Improving the performance of deep learning models in predicting and classifying gamma passing rates with discriminative features and a class balancing technique: a retrospective cohort study.

BACKGROUND: The purpose of this study was to improve the deep learning (DL) model performance in predicting and classifying IMRT gamma passing rate (GPR) by using input features related to machine parameters and a class balancing technique. METHODS: A total of 2348 fields from 204 IMRT plans for patients with nasopharyngeal carcinoma were retrospectively collected to form a dataset. Input feature maps, including fluence, leaf gap, leaf speed of both banks, and corresponding errors, were constructed from the dynamic log files. The SHAP framework was employed to compute the impact of each feature on the model output for recursive feature elimination. A series of UNet++ based models were trained on the obtained eight feature sets with three fine-tuning methods including the standard mean squared error (MSE) loss, a re-sampling technique, and a proposed weighted MSE loss (WMSE). Differences in mean absolute error, area under the receiver operating characteristic curve (AUC), sensitivity, and specificity were compared between the different models. RESULTS: The models trained with feature sets including leaf speed and leaf gap features predicted GPR for failed fields more accurately than the other models (F(7, 147) = 5.378, p < 0.001). The WMSE loss had the highest accuracy in predicting GPR for failed fields among the three fine-tuning methods (F(2, 42) = 14.149, p < 0.001), while an opposite trend was observed in predicting GPR for passed fields (F(2, 730) = 9.907, p < 0.001). The WMSE_FS5 model achieved a superior AUC (0.92) and more balanced sensitivity (0.77) and specificity (0.89) compared to the other models. CONCLUSIONS: Machine parameters can provide discriminative input features for GPR prediction in DL. The novel weighted loss function demonstrates the ability to balance the prediction and classification accuracy between the passed and failed fields. The proposed approach is able to improve the DL model performance in predicting and classifying GPR, and can potentially be integrated into the plan optimization process to generate higher deliverability plans. TRIAL REGISTRATION: This clinical trial was registered in the Chinese Clinical Trial Registry on March 26th, 2020 (registration number: ChiCTR2000031276). https://clinicaltrials.gov/ct2/show/ChiCTR2000031276.

Integrative analysis of the transcriptome and proteome reveals the molecular responses of tobacco to boron deficiency.

BACKGROUND: Boron (B) is an essential micronutrient for plants. Inappropriate B supply detrimentally affects the productivity of numerous crops. Understanding of the molecular responses of plants to different B supply levels would be of significance in crop improvement and cultivation practices to deal with the problem. RESULTS: We conducted a comprehensive analysis of the transcriptome and proteome of tobacco seedlings to investigate the expression changes of genes/proteins in response to different B supply levels, with a particular focus on B deficiency. The global gene and protein expression profiles revealed the potential mechanisms involved in the responses of tobacco to B deficiency, including up-regulation of the NIP5;1-BORs module, complex regulation of genes/proteins related to cell wall metabolism, and up-regulation of the antioxidant machinery. CONCLUSION: Our results demonstrated that B deficiency caused severe morphological and physiological disorders in tobacco seedlings, and revealed dynamic expression changes of tobacco genes/proteins in response to different B supply levels, especially to B deficiency, thus offering valuable insights into the molecular responses of tobacco to B deficiency.

Relationship between triglyceride-glucose index and gallstones risk: a population-based study.

Background: Globally, gallstones represented a prevalent condition of the digestive system, heavily affected by metabolic dysfunctions such as obesity, dyslipidemia, insulin resistance, and diabetes. The triglyceride-glucose (TyG) index served as an accessible novel indicator for evaluating insulin resistance, offering a precise reflection of metabolic conditions. However, no studies have yet explored their relationship. The link between the TyG and gallstone risk was the primary purpose of this study. Methods: Utilized data from the public database, the National Health and Nutrition Examination Survey, for the years 2017-2020. The logit model was utilized to elucidate the connection between the TyG and the gallstones risk. The restricted cubic spline (RCS) analysis served to verify any non-linear relationships existing between them. Sensitivity analyses, encompassing both stratified and interaction analyses, were conducted to identify populations of particular interest and assess potential interactions between covariates and the TyG index. Results: A total of 4544 individuals were included. The risk of gallstones in high group was 1.6 times that of the low group. The potential cut-off value for the TyG index was 6.19. Above this threshold, there was a 40% heightened risk of gallstones with each one-unit increment in the TyG. The RCS analysis revealed the absence of a non-linear association between them. The populations warranting particular focus included those over 60 years, non-White people, individuals with a body mass index ≥25, smokers, drinkers, those with hypertension, and diabetes. Apart from smoking history, alcohol consumption, and history of diabetes, there were no interactions between other variables and the TyG index. Conclusion: The current study represented the inaugural investigation into the link between TyG index and the risk of gallstones. A positive correlation existed between them, signifying that an increase in TyG paralleled an elevated risk of gallstones. No non-linear relationship has been found between them. Besides, a 40% increase in gallstone risk accompanied each unit rise in TyG. Considering the convenience and accessibility of TyG in clinical settings, it has a promising potential for clinical application.

Dynamic response of allelopathic potency of Taxus cuspidata Sieb. et Zucc. mediated by allelochemicals in Ficus carica Linn. root exudates.

In a mixed forest, certain plants can release allelochemicals that exert allelopathic effects on neighboring plants, thereby facilitating interspecific coexistence of two species. Previous studies have demonstrated that allelochemicals released from Ficus carica Linn. roots in mixed forest of F. carica and Taxus cuspidata Sieb. et Zucc. has phase characteristics over time, which can improve the soil physicochemical properties, enzyme activity and microbial diversity, thus promoting the growth of T. cuspidata. Based on the irrigation of exogenous allelochemicals, changes in soil fertility (soil physical and chemical properties, soil enzyme activity and soil microelement content) were observed in response to variations in allelochemicals during five phases of irrigation: initial disturbance phase (0-2 d), physiological compensation phase (2-8 d), screening phase (8-16 d), restore phase (16-32 d) and maturity phase (32-64 d), which was consistent with the response of soil microorganisms. The allelopathic response of growth physiological indexes of T. cuspidata, however, exhibited a slight lag behind the soil fertility, with distinct phase characteristics becoming evident on the 4th day following irrigation of allelochemicals. The findings demonstrated that the allelochemicals released by the root of F. carica induced a synergistic effect on soil fertility and microorganisms, thereby facilitating the growth of T. cuspidata. This study provides a comprehensive elucidation of the phased dynamic response-based allelopathic mechanism employed by F. carica to enhance the growth of T. cuspidata, thus establishing a theoretical basis for optimizing forest cultivation through allelopathic pathways.

Invasion of Trifolium repens L. aggravated by biodegradable plastics: adjustable strategy for foraging N and P.

The invasion of alien plant and the pollution caused by soil microplastics have emerged as significant ecological threats. Recent studies have demonstrated aggravating effect of non-biodegradable microplastics on plant invasion. However, the impact of biodegradable microplastics (BMPs) on plant invasion remains unclear. Therefore, it is imperative to explore the impact of BMPs on plant invasion. In this study, a 30-day potting experiment with Trifolium repens L. (an invasive plant) and Oxalis corniculata L. (a native plant) was conducted to evaluate the influence of BMPs on T. repens's invasion. The findings revealed that BMPs results in a reduction in available N and P contents, thereby facilitating the colonization of arbuscular mycorrhizal fungi on T. repens 's roots. Consequently, T. repens adjusted its N and P foraging strategy by increasing P absorption ratio, and enhancing the accumulation of N and P in leaves. This ultimately led to the decrease of relative neighbor effect index of T. repens, indicating an aggravated invasion by T. repens. This study significantly enhances and expands the understanding of mechanisms by which microplastics aggravate plant invasion.

Molecularly imprinted beads based on modified cellulose hydrogel:A novel solid phase extraction filler for specific adsorption of camptothecin.

Traditional solid phase extraction (SPE) suffers from a lack of specific adsorption. To overcome this problem, a combination of adsorption method and molecular imprinting technology by polydopamine modification was proposed to realize specific recognition of target compounds in SPE, which is of great significance to improve the separation efficiency of SPE. Cellulose hydrogel beads were prepared by dual cross-linking curing method and modified with polydopamine to make them hydrophilic and biocompatible. Subsequently, cellulose hydrogel-based molecularly imprinted beads (MIBs) were synthesized by surface molecular imprinting technology and used as novel column fillers in SPE to achieve efficient adsorption (34.16 mg·g-1) with specific selectivity towards camptothecin (CPT) in 120 min. The simulation and NMR analysis revealed that recognition mechanism of MIBs involved hydrogen bond interactions and Van der Waals effect. The MIBs were successful used in separating CPT from Camptotheca acuminata fruits, exhibiting impressive adsorption capacity (1.19 mg·g-1) and efficient recovery of CPT (81.54 %). Thus, an environmentally friendly column filler for SPE was developed, offering a promising avenue for utilizing cellulose-based materials in the selective separation of natural products.

Review of clinical characteristics and mortality outcomes in patients on maintenance hemodialysis during the Omicron surge: a single center experience.

BACKGROUND: This hemodialysis center experienced the pandemic from December 2022 to January 2023. Therefore, we sought to describe the clinical characteristics and mortality outcomes in hemodialysis patients during this Omicron surge. METHODS: According to whether they are infected, they are divided into two groups: SARS-CoV-2-positive and SARS-CoV-2-negative. The SARS-CoV-2-positive group was divided into a survival group and a non-survival group for comparison. RESULTS: 366 of 457 hemodialysis patients were infected with SARS-CoV-2. The most common symptoms observed were fever (43.2%) and cough (29.8%), Followed by diarrhea (1.4%). Hemodialysis patients with hypertension were more susceptible to SARS-CoV-2 infection. The lymphocyte count, serum creatinine, serum potassium, and serum phosphorus in the SARS-CoV-2-positive group were significantly lower than those in the SARS-CoV-2-negative group. The all-cause mortality rate for infection with SARS-CoV-2 was 5.2%. Only 7 of 366 SARS-CoV-2-positive patients were admitted to the intensive care unit, but 6 of them died. Intensive care unit hospitalization rates were significantly higher in the non-survival group compared with the survival group. White blood cells count, neutrophil count, C-reactive protein, AST, and D-dimer in the non-survival group were higher than those in the survival group. The lymphocyte count, hemoglobin concentration, serum creatinine, serum albumin, serum phosphorus and parathyroid hormone in the non-survival group were lower than those in the survival group. Age > 65 years, elevated C-reactive protein and AST are independent risk factors for death. Finally, no significant difference in vaccination status was found between the SARS-CoV-2-positive group and the negative group. CONCLUSIONS: Hemodialysis patients are at high risk for SARS-CoV-2 infection. Ensuring the adequacy of hemodialysis treatment and maintaining good physical condition of patients are the top priorities.

3D-hUMSCs Exosomes Ameliorate Vitiligo by Simultaneously Potentiating Treg Cells-Mediated Immunosuppression and Suppressing Oxidative Stress-Induced Melanocyte Damage.

Vitiligo is an autoimmune disease characterized by epidermal melanocyte destruction, with abnormal autoimmune responses and excessive oxidative stress as two cardinal mechanisms. Human umbilical mesenchymal stem cells-derived exosomes (hUMSCs-Exos) are regarded as promising therapeutic choice for autoimmune diseases due to potent immunosuppressive and anti-oxidative properties, which can be potentiated under 3D cell culture condition. Nevertheless, whether exosomes derived from 3D spheroids of hUMSCs (3D-Exos) exhibit considerable therapeutic effect on vitiligo and the underlying mechanism remain elusive. In this study, systemic administration of 3D-Exos showed a remarkable effect in treating mice with vitiligo, as revealed by ameliorated skin depigmentation, less CD8+T cells infiltration, and expanded Treg cells in skin, and 3D-Exos exerted a better effect than 2D-Exos. Mechanistically, 3D-Exos can prominently facilitate the expansion of Treg cells in vitiligo lesion and suppress H2O2-induced melanocytes apoptosis. Forward miRNA profile analysis and molecular experiments have demonstrated that miR-132-3p and miR-125b-5p enriched in 3D-Exos greatly contributed to these biological effects by targeting Sirt1 and Bak1 respectively. In aggregate, 3D-Exos can efficiently ameliorate vitiligo by simultaneously potentiating Treg cells-mediated immunosuppression and suppressing oxidative stress-induced melanocyte damage via the delivery of miR-132-3p and miR-125b-5p. The employment of 3D-Exos will be a promising treament for vitiligo.

Leveraging DNA-Encoded Cell-Mimics for Environment-Adaptive Transmembrane Channel Release-Induced Cell Death.

The advancement of cell-mimic materials, which can forge sophisticated physicochemical dialogues with living cells, has unlocked a realm of intriguing prospects within the fields of synthetic biology and biomedical engineering. Inspired by the evolutionarily acquired ability of T lymphocytes to release perforin and generate transmembrane channels on targeted cells for killing, herein we present a pioneering DNA-encoded artificial T cell mimic model (ARTC) that accurately mimics T-cell-like behavior. ARTC responds to acidic conditions similar to those found in the tumor microenvironment and then selectively releases a G-rich DNA strand (LG4) embedded with C12 lipid and cholesterol molecules. Once released, LG4 effectively integrates into the membranes of neighboring live cells, behaving as an artificial transmembrane channel that selectively transports K+ ions and disrupts cellular homeostasis, ultimately inducing apoptosis. We hope that the emergence of ARTC will usher in new perspectives for revolutionizing future disease treatment and catalyzing the development of advanced biomedical technologies.

Long-term toxicity evaluation of aristolochic acid-IIIa in mice.

Aristolochic acid (AA)-IIIa is an AA analog present in Aristolochiaceae plants. To evaluate the chronic toxicity of AA-IIIa, mice were intragastrically administered with media control, 1 mg/kg AA-IIIa, and 10 mg/kg AA-IIIa, and designated as the control (CTL), AA-IIIa low dose (AA-IIIa-L), and AA-IIIa high dose (AA-IIIa-H) groups, respectively. AA-IIIa was administered three times a week, every other day, for 24 weeks (24-week time point). Thereafter, some mice were sacrificed immediately, while others were sacrificed 29 or 50 weeks after AA-IIIa withdrawal (53- or 74-week time point). Serum and organs were collected for biochemical and pathological analyses, respectively. Whole-genome sequencing was performed on the kidney, liver, and stomach tissues of AA-IIIa-treated mice for single-nucleotide polymorphism (SNP) detection. AA-IIIa-H mice died at 66 weeks, and the remaining mice showed moribund conditions at the 69 weeks. AA-IIIa induced minor kidney tubule injury, fibroblast hyperplasia, and forestomach carcinoma in mice. Bladder, intestine, liver, heart, spleen, lung, and testis tissues were not pathologically altered by AA-IIIa. In addition, AA-IIIa increased the C:G > A:T mutation in the kidney; however, no SNP mutation changes were observed in the liver and forestomach tissues of AA-IIIa-H mice at the 24-week time point compared with control mice. Therefore, we suspect that AA-IIIa is potentially mutagenic for mice after overdose and long-term administration. On the other hand, the forestomach is a unique organ in mice, but it does not exist in humans; thus, we hypothesize that the stomach toxicity induced by AA-IIIa is not a suitable reference for toxicological evaluation in humans. We recommend that Aristolochiaceae plants containing AA-IIIa should be properly supervised, and overdosing and long-term administration of drugs containing AA-IIIa should be avoided.

Oxidative stress-induced hypermethylation and low expression of ANXA2R: Novel insights into the dysfunction of melanocytes in vitiligo.

BACKGROUND: Vitiligo is a skin disorder with melanocyte destruction caused by complex interplay between multiple genetic and environmental factors. Recent studies have suggested DNA methylation is involved in the melanocyte damage, but the underlying mechanism remains unknown. OBJECTIVE: To explore the abnormal DNA methylation patterns in vitiligo lesional and nonlesional skin, and the mechanism of DNA methylation involved in vitiligo pathogenesis. METHODS: Initially, the genome-wide aberrant DNA methylation profiles in lesional and nonlesional skin of vitiligo were detect via Illumina methylation EPIC 850k Beadchip. Subsequently, a comprehensive analysis was conduct to investigate the genomic characteristics of differentially methylated regions (DMRs). Furthermore, the effects of key aberrant methylated genes on cell apoptosis and function of both melanocytes and keratinocytes were further identified and validated by western bloting, ELISA, and immunofluorescence. RESULTS: Compared with nonlesional skins, we discovered 79 significantly differentially methylated CpG sites in vitiligo lesions. These DMRs were mainly located in the gene body and the TS1500 region. Annexin A2 receptor (ANXA2R), a crucial gene in cell apoptosis, was hypermethylated in vitiligo lesions. Furthermore, we showed that ANXA2R displayed hypermethylation and low expression levels in both keratinocytes and melanocytes of vitiligo patients, and the hypermethylated-triggered downregulation of ANXA2R under oxidative stress induced melanocyte apoptosis, and inhibited the secretion of stem cell factor (SCF) from keratinocytes thus impaired the survival of melanocytes. CONCLUSIONS: Our study illustrates the DNA methylation modification in vitiligo, and further demonstrates the molecular mechanism of hypermethylated ANXA2R in the dysfunction of melanocytes under oxidative stress.

SIRT7 promotes mitochondrial biogenesis to render the adaptive resistance to MAPK inhibition in melanoma.

Melanoma, arising from the malignant transformation of melanocytes, stands as the most lethal type of skin cancer. While significant strides have been made in targeted therapy and immunotherapy, substantially enhancing therapeutic efficacy, the prognosis for melanoma patients remains unoptimistic. SIRT7, a nuclear-localized deacetylase, plays a pivotal role in maintaining cellular homeostasis and adapting to external stressors in melanoma, with its activity closely tied to intracellular nicotinamide adenine dinucleotide (NAD+). However, its involvement in adaptive resistance to targeted therapy remains unclear. Herein, we unveil that up-regulated SIRT7 promotes mitochondrial biogenesis to render the adaptive resistance to MAPK inhibition in melanoma. Initially, we observed a significant increase of SIRT7 expression in publicly available datasets following targeted therapy within a short duration. In consistent, we found elevated SIRT7 expression in melanoma cells subjected to BRAF or MEK inhibitors in vitro. The up-regulation of SIRT7 expression was also confirmed in xenograft tumors in mice after targeted therapy in vivo. Furthermore, we proved that SIRT7 deficiency led to decreased cell viability upon prolonged exposure to BRAF or MEK inhibitors, accompanied by an increase in cell apoptosis. Mechanistically, SIRT7 deficiency restrained the upregulation of genes associated with mitochondrial biogenesis and intracellular ATP levels in response to targeted therapy treatment in melanoma cells. Ultimately, we proved that SIRT7 deficieny could sensitize BRAF-mutant melanoma cells to MAPK inhibition targeted therapy in vivo. In conclusion, our findings underscore the role of SIRT7 in fostering adaptive resistance to targeted therapy through the facilitation of mitochondrial biogenesis. Targeting SIRT7 emerges as a promising strategy to overcome MAPK inhibitor adaptive resistance in melanoma.

Combined PET/CT and Autoantibody Detection in Lung Nodules Diagnosis.

Objective: To assess the usefulness of combining positron emission tomography/computed tomography (PET/CT) with lung cancer autoantibody detection in identifying and managing lung nodules. Methods: The researchers identified 160 patients with pulmonary nodules admitted to their hospital between January 2018 and January 2021. These patients were designated as the experimental group. Additionally, 60 healthy individuals without pulmonary nodules were admitted to the hospital during the same period. The individuals constituted the control group. All study participants underwent digital PET/CT detection and had their lung cancer autoantibody levels determined through enzyme-linked immunosorbent assay. Further testing, such as puncture or surgical pathology, was performed for patients with lung nodules. The aim was to evaluate the significance of combining PET/CT with autoantibody detection in diagnosing and treating lung nodules. Results: The study found that testing multiple autoantibodies together increased sensitivity and accuracy compared to testing individual autoantibodies. Combining PET/CT screening with autoantibody detection improved the diagnostic rate for identifying lung nodules, including benign and suspected malignant ones. Several autoantibodies were significantly higher in the experimental group compared to the control group. Testing for multiple autoantibodies showed higher sensitivity and accuracy than testing for one. Pathological examination confirmed 129 benign nodules and 31 malignant nodules. The median SUVmax values were measured at 0.7 for benign nodules and 4.8 for malignant nodules. The diagnostic efficacy of PET/CT combined with autoantibodies was determined through comparison with pathology testing and was as follows: PET/CT combined with autoantibody detection > PET/CT > autoantibody detection. Conclusion: Combining PET/CT with the detection of autoantibodies enhances the positive diagnostic rate and accuracy of lung nodules in the case of lung cancer. The SUVmax also shows excellent potential as a supplement in diagnosing both benign and malignant lung nodules, providing valuable guidance in determining the pathological types.

Long-term oral administration of Kelisha capsule does not cause hepatorenal toxicity in rats.

ETHNOPHARMACOLOGICAL RELEVANCE: Kelisha capsules (KLS) are often used to treat acute diarrhoea, bacillary dysentery, heat stroke, and other diseases. One of its components, Asarum, contains aristolochic acid I which is both nephrotoxic and carcinogenic. However, the aristolochic acid (AA) content in KLS and its toxicity remain unclear. AIM OF THE STUDY: The aims of this study were to quantitatively determine the contents of five aristolochic acid analogues (AAAs) in Asarum and KLS, and systematically evaluate the in vivo toxicity of KLS in rats. MATERIALS AND METHODS: Ultra-performance liquid chromatography-tandem mass spectrometry (UPLC-MS/MS) was used to determine the content of the five AAAs in Asarum and KLS. Sprague-Dawley rats were administered KLS at 0, 0.75, 1.5, and 3.0 g/kg respectively, and then sacrificed after 4 weeks of administration or after an additional 2 weeks of recovery. The endpoints assessed included body weight measurements, serum biochemistry and haematology indices, and clinical and histopathological observations. RESULTS: The AAAs content in Asarum sieboldii Miq. (HB-ESBJ) were much lower than those of the other Asarums. The contents of AA I, AA IVa, and aristolactam I in KLS were in the ranges of 0.03-0.06 µg/g, 1.89-2.16 µg/g, and 0.55-1.60 µg/g, respectively, whereas AA II and AA IIIa were not detected. None of the rats showed symptoms of toxic reactions and KLS was well tolerated throughout the study. Compared to the control group, the activated partial thromboplastin time values of rats in the 1.5 and 3.0 g/kg groups significantly reduced after administration (P < 0.05). In addition, the serum triglycerides of male rats in the 0.75 and 1.5 g/kg groups after administration, and the 0.75, 1.5, 3.0 g/kg groups after recovery were significantly decreased (P < 0.01 or P < 0.001). No significant drug-related toxicological changes were observed in other serum biochemical indices, haematology, or histopathology. CONCLUSIONS: The AA I content in KLS met the limit requirements (<0.001%) of the Chinese Pharmacopoeia. Therefore, it is safe to use KLS in the short-term. However, for safety considerations, attention should be paid to the effects of long-term KLS administration on coagulation function and triglyceride metabolism.

Melanocyte stem cells in the skin: Origin, biological characteristics, homeostatic maintenance and therapeutic potential.

Melanocyte stem cells (MSCs), melanocyte lineage-specific skin stem cells derived from the neural crest, are observed in the mammalian hair follicle, the epidermis or the sweat gland. MSCs differentiate into mature melanin-producing melanocytes, which confer skin and hair pigmentation and uphold vital skin functions. In controlling and coordinating the homeostasis, repair and regeneration of skin tissue, MSCs play a vital role. Decreased numbers or impaired functions of MSCs are closely associated with the development and therapy of many skin conditions, such as hair graying, vitiligo, wound healing and melanoma. With the advancement of stem cell technology, the relevant features of MSCs have been further elaborated. In this review, we provide an exhaustive overview of cutaneous MSCs and highlight the latest advances in MSC research. A better understanding of the biological characteristics and micro-environmental regulatory mechanisms of MSCs will help to improve clinical applications in regenerative medicine, skin pigmentation disorders and cancer therapy. KEY POINTS: This review provides a concise summary of the origin, biological characteristics, homeostatic maintenance and therapeutic potential of cutaneous MSCs. The role and potential application value of MSCs in skin pigmentation disorders are discussed. The significance of single-cell RNA sequencing, CRISPR-Cas9 technology and practical models in MSCs research is highlighted.

Identification of sensitive endpoints for the assessment of phthalates-induced reproductive and developmental toxicity: A literature mining study.

Dibutyl phthalate (DBP) and di(2-ethylhexyl) phthalate (DEHP), two common types of phthalates, are known to cause reproductive and developmental toxicity in animals and humans. The reference doses (RfD) of DBP and DEHP should be determined by sensitive endpoints. We here aimed to identify sensitive endpoints for DBP- and DEHP-induced such toxicity using published literatures. By examining the impacts of maternal exposure to DBP or DEHP on anogenital distance (AGD) and semen quality of offspring, we discovered that DBP or DEHP caused AGD decline in boys but increase in girls with DBP being more potent and the first 14weeks of pregnancy being more susceptible, suggesting a chemical- and time-dependent phenomenon. We also identified AGD shortening and total sperm count reduction as two sensitive endpoints for DBP- or DEHP-induced reproductive and developmental toxicity, respectively. Based upon these two endpoints and the employment of the Bayesian benchmark dose approach with an uncertainty factor of 3,000, we estimated the RfD values of DBP and DEHP were 15 µg/kg/day and 36 µg/kg/day, respectively. Thus, we uncover previously unrecognized phenomena of DBP- or DEHP-induced reproductive and developmental toxicity and establish new and comparable or more conservative RfDs for the risk assessment of phthalates exposure in humans.

Optimizing vitiligo diagnosis with ResNet and Swin transformer deep learning models: a study on performance and interpretability.

Vitiligo is a hypopigmented skin disease characterized by the loss of melanin. The progressive nature and widespread incidence of vitiligo necessitate timely and accurate detection. Usually, a single diagnostic test often falls short of providing definitive confirmation of the condition, necessitating the assessment by dermatologists who specialize in vitiligo. However, the current scarcity of such specialized medical professionals presents a significant challenge. To mitigate this issue and enhance diagnostic accuracy, it is essential to build deep learning models that can support and expedite the detection process. This study endeavors to establish a deep learning framework to enhance the diagnostic accuracy of vitiligo. To this end, a comparative analysis of five models including ResNet (ResNet34, ResNet50, and ResNet101 models) and Swin Transformer series (Swin Transformer Base, and Swin Transformer Large models), were conducted under the uniform condition to identify the model with superior classification capabilities. Moreover, the study sought to augment the interpretability of these models by selecting one that not only provides accurate diagnostic outcomes but also offers visual cues highlighting the regions pertinent to vitiligo. The empirical findings reveal that the Swin Transformer Large model achieved the best performance in classification, whose AUC, accuracy, sensitivity, and specificity are 0.94, 93.82%, 94.02%, and 93.5%, respectively. In terms of interpretability, the highlighted regions in the class activation map correspond to the lesion regions of the vitiligo images, which shows that it effectively indicates the specific category regions associated with the decision-making of dermatological diagnosis. Additionally, the visualization of feature maps generated in the middle layer of the deep learning model provides insights into the internal mechanisms of the model, which is valuable for improving the interpretability of the model, tuning performance, and enhancing clinical applicability. The outcomes of this study underscore the significant potential of deep learning models to revolutionize medical diagnosis by improving diagnostic accuracy and operational efficiency. The research highlights the necessity for ongoing exploration in this domain to fully leverage the capabilities of deep learning technologies in medical diagnostics.