publication bias showed no significant evidence of publication bias. Meta-regression analyses revealed a significant association between treatment duration and improvement in motor function. Furthermore, adverse events and dropout rates did not significantly differ between the 2 groups. CONCLUSION: Rhythmically cued exercise interventions are effective in improving motor function in the early to middle stages of Parkinson disease. More than 10 weeks of intervention yielded better results. However, these interventions do not have a significant impact on cognition and mental states. Importantly, rhythmically cued exercise interventions are safe and well tolerated. Large-scale trials are needed for further confirmation. IMPACT: This study contributes to the development of safe and reliable home rehabilitation programs, aiming to enhance the quality of life for patients with Parkinson disease.

Timing Matters: Time of Day Impacts the Ergogenic Effects of Caffeine-A Narrative Review.

Caffeine has attracted significant attention from researchers in the sports field due to its well-documented ergogenic effects across various athletic disciplines. As research on caffeine continues to progress, there has been a growing emphasis on evaluating caffeine dosage and administration methods. However, investigations into the optimal timing of caffeine intake remain limited. Therefore, this narrative review aimed to assess the ergogenic effects of caffeine administration at different times during the morning (06:00 to 10:00) and evening (16:00 to 21:00). The review findings suggest that circadian rhythms play a substantial role in influencing sports performance, potentially contributing to a decline in morning performance. Caffeine administration has demonstrated effectiveness in mitigating this phenomenon, resulting in ergogenic effects and performance enhancement, even comparable to nighttime levels. While the specific mechanisms by which caffeine regulates circadian rhythms and influences sports performance remain unclear, this review also explores the mechanisms underlying caffeine's ergogenic effects, including the adenosine receptor blockade, increased muscle calcium release, and modulation of catecholamines. Additionally, the narrative review underscores caffeine's indirect impact on circadian rhythms by enhancing responsiveness to light-induced phase shifts. Although the precise mechanisms through which caffeine improves morning performance declines via circadian rhythm regulation necessitate further investigations, it is noteworthy that the timing of caffeine administration significantly affects its ergogenic effects during exercise. This emphasizes the importance of considering caffeine intake timing in future research endeavors to optimize its ergogenic potential and elucidate its mechanisms.

Asprosin contributes to vascular remodeling in hypertensive rats via superoxide signaling.

OBJECTIVE: Proliferation and migration of vascular smooth muscle cells (VSMCs) contribute to vascular remodeling. Asprosin, a newly discovered protein hormone, is involved in metabolic diseases. Little is known about the roles of asprosin in cardiovascular diseases. This study focused on the role and mechanism of asprosin on VSMC proliferation and migration, and vascular remodeling in a rat model of hypertension. METHODS AND RESULTS: VSMCs were obtained from the aortic media of 8-week-old male Wistar-Kyoto rats (WKY) and spontaneously hypertensive rats (SHR). Asprosin was upregulated in the VSMCs of SHR. For in vitro studies, asprosin promoted VSMC proliferation and migration of WKY and SHR, and increased Nicotinamide adenine dinucleotide phosphate (NADPH) oxidase (NOX) activity, NOX1/2/4 protein expressions and superoxide production. Knockdown of asprosin inhibited the proliferation, migration, NOX activity, NOX1/2 expressions and superoxide production in the VSMCs of SHR. The roles of asprosin in promoting VSMC proliferation and migration were not affected by hydrogen peroxide scavenger, but attenuated by superoxide scavenger, selective NOX1 or NOX2 inhibitor. Toll-like receptor 4 (TLR4) was upregulated in SHR, TLR4 knockdown inhibited asprosin overexpression-induced proliferation, migration and oxidative stress in VSMCs of WKY and SHR. Asprosin was upregulated in arteries of SHR, and knockdown of asprosin in vivo not only attenuated oxidative stress and vascular remodeling in aorta and mesentery artery, but also caused a subsequent persistent antihypertensive effect in SHR. CONCLUSIONS: Asprosin promotes VSMC proliferation and migration via NOX-mediated superoxide production. Inhibition of endogenous asprosin expression attenuates VSMC proliferation and migration, and vascular remodeling of SHR.

Ile-Pro-Pro attenuates sympathetic activity and hypertension.

Isoleucine-proline-proline (Ile-Pro-Pro, IPP) is a natural food source tripeptide that inhibits angiotensin-converting enzyme (ACE) activity. The aim of this study was to determine the central and peripheral roles of IPP in attenuating sympathetic activity, oxidative stress and hypertension. Male Sprague-Dawley rats were subjected to sham-operated surgery (Sham) or two-kidney one-clip (2K1C) surgery to induce renovascular hypertension. Renal sympathetic nerve activity and blood pressure were recorded. Bilateral microinjections of IPP to hypothalamic paraventricular nucleus (PVN) attenuated sympathetic activity (-16.1 ± 2.5%, P < 0.001) and hypertension (-8.7 ± 1.5 mmHg, P < 0.01) in 2K1C rats by inhibiting ACE activity and subsequent angiotensin II and superoxide production in the PVN. Intravenous injections of IPP also attenuated sympathetic activity (-15.1 ± 2.1%, P < 0.001) and hypertension (-16.8 ± 2.3 mmHg, P < 0.001) via inhibiting ACE activity and oxidative stress in both PVN and arteries of 2K1C rats. The duration of the effects of the intravenous IPP was longer than those of the PVN microinjection, but the sympatho-inhibitory effect of intravenous injections occurred later than that of the PVN microinjection. Intraperitoneal injection of IPP (400 pmol/day for 20 days) attenuated hypertension and vascular remodeling via inhibiting ACE activity and oxidative stress in both PVN and arteries of 2K1C rats. These results indicate that IPP attenuates hypertension and sympathetic activity by inhibiting ACE activity and oxidative stress. The sympathoinhibitory effect of peripheral IPP is mainly caused by the ACE inhibition in PVN, and the antihypertensive effect is related to the sympathoinhibition and the arterial ACE inhibition. Long-term intraperitoneal IPP therapy attenuates hypertension, oxidative stress and vascular remodeling.

Causal and structural connectivity of pulse-coupled nonlinear networks.

We study the reconstruction of structural connectivity for a general class of pulse-coupled nonlinear networks and show that the reconstruction can be successfully achieved through linear Granger causality (GC) analysis. Using spike-triggered correlation of whitened signals, we obtain a quadratic relationship between GC and the network couplings, thus establishing a direct link between the causal connectivity and the structural connectivity within these networks. Our work may provide insight into the applicability of GC in the study of the function of general nonlinear networks.

A massive calcification and ossification of the transverse sinus and the neighbouring dura mimicking meningioma.

BACKGROUND: Although small calcifications of the dura and the transverse sinus occur frequently, large, single intracranial calcifications originating from the transverse sinus and the neighbouring dura are rare. CASE PRESENTATION: A 47-year-old man was admitted to the hospital for a right occipital headache that had persisted for two weeks. There was no neurological deficit. Normal skull X-ray and computed tomography (CT) scans revealed an irregular, calcified, intracranial lesion of approximately 4.4 × 4.0 × 2.5 cm in volume in the right occipital region. Via surgery, a bone-hard, poorly vascularised, pink mass originating from the right transverse sinus and the convex dura of the right cerebellar hemisphere, as well as the cerebellar tentorium, was completely removed. Pathological examination yielded a diagnosis of fibrous connective tissue with hyaline degeneration, calcification and ossification with no indication of neoplasia or inflammation. CONCLUSIONS: We report a rare case of massive calcification and ossification of the transverse sinus and the neighbouring dura mimicking meningioma. Degenerative calcification and ossification may serve as a rare differential diagnosis of diseases, such as meningiomas, in the transverse sinus and the neighbouring dura.

Diversity-Oriented Biosynthesis Yields l-Kynurenine Derivative-Based Neurological Drug Candidate Collection.

Natural product libraries with a remarkable range of biological activities play pivotal roles in drug discoveries due to their extraordinary structural complexity and immense diversity. l-Kynurenine (l-Kyn)-based derivatives are privileged pharmacophores that exhibit diverse therapeutic implications in neurological disorders. However, the difficulty in obtaining l-Kyn analogues with different skeletal structures has recently led to a decline in its medicinal research. Herein, we report a two-step, one-pot protocol for diversity-oriented biosynthesis of a collection of previously intractable l-Kyn-like compounds. The success of these challenging transformations mainly depends on unlocking the new catalytic scope of tryptophan 2,3-dioxygenases, followed by rational site-directed mutagenesis to modify the substrate domains further. As a result, 18 kynurenine analogues with diverse molecular scaffolds can be rapidly assembled in a predictable manner with 20-83% isolated yields, which not only fill the voids of the catalytic profile of tryptophan 2,3-dioxygenases with an array of substituent groups (e.g., F, Cl, Br, I, CH3, OCH3, and NO2) but also update the current understanding of its substrate spectrum. Our work highlights the great potential of existing enzymes in addressing long-standing synthetic challenges for facilitating the development or discovery of new drug candidates. Furthermore, our approach enables translating the reaction parameters from Eppendorf tubes to 1 L scale, affording l-4-Cl-Kyn and l-5-Cl-Kyn both on a gram scale with more than 80% isolated yields, and provides a promising alternative to further industrial applications.

Continuous theta burst stimulation over right cerebellum for speech impairment in Parkinson's disease: study protocol for a randomized, sham-controlled, clinical trial.

Background: Speech impairment is a common symptom of Parkinson's disease (PD) that worsens with disease progression and affects communication and quality of life. Current pharmacological and surgical treatments for PD have inconsistent effects on speech impairment. The cerebellum is an essential part of sensorimotor network that regulates speech production and becomes dysfunctional in PD. Continuous theta-burst stimulation (cTBS) is a non-invasive brain stimulation technique that can modulate the cerebellum and its connections with other brain regions. Objective: To investigate whether cTBS over the right cerebellum coupled with speech-language therapy (SLT) can improve speech impairment in PD. Methods: In this randomized controlled trial (RCT), 40 patients with PD will be recruited and assigned to either an experimental group (EG) or a control group (CG). Both groups will receive 10 sessions of standard SLT. The EG will receive real cTBS over the right cerebellum, while the CG will receive sham stimulation. Blinded assessors will evaluate the treatment outcome at three time points: pre-intervention, post-intervention, and at a 12-week follow-up. The primary outcome measures are voice/speech quality and neurobehavioral parameters of auditory-vocal integration. The secondary outcome measures are cognitive function, quality of life, and functional connectivity determined by resting-state functional magnetic resonance imaging (fMRI). Significance: This trial will provide evidence for the efficacy and safety of cerebellar cTBS for the treatment of speech impairment in PD and shed light on the neural mechanism of this intervention. It will also have implications for other speech impairment attributed to cerebellar dysfunctions. Clinical trial registration: www.chictr.org.cn, identifier ChiCTR2100050543.

Data-informed deep optimization.

Motivated by the impressive success of deep learning in a wide range of scientific and industrial applications, we explore in this work the application of deep learning into a specific class of optimization problems lacking explicit formulas for both objective function and constraints. Such optimization problems exist in many design problems, e.g., rotor profile design, in which objective and constraint values are available only through experiment or simulation. They are especially challenging when design parameters are high-dimensional due to the curse of dimensionality. In this work, we propose a data-informed deep optimization (DiDo) approach emphasizing on the adaptive fitting of the the feasible region as follows. First, we propose a deep neural network (DNN) based adaptive fitting approach to learn an accurate DNN classifier of the feasible region. Second, we use the DNN classifier to efficiently sample feasible points and train a DNN surrogate of the objective function. Finally, we find optimal points of the DNN surrogate optimization problem by gradient descent. To demonstrate the effectiveness of our DiDo approach, we consider a practical design case in industry, in which our approach yields good solutions using limited size of training data. We further use a 100-dimension toy example to show the effectiveness of our approach for higher dimensional problems. Our results indicate that, by properly dealing with the difficulty in fitting the feasible region, a DNN-based method like our DiDo approach is flexible and promising for solving high-dimensional design problems with implicit objective and constraints.