Modeling naturalistic face processing in humans with deep convolutional neural networks.

Deep convolutional neural networks (DCNNs) trained for face identification can rival and even exceed human-level performance. The ways in which the internal face representations in DCNNs relate to human cognitive representations and brain activity are not well understood. Nearly all previous studies focused on static face image processing with rapid display times and ignored the processing of naturalistic, dynamic information. To address this gap, we developed the largest naturalistic dynamic face stimulus set in human neuroimaging research (700+ naturalistic video clips of unfamiliar faces). We used this naturalistic dataset to compare representational geometries estimated from DCNNs, behavioral responses, and brain responses. We found that DCNN representational geometries were consistent across architectures, cognitive representational geometries were consistent across raters in a behavioral arrangement task, and neural representational geometries in face areas were consistent across brains. Representational geometries in late, fully connected DCNN layers, which are optimized for individuation, were much more weakly correlated with cognitive and neural geometries than were geometries in late-intermediate layers. The late-intermediate face-DCNN layers successfully matched cognitive representational geometries, as measured with a behavioral arrangement task that primarily reflected categorical attributes, and correlated with neural representational geometries in known face-selective topographies. Our study suggests that current DCNNs successfully capture neural cognitive processes for categorical attributes of faces but less accurately capture individuation and dynamic features.

CellGO: a novel deep learning-based framework and webserver for cell-type-specific gene function interpretation.

Interpreting the function of genes and gene sets identified from omics experiments remains a challenge, as current pathway analysis tools often fail to consider the critical biological context, such as tissue or cell-type specificity. To address this limitation, we introduced CellGO. CellGO tackles this challenge by leveraging the visible neural network (VNN) and single-cell gene expressions to mimic cell-type-specific signaling propagation along the Gene Ontology tree within a cell. This design enables a novel scoring system to calculate the cell-type-specific gene-pathway paired active scores, based on which, CellGO is able to identify cell-type-specific active pathways associated with single genes. In addition, by aggregating the activities of single genes, CellGO extends its capability to identify cell-type-specific active pathways for a given gene set. To enhance biological interpretation, CellGO offers additional features, including the identification of significantly active cell types and driver genes and community analysis of pathways. To validate its performance, CellGO was assessed using a gene set comprising mixed cell-type markers, confirming its ability to discern active pathways across distinct cell types. Subsequent benchmarking analyses demonstrated CellGO's superiority in effectively identifying cell types and their corresponding cell-type-specific pathways affected by gene knockouts, using either single genes or sets of genes differentially expressed between knockout and control samples. Moreover, CellGO demonstrated its ability to infer cell-type-specific pathogenesis for disease risk genes. Accessible as a Python package, CellGO also provides a user-friendly web interface, making it a versatile and accessible tool for researchers in the field.

Identification of novel immune-related signatures for keloid diagnosis and treatment: insights from integrated bulk RNA-seq and scRNA-seq analysis.

BACKGROUND: Keloid is a disease characterized by proliferation of fibrous tissue after the healing of skin tissue, which seriously affects the daily life of patients. However, the clinical treatment of keloids still has limitations, that is, it is not effective in controlling keloids, resulting in a high recurrence rate. Thus, it is urgent to identify new signatures to improve the diagnosis and treatment of keloids. METHOD: Bulk RNA seq and scRNA seq data were downloaded from the GEO database. First, we used WGCNA and MEGENA to co-identify keloid/immune-related DEGs. Subsequently, we used three machine learning algorithms (Randomforest, SVM-RFE, and LASSO) to identify hub immune-related genes of keloid (KHIGs) and investigated the heterogeneous expression of KHIGs during fibroblast subpopulation differentiation using scRNA-seq. Finally, we used HE and Masson staining, quantitative reverse transcription-PCR, western blotting, immunohistochemical, and Immunofluorescent assay to investigate the dysregulated expression and the mechanism of retinoic acid in keloids. RESULTS: In the present study, we identified PTGFR, RBP5, and LIF as KHIGs and validated their diagnostic performance. Subsequently, we constructed a novel artificial neural network molecular diagnostic model based on the transcriptome pattern of KHIGs, which is expected to break through the current dilemma faced by molecular diagnosis of keloids in the clinic. Meanwhile, the constructed IG score can also effectively predict keloid risk, which provides a new strategy for keloid prevention. Additionally, we observed that KHIGs were also heterogeneously expressed in the constructed differentiation trajectories of fibroblast subtypes, which may affect the differentiation of fibroblast subtypes and thus lead to dysregulation of the immune microenvironment in keloids. Finally, we found that retinoic acid may treat or alleviate keloids by inhibiting RBP5 to differentiate pro-inflammatory fibroblasts (PIF) to mesenchymal fibroblasts (MF), which further reduces collagen secretion. CONCLUSION: In summary, the present study provides novel immune signatures (PTGFR, RBP5, and LIF) for keloid diagnosis and treatment, and identifies retinoic acid as potential anti-keloid drugs. More importantly, we provide a new perspective for understanding the interactions between different fibroblast subtypes in keloids and the remodeling of their immune microenvironment.

Development of novel lysosome-related signatures and their potential target drugs based on bulk RNA-seq and scRNA-seq for diabetic foot ulcers.

BACKGROUND: Diabetic foot ulcers (DFU) is the most serious complication of diabetes mellitus, which has become a global health problem due to its high morbidity and disability rates and the poor efficacy of conventional treatments. Thus, it is urgent to identify novel molecular targets to improve the prognosis and reduce disability rate in DFU patients. RESULTS: In the present study, bulk RNA-seq and scRNA-seq associated with DFU were downloaded from the GEO database. We identified 1393 DFU-related DEGs by differential analysis and WGCNA analysis together, and GO/KEGG analysis showed that these genes were associated with lysosomal and immune/inflammatory responses. Immediately thereafter, we identified CLU, RABGEF1 and ENPEP as DLGs for DFU using three machine learning algorithms (Randomforest, SVM-RFE and LASSO) and validated their diagnostic performance in a validation cohort independent of this study. Subsequently, we constructed a novel artificial neural network model for molecular diagnosis of DFU based on DLGs, and the diagnostic performance in the training and validation cohorts was sound. In single-cell sequencing, the heterogeneous expression of DLGs also provided favorable evidence for them to be potential diagnostic targets. In addition, the results of immune infiltration analysis showed that the abundance of mainstream immune cells, including B/T cells, was down-regulated in DFUs and significantly correlated with the expression of DLGs. Finally, we found latamoxef, parthenolide, meclofenoxate, and lomustine to be promising anti-DFU drugs by targeting DLGs. CONCLUSIONS: CLU, RABGEF1 and ENPEP can be used as novel lysosomal molecular signatures of DFU, and by targeting them, latamoxef, parthenolide, meclofenoxate and lomustine were identified as promising anti-DFU drugs. The present study provides new perspectives for the diagnosis and treatment of DFU and for improving the prognosis of DFU patients.

Artificial neural network-based gene screening and immune cell infiltration analysis of osteosarcoma feature.

BACKGROUND: The present study aimed to construct an artificial neural network (ANN) model that leverages characteristic genes associated with osteosarcoma (OS) to enable accurate prognostication for OS patients. METHODS: Our research revealed 467 differentially expressed genes (DEGs) via gene expression contrast analysis, consisting of 345 downregulated genes and 122 upregulated genes. Gene Ontology (GO) enrichment analysis illuminated functions primarily encompassing T-cell activation, secretory granule lumen and antioxidant activity, among others. Through Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis, we discovered significant correlations between the DEGs and certain pathways, including phagosome, Staphylococcus aureus infection and human T-cell leukemia virus 1 infection. We then screened out 30 characteristic DEGs (CDEGs) based on random forest analysis and constructed the ANN model using the gene score matrix. To verify the credibility and accuracy of the ANN model, we performed internal and external validation processes, which affirmed our model's predictive capabilities. RESULTS: The study further delved into the analysis of immune cell infiltration and its correlation with the target CDEGs, revealing disparities in the infiltration of 22 types of immune cells across different groups and their interrelationships. Moreover, we probed the expression of the two foremost CDEGs (YES1 and MFNG) in OS and normal tissues. We noted a positive relationship between the expression of YES1 and MFNG in OS tissues and the clinicopathological characteristics of OS patients. CONCLUSIONS: Collectively, the findings of the present study validate the effectiveness of the CDEGs-based ANN model in predicting OS patients, which might facilitate early diagnosis and treatment of OS.

GeTe/MoTe2 Van der Waals Heterostructures: Enabling Ultralow Voltage Memristors for Nonvolatile Memory and Neuromorphic Computing Applications.

Advanced electronic semiconducting Van der Waals heterostructures (HSs) are promising candidates for exploring next-generation nanoelectronics owing to their exceptional electronic properties, which present the possibility of extending their functionalities to diverse potential applications. In this study, GeTe/MoTe2 HS are explored for nonvolatile memory and neuromorphic-computing applications. Sputter-deposited Ag/GeTe/MoTe2/Pt HS cross-point devices are fabricated, and they demonstrate memristor behavior at ultralow switching voltages (VSET: 0.15 V and VRESET: -0.14 V) with very low energy consumption (≈30 nJ), high memory window, long retention time (104 s), and excellent endurance (105 cycles). Resistive switching is achieved by adjusting the interface between the Ag top electrode and the heterojunction switching layer. Cross-sectional transmission electron microscope images and conductive atomic force microscopy analysis confirm the presence of a conducting filament in the heterojunction switching layer. Further, emulating various synaptic functions of a biological synapse reveals that GeTe/MoTe2 HS can be utilized for energy-efficient neuromorphic-computing applications. A multilayer perceptron is implemented using the synaptic weights of the Ag/GeTe/MoTe2/Pt HS device, revealing high pattern accuracy (81.3%). These results indicate that HS devices can be considered a potential solution for high-density memory and artificial intelligence applications.

Endoplasmic reticulum stress promotes hepatocellular carcinoma by modulating immunity: a study based on artificial neural networks and single-cell sequencing.

INTRODUCTION: Hepatocellular carcinoma (HCC) is characterized by the complex pathogenesis, limited therapeutic methods, and poor prognosis. Endoplasmic reticulum stress (ERS) plays an important role in the development of HCC, therefore, we still need further study of molecular mechanism of HCC and ERS for early diagnosis and promising treatment targets. METHOD: The GEO datasets (GSE25097, GSE62232, and GSE65372) were integrated to identify differentially expressed genes related to HCC (ERSRGs). Random Forest (RF) and Support Vector Machine (SVM) machine learning techniques were applied to screen ERSRGs associated with endoplasmic reticulum stress, and an artificial neural network (ANN) diagnostic prediction model was constructed. The ESTIMATE algorithm was utilized to analyze the correlation between ERSRGs and the immune microenvironment. The potential therapeutic agents for ERSRGs were explored using the Drug Signature Database (DSigDB). The immunological landscape of the ERSRGs central gene PPP1R16A was assessed through single-cell sequencing and cell communication, and its biological function was validated using cytological experiments. RESULTS: An ANN related to the ERS model was constructed based on SRPX, THBS4, CTH, PPP1R16A, CLGN, and THBS1. The area under the curve (AUC) of the model in the training set was 0.979, and the AUC values in three validation sets were 0.958, 0.936, and 0.970, respectively, indicating high reliability and effectiveness. Spearman correlation analysis suggests that the expression levels of ERSRGs are significantly correlated with immune cell infiltration and immune-related pathways, indicating their potential as important targets for immunotherapy. Mometasone was predicted to be the most promising treatment drug based on its highest binding score. Among the six ERSRGs, PPP1R16A had the highest mutation rate, predominantly copy number mutations, which may be the core gene of the ERSRGs model. Single-cell analysis and cell communication indicated that PPP1R16A is predominantly distributed in liver malignant parenchymal cells and may reshape the tumor microenvironment by enhancing macrophage migration inhibitory factor (MIF)/CD74 + CXCR4 signaling pathways. Functional experiments revealed that after siRNA knockdown, the expression of PPP1R16A was downregulated, which inhibited the proliferation, migration, and invasion capabilities of HCCLM3 and Hep3B cells in vitro. CONCLUSION: The consensus of various machine learning algorithms and artificial intelligence neural networks has established a novel predictive model for the diagnosis of liver cancer associated with ERS. This study offers a new direction for the diagnosis and treatment of HCC.

Predicting dynamic, motion-related changes in B0 field in the brain at a 7T MRI using a subject-specific fine-trained U-net.

PURPOSE: Subject movement during the MR examination is inevitable and causes not only image artifacts but also deteriorates the homogeneity of the main magnetic field (B0 ), which is a prerequisite for high quality data. Thus, characterization of changes to B0 , for example induced by patient movement, is important for MR applications that are prone to B0 inhomogeneities. METHODS: We propose a deep learning based method to predict such changes within the brain from the change of the head position to facilitate retrospective or even real-time correction. A 3D U-net was trained on in vivo gradient-echo brain 7T MRI data. The input consisted of B0 maps and anatomical images at an initial position, and anatomical images at a different head position (obtained by applying a rigid-body transformation on the initial anatomical image). The output consisted of B0 maps at the new head positions. We further fine-trained the network weights to each subject by measuring a limited number of head positions of the given subject, and trained the U-net with these data. RESULTS: Our approach was compared to established dynamic B0 field mapping via interleaved navigators, which suffer from limited spatial resolution and the need for undesirable sequence modifications. Qualitative and quantitative comparison showed similar performance between an interleaved navigator-equivalent method and proposed method. CONCLUSION: It is feasible to predict B0 maps from rigid subject movement and, when combined with external tracking hardware, this information could be used to improve the quality of MR acquisitions without the use of navigators.

Development and validation of a machine learning-based model to predict isolated post-challenge hyperglycemia in middle-aged and elder adults: Analysis from a multicentric study.

INTRODUCTION: Due to the high cost and complexity, the oral glucose tolerance test is not adopted as the screening method for identifying diabetes patients, which leads to the misdiagnosis of patients with isolated post-challenge hyperglycemia (IPH), that is., patients with normal fasting plasma glucose (<7.0 mmoL/L) and abnormal 2-h postprandial blood glucose (≥11.1 mmoL/L). We aimed to develop a model to differentiate individuals with IPH from the normal population. METHODS: Data from 54301 eligible participants were obtained from the Risk Evaluation of Cancers in Chinese Diabetic Individuals: a longitudinal (REACTION) study in China. Data from 37740 participants were used to develop the diagnostic system. External validation was performed among 16561 participants. Three machine learning algorithms were used to create the predictive models, which were further evaluated by various classification algorithms to establish the best predictive model. RESULTS: Ten features were selected to develop an IPH diagnosis system (IPHDS) based on an artificial neural network. In external validation, the AUC of the IPHDS was 0.823 (95% CI 0.811-0.836), which was significantly higher than the AUC of the Taiwan model [0.799 (0.786-0.813)] and that of the Chinese Diabetes Risk Score model [0.648 (0.635-0.662)]. The IPHDS model had a sensitivity of 75.6% and a specificity of 74.6%. This model outperformed the Taiwan and CDRS models in subgroup analyses. An online site with instant predictions was deployed at https://app-iphds-e1fc405c8a69.herokuapp.com/. CONCLUSIONS: The proposed IPHDS could be a convenient and user-friendly screening tool for diabetes during health examinations in a large general population.

Elevated liver enzymes in the first trimester are associated with gestational diabetes mellitus: A prospective cohort study.

AIMS: Previous studies have found that a single liver enzyme may predict gestational diabetes mellitus (GDM), but the results have been inconsistent. This study aimed to explore the associations of liver enzymes in early pregnancy with risk of GDM, as well as to independently rank risk factors. METHODS: This prospective cohort study included 1295 women who underwent liver enzyme measurements during early pregnancy and completed GDM assessment in mid-pregnancy. Logistic regression and restricted cubic spline analyses were conducted to assess the relationship between liver enzymes and risk of GDM. Back-propagation artificial neural network was performed to rank independently risk factors of GDM. RESULTS: Women diagnosed with GDM exhibited significantly higher levels of liver enzymes than those without GDM (all p < 0.05). The highest quartile of liver enzymes was associated with higher risk of GDM compared with the lowest quartile, with adjusted odds ratio (ORs) ranging from 2.76 to 8.11 (all p < 0.05). Moreover, the ORs of GDM increased linearly with liver enzymes level (all P for overall association <0.001). Furthermore, Back-propagation artificial neural network identified γ-gamma-glutamyl transferase (GGT) as accounting for the highest proportion in the ranking of GDM risk prediction weights (up to 20.8%). CONCLUSIONS: Single or total elevations of liver enzymes in early pregnancy could predict the GDM occurrence, in which GGT, alkaline Phosphatase, and aspartate aminotransferase were the three most important independent risk factors.

Integrated clinical and genomic models using machine-learning methods to predict the efficacy of paclitaxel-based chemotherapy in patients with advanced gastric cancer.

BACKGROUND: Paclitaxel is commonly used as a second-line therapy for advanced gastric cancer (AGC). The decision to proceed with second-line chemotherapy and select an appropriate regimen is critical for vulnerable patients with AGC progressing after first-line chemotherapy. However, no predictive biomarkers exist to identify patients with AGC who would benefit from paclitaxel-based chemotherapy. METHODS: This study included 288 patients with AGC receiving second-line paclitaxel-based chemotherapy between 2017 and 2022 as part of the K-MASTER project, a nationwide government-funded precision medicine initiative. The data included clinical (age [young-onset vs. others], sex, histology [intestinal vs. diffuse type], prior trastuzumab use, duration of first-line chemotherapy), and genomic factors (pathogenic or likely pathogenic variants). Data were randomly divided into training and validation sets (0.8:0.2). Four machine learning (ML) methods, namely random forest (RF), logistic regression (LR), artificial neural network (ANN), and ANN with genetic embedding (ANN with GE), were used to develop the prediction model and validated in the validation sets. RESULTS: The median patient age was 64 years (range 25-91), and 65.6% of those were male. A total of 288 patients were divided into the training (n = 230) and validation (n = 58) sets. No significant differences existed in baseline characteristics between the training and validation sets. In the training set, the areas under the ROC curves (AUROC) for predicting better progression-free survival (PFS) with paclitaxel-based chemotherapy were 0.499, 0.679, 0.618, and 0.732 in the RF, LR, ANN, and ANN with GE models, respectively. The ANN with the GE model that achieved the highest AUROC recorded accuracy, sensitivity, specificity, and F1-score performance of 0.458, 0.912, 0.724, and 0.579, respectively. In the validation set, the ANN with GE model predicted that paclitaxel-sensitive patients had significantly longer PFS (median PFS 7.59 vs. 2.07 months, P = 0.020) and overall survival (OS) (median OS 14.70 vs. 7.50 months, P = 0.008). The LR model predicted that paclitaxel-sensitive patients showed a trend for longer PFS (median PFS 6.48 vs. 2.33 months, P = 0.078) and OS (median OS 12.20 vs. 8.61 months, P = 0.099). CONCLUSIONS: These ML models, integrated with clinical and genomic factors, offer the possibility to help identify patients with AGC who may benefit from paclitaxel chemotherapy.

Retrospective validation study of an artificial neural network-based preoperative decision-support tool for noninvasive lymph node staging (NILS) in women with primary breast cancer (ISRCTN14341750).

BACKGROUND: Surgical sentinel lymph node biopsy (SLNB) is routinely used to reliably stage axillary lymph nodes in early breast cancer (BC). However, SLNB may be associated with postoperative arm morbidities. For most patients with BC undergoing SLNB, the findings are benign, and the procedure is currently questioned. A decision-support tool for the prediction of benign sentinel lymph nodes based on preoperatively available data has been developed using artificial neural network modelling. METHODS: This was a retrospective geographical and temporal validation study of the noninvasive lymph node staging (NILS) model, based on preoperatively available data from 586 women consecutively diagnosed with primary BC at two sites. Ten preoperative clinicopathological characteristics from each patient were entered into the web-based calculator, and the probability of benign lymph nodes was predicted. The performance of the NILS model was assessed in terms of discrimination with the area under the receiver operating characteristic curve (AUC) and calibration, that is, comparison of the observed and predicted event rates of benign axillary nodal status (N0) using calibration slope and intercept. The primary endpoint was axillary nodal status (discrimination, benign [N0] vs. metastatic axillary nodal status [N+]) determined by the NILS model compared to nodal status by definitive pathology. RESULTS: The mean age of the women in the cohort was 65 years, and most of them (93%) had luminal cancers. Approximately three-fourths of the patients had no metastases in SLNB (N0 74% and 73%, respectively). The AUC for the predicted probabilities for the whole cohort was 0.6741 (95% confidence interval: 0.6255-0.7227). More than one in four patients (n = 151, 26%) were identified as candidates for SLNB omission when applying the predefined cut-off for lymph node-negative status from the development cohort. The NILS model showed the best calibration in patients with a predicted high probability of healthy axilla. CONCLUSION: The performance of the NILS model was satisfactory. In approximately every fourth patient, SLNB could potentially be omitted. Considering the shift from postoperatively to preoperatively available predictors in this validation study, we have demonstrated the robustness of the NILS model. The clinical usability of the web interface will be evaluated before its clinical implementation. TRIAL REGISTRATION: Registered in the ISRCTN registry with study ID ISRCTN14341750. Date of registration 23/11/2018.

Developing neural network diagnostic models and potential drugs based on novel identified immune-related biomarkers for celiac disease.

BACKGROUND: As one of the most common intestinal inflammatory diseases, celiac disease (CD) is typically characterized by an autoimmune disorder resulting from ingesting gluten proteins. Although the incidence and prevalence of CD have increased over time, the diagnostic methods and treatment options are still limited. Therefore, it is urgent to investigate the potential biomarkers and targeted drugs for CD. METHODS: Gene expression data was downloaded from GEO datasets. Differential gene expression analysis was performed to identify the dysregulated immune-related genes. Multiple machine algorithms, including randomForest, SVM-RFE, and LASSO, were used to select the hub immune-related genes (HIGs). The immune-related genes score (IG score) and artificial neural network (ANN) were constructed based on HIGs. Potential drugs targeting HIGs were identified by using the Enrichr platform and molecular docking method. RESULTS: We identified the dysregulated immune-related genes at a genome-wide level and demonstrated their roles in CD-related immune pathways. The hub genes (MR1, CCL25, and TNFSF13B) were further screened by integrating several machine algorithms. Meanwhile, the CD patients were divided into distinct subtypes with either high- or low-immunoactivity using single-sample gene set enrichment analysis (ssGSEA) and consensus clustering. By constructing IG score based on HIGs, we found that patients with high IG score were mainly attributed to high-immunoactivity subgroups, which suggested a strong link between HIGs and immunoactivity of CD patients. In addition, the novel constructed ANN model showed the sound diagnostic ability of HIGs. Mechanistically, we validated that the HIGs play pivotal roles in regulating CD's immune and inflammatory state. Through targeting the HIGs, we also found potential drugs for anti-CD treatment by using the Enrichr platform and molecular docking method. CONCLUSIONS: This study unveils the HIGs and elucidates the networks regulated by these genes in the context of CD. It underscores the pivotal significance of HIGs in accurately predicting the presence or absence of CD in patients. Consequently, this research offers promising prospects for the development of diagnostic biomarkers and therapeutic targets for CD.

Risk exploration and prediction model construction for linezolid-resistant Enterococcus faecalis based on big data in a province in southern China.

BACKGROUND: Enterococcus faecalis is a common cause of healthcare-associated infections. Its resistance to linezolid, the antibiotic of last resort for vancomycin-resistant enterococci, has become a growing threat in healthcare settings. METHODS: We analyzed the data of E. faecalis isolates from 26 medical institutions between 2018 and 2020 and performed univariate and multivariate logistic regression analyses to determine the independent predictors for linezolid-resistant E. faecalis (LREFs). Then, we used the artificial neural network (ANN) and logistic regression (LR) to build a prediction model for linezolid resistance and performed a performance evaluation and comparison. RESULTS: Of 12,089 E. faecalis strains, 755 (6.25%) were resistant to linezolid. Among vancomycin-resistant E. faecalis, the linezolid-resistant rate was 24.44%, higher than that of vancomycin-susceptible E. faecalis (p < 0.0001). Univariate and multivariate regression analyses showed that gender, age, specimen type, length of stay before culture, season, region, GDP (gross domestic product), number of beds, and hospital level were predictors of linezolid resistance. Both the ANN and LR models constructed in the study performed well in predicting linezolid resistance in E. faecalis, with AUCs of 0.754 and 0.741 in the validation set, respectively. However, synthetic minority oversampling technique (SMOTE) did not improve the prediction ability of the models. CONCLUSION: E. faecalis linezolid-resistant rates varied by specimen site, geographic region, GDP level, facility level, and the number of beds. At the same time, community-acquired E. faecalis with linezolid resistance should be monitored closely. We can use the prediction model to guide clinical medication and take timely prevention and control measures.

A novel non-negative Bayesian stacking modeling method for Cancer survival prediction using high-dimensional omics data.

BACKGROUND: Survival prediction using high-dimensional molecular data is a hot topic in the field of genomics and precision medicine, especially for cancer studies. Considering that carcinogenesis has a pathway-based pathogenesis, developing models using such group structures is a closer mimic of disease progression and prognosis. Many approaches can be used to integrate group information; however, most of them are single-model methods, which may account for unstable prediction. METHODS: We introduced a novel survival stacking method that modeled using group structure information to improve the robustness of cancer survival prediction in the context of high-dimensional omics data. With a super learner, survival stacking combines the prediction from multiple sub-models that are independently trained using the features in pre-grouped biological pathways. In addition to a non-negative linear combination of sub-models, we extended the super learner to non-negative Bayesian hierarchical generalized linear model and artificial neural network. We compared the proposed modeling strategy with the widely used survival penalized method Lasso Cox and several group penalized methods, e.g., group Lasso Cox, via simulation study and real-world data application. RESULTS: The proposed survival stacking method showed superior and robust performance in terms of discrimination compared with single-model methods in case of high-noise simulated data and real-world data. The non-negative Bayesian stacking method can identify important biological signal pathways and genes that are associated with the prognosis of cancer. CONCLUSIONS: This study proposed a novel survival stacking strategy incorporating biological group information into the cancer prognosis models. Additionally, this study extended the super learner to non-negative Bayesian model and ANN, enriching the combination of sub-models. The proposed Bayesian stacking strategy exhibited favorable properties in the prediction and interpretation of complex survival data, which may aid in discovering cancer targets.

Artificial Neural Network Analysis of Determinants of Tacrolimus Pharmacokinetics in Liver Transplant Recipients.

BACKGROUND: The efficacy and toxicity of tacrolimus are closely related to its trough blood concentrations. Identifying the influencing factors of pharmacokinetics of tacrolimus in the early postoperative period is conducive to the optimization of the individualized tacrolimus administration protocol and to help liver transplant (LT) recipients achieve the target blood concentrations. OBJECTIVE: This study aimed to develop an artificial neural network (ANN) for predicting the blood concentration of tacrolimus soon after liver transplantation and for identifying determinants of the concentration based on Shapley additive explanation (SHAP). METHODS: In this retrospective study, we enrolled 31 recipients who were first treated with liver transplantation from the Department of Liver Transplantation and Hepatic Surgery, the First Affiliated Hospital of Shandong First Medical University (Shandong Provincial Qianfoshan Hospital) from November 2020 to May 2021. The basic information, biochemical indexes, use of concomitant drugs, and genetic factors of organ donors and recipients were used for the ANN model inputs, and the output was the steady-state trough concentration (C0) of tacrolimus after oral administration in LT recipients. The ANN model was established to predict C0 of tacrolimus, SHAP was applied to the trained model, and the SHAP value of each input was calculated to analyze quantitatively the influencing factors for the output C0. RESULTS: A back-propagation ANN model with 3 hidden layers was established using deep learning. The mean prediction error was 0.27 ± 0.75 ng/mL; mean absolute error, 0.60 ± 0.52 ng/mL; correlation coefficient between predicted and actual C0 values, 0.9677; and absolute prediction error of all blood concentrations obtained by the ANN model, ≤3.0 ng/mL. The results indicated that the following factors had the most significant effect on C0: age, daily drug dose, genotype at CYP3A5 polymorphism rs776746 in both recipient and donor, and concomitant use of caspofungin. The predicted C0 value of tacrolimus in LT recipients increased in a dose-dependent manner when the daily dose exceeded 3 mg, whereas it decreased with age when LT recipients were older than 48 years. The predicted C0 was higher when recipients and donors had the genotype CYP3A5*3*3 than when they had the genotype CYP3A5*1. The predicted C0 value also increased with the use of caspofungin or Wuzhi capsule. CONCLUSION AND RELEVANCE: The established ANN model can be used to predict the C0 value of tacrolimus in LT recipients with high accuracy and good predictive ability, serving as a reference for personalized treatment in the early stage after liver transplantation.

Machine learning algorithms for prediction of entrapment efficiency in nanomaterials.

Exploitation of machine learning in predicting performance of nanomaterials is a rapidly growing dynamic area of research. For instance, incorporation of therapeutic cargoes into nanovesicles (i.e., entrapment efficiency) is one of the critical parameters that ensures proper entrapment of drugs in the developed nanosystems. Several factors affect the entrapment efficiency of drugs and thus multiple assessments are required to ensure drug retention, and to reduce cost and time. Supervised machine learning can allow for the construction of algorithms that can mine data available from earlier studies to predict performance of specific types of nanoparticles. Comparative studies that utilize multiple regression algorithms to predict entrapment efficiency in nanomaterials are scarce. Herein, we report on a detailed methodology for prediction of entrapment efficiency in nanomaterials (e.g., niosomes) using different regression algorithms (i.e., CatBoost, linear regression, support vector regression and artificial neural network) to select the model that demonstrates the best performance for estimation of entrapment efficiency. The study concluded that CatBoost algorithm demonstrated the best performance with maximum R2 score (0.98) and mean square error (< 10-4). Among the various parameters that possess a role in entrapment efficiency of drugs into niosomes, the results obtained from CatBoost model revealed that the drug:lipid ratio is the major contributing factor affecting entrapment efficiency, followed by the lipid:surfactant molar ratio. Hence, supervised machine learning may be applied for future selection of the components of niosomes that achieve high entrapment efficiency of drugs while minimizing experimental procedures and cost.

Artificial neural networks in the modeling of the catalytic activity of a biosensor composed of conjugated polymers and urease.

Thin films of conjugated polymer and enzyme can be used to unravel the interaction between components in a biosensor. Using artificial neural networks (ANNs) improves data interpretability and helps construct models with great capacity for classifying and processing information. The present work used kinetic data from the catalytic activity of urease immobilized in different conjugated polymers to create ANN models using time, substrate concentration, and absorbance as input variables since the models had absorbance in a posterior instant as output value to explore the predictivity of the ANNs. The performance of the models was evaluated by Pearson's correlation coefficient (ρ) and mean squared error (MSE) values. After the learning process, a series of new experiments were performed to verify the generality of the models. As the main results, the best ANN model presented 0.9980 and 3.0736 × 10-5 for ρ and MSE, respectively. For the simulation step, intermediary values of substrate concentration were used. The mean absolute percentage error (MAPE) values were 3.34, 3.07, and 3.78 for 12 mM, 22 mM, and 32 mM concentrations, respectively. Overall, with the simulations, it was possible to ascertain the interpolatory capacity of the model, which has a learning mechanism based on absorbance and time as variables. Thus, the potential of ANNs would be in their use in pre-evaluations, helping to determine the substrate concentration at which there is higher catalytic activity or in determining the linear range of the sensor.

Advances in the co-production of biosurfactant and other biomolecules: statistical approaches for process optimization.

An efficient microbial conversion for simultaneous synthesis of multiple high-value compounds, such as biosurfactants and enzymes, is one of the most promising aspects for an economical bioprocess leading to a marked reduction in production cost. Although biosurfactant and enzyme production separately have been much explored, there are limited reports on the predictions and optimization studies on simultaneous production of biosurfactants and other industrially important enzymes, including lipase, protease, and amylase. Enzymes are suited for an integrated production process with biosurfactants as multiple common industrial processes and applications are catalysed by these molecules. However, the complexity in microbial metabolism complicates the production process. This study details the work done on biosurfactant and enzyme co-production and explores the application and scope of various statistical tools and methodologies in this area of research. The use of advanced computational tools is yet to be explored for the optimization of downstream strategies in the co-production process. Given the complexity of the co-production process and with various new methodologies based on artificial intelligence (AI) being invented, the scope of AI in shaping the biosurfactant-enzyme co-production process is immense and would lead to not only efficient and rapid optimization, but economical extraction of multiple biomolecules as well.

Breaking boundaries: Artificial intelligence for pesticide detection and eco-friendly degradation.

Pesticides are extensively used agrochemicals across the world to control pest populations. However, irrational application of pesticides leads to contamination of various components of the environment, like air, soil, water, and vegetation, all of which build up significant levels of pesticide residues. Further, these environmental contaminants fuel objectionable human toxicity and impose a greater risk to the ecosystem. Therefore, search of methodologies having potential to detect and degrade pesticides in different environmental media is currently receiving profound global attention. Beyond the conventional approaches, Artificial Intelligence (AI) coupled with machine learning and artificial neural networks are rapidly growing branches of science that enable quick data analysis and precise detection of pesticides in various environmental components. Interestingly, nanoparticle (NP)-mediated detection and degradation of pesticides could be linked to AI algorithms to achieve superior performance. NP-based sensors stand out for their operational simplicity as well as their high sensitivity and low detection limits when compared to conventional, time-consuming spectrophotometric assays. NPs coated with fluorophores or conjugated with antibody or enzyme-anchored sensors can be used through Surface-Enhanced Raman Spectrometry, fluorescence, or chemiluminescence methodologies for selective and more precise detection of pesticides. Moreover, NPs assist in the photocatalytic breakdown of various organic and inorganic pesticides. Here, AI models are ideal means to identify, classify, characterize, and even predict the data of pesticides obtained through NP sensors. The present study aims to discuss the environmental contamination and negative impacts of pesticides on the ecosystem. The article also elaborates the AI and NP-assisted approaches for detecting and degrading a wide range of pesticide residues in various environmental and agrecultural sources including fruits and vegetables. Finally, the prevailing limitations and future goals of AI-NP-assisted techniques have also been dissected.