RBProkCNN: Deep learning on appropriate contextual evolutionary information for RNA binding protein discovery in prokaryotes

RNA-binding proteins (RBPs) are central to key functions such as post-transcriptional regulation, mRNA stability, and adaptation to varied environmental conditions in prokaryotes. While the majority of research has concentrated on eukaryotic RBPs, recent developments underscore the crucial involvement of prokaryotic RBPs. Although computational methods have emerged in recent years to identify RBPs, they have fallen short in accurately identifying prokaryotic RBPs due to their generic nature. To bridge this gap, we introduce RBProkCNN, a novel machine learning-driven computational model meticulously designed for the accurate prediction of prokaryotic RBPs. The prediction process involves the utilization of eight shallow learning algorithms and four deep learning models, incorporating PSSM-based evolutionary features. By leveraging a convolutional neural network (CNN) and evolutionarily significant features selected through extreme gradient boosting variable importance measure, RBProkCNN achieved the highest accuracy in five-fold cross-validation, yielding 98.04% auROC and 98.19% auPRC. Furthermore, RBProkCNN demonstrated robust performance with an independent dataset, showcasing a commendable 95.77% auROC and 95.78% auPRC. Noteworthy is its superior predictive accuracy when compared to several state-of-the-art existing models. RBProkCNN is available as an online prediction tool (https://iasri-sg.icar.gov.in/rbprokcnn/), offering free access to interested users. This tool represents a substantial contribution, enriching the array of resources available for the accurate and efficient prediction of prokaryotic RBPs.

Enhancing ASD detection accuracy: a combined approach of machine learning and deep learning models with natural language processing

Purpose

The main aim of our study was to explore the utility of artificial intelligence (AI) in diagnosing autism spectrum disorder (ASD). The study primarily focused on using machine learning (ML) and deep learning (DL) models to detect ASD potential cases by analyzing text inputs, especially from social media platforms like Twitter. This is to overcome the ongoing challenges in ASD diagnosis, such as the requirement for specialized professionals and extensive resources. Timely identification, particularly in children, is essential to provide immediate intervention and support, thereby improving the quality of life for affected individuals.

Methods

We employed natural language processing (NLP) techniques along with ML models like decision trees, extreme gradient boosting (XGB), k-nearest neighbors algorithm (KNN), and DL models such as recurrent neural networks (RNN), long short-term memory (LSTM), bidirectional long short-term memory (Bi-LSTM), bidirectional encoder representations from transformers (BERT and BERTweet). We extracted a dataset of 404,627 tweets from Twitter users using the platform's API and classified them based on whether they were written by individuals claiming to have ASD (ASD users) or by those without ASD (non-ASD users). From this dataset, we used a subset of 90,000 tweets (45,000 from each classification group) for the training and testing of these models.

Results

The application of our AI models yielded promising results, with the predictive model reaching an accuracy of almost 88% when classifying texts that potentially originated from individuals with ASD.

Conclusion

Our research demonstrated the potential of using AI, particularly DL models, in enhancing the accuracy of ASD detection and diagnosis. This innovative approach signifies the critical role AI can play in advancing early diagnostic techniques, enabling better patient outcomes and underlining the importance of early identification of ASD, especially in children.

Depression assessment using integrated multi-featured EEG bands deep neural network models: Leveraging ensemble learning techniques

Mental Status Assessment (MSA) holds significant importance in psychiatry. In recent years, several studies have leveraged Electroencephalogram (EEG) technology to gauge an individual's mental state or level of depression. This study introduces a novel multi-tier ensemble learning approach to integrate multiple EEG bands for conducting mental state or depression assessments. Initially, the EEG signal is divided into eight sub-bands, and then a Long Short-Term Memory (LSTM)-based Deep Neural Network (DNN) model is trained for each band. Subsequently, the integration of multi-band EEG frequency models and the evaluation of mental state or depression level are facilitated through a two-tier ensemble learning approach based on Multiple Linear Regression (MLR). The authors conducted numerous experiments to validate the performance of the proposed method under different evaluation metrics. For clarity and conciseness, the research employs the simplest commercialized one-channel EEG sensor, positioned at FP1, to collect data from 57 subjects (49 depressed and 18 healthy subjects). The obtained results, including an accuracy of 0.897, F1-score of 0.921, precision of 0.935, negative predictive value of 0.829, recall of 0.908, specificity of 0.875, and AUC of 0.8917, provide evidence of the superior performance of the proposed method compared to other ensemble learning techniques. This method not only proves effective but also holds the potential to significantly enhance the accuracy of depression assessment.

Autism spectrum disorder detection with kNN imputer and machine learning classifiers via questionnaire mode of screening

Autism spectrum disorder (ASD) is a neurodevelopmental disorder. ASD cannot be fully cured, but early-stage diagnosis followed by therapies and rehabilitation helps an autistic person to live a quality life. Clinical diagnosis of ASD symptoms via questionnaire and screening tests such as Autism Spectrum Quotient-10 (AQ-10) and Quantitative Check-list for Autism in Toddlers (Q-chat) are expensive, inaccessible, and time-consuming processes. Machine learning (ML) techniques are beneficial to predict ASD easily at the initial stage of diagnosis. The main aim of this work is to classify ASD and typical developed (TD) class data using ML classifiers. In our work, we have used different ASD data sets of all age groups (toddlers, adults, children, and adolescents) to classify ASD and TD cases. We implemented One-Hot encoding to translate categorical data into numerical data during preprocessing. We then used kNN Imputer with MinMaxScaler feature transformation to handle missing values and data normalization. ASD and TD class data is classified using Support vector machine, k-nearest-neighbor (KNN), random forest (RF), and artificial neural network classifiers. RF gives the best performance in terms of the accuracy of 100% with different training and testing data split for all four types of data sets and has no over-fitting issue. We have also examined our results with already published work, including recent methods like Deep Neural Network (DNN) and Convolution Neural Network (CNN). Even using complex architectures like DNN and CNN, our proposed methods provide the best results with low-complexity models. In contrast, existing methods have shown accuracy upto 98% with log-loss upto 15%. Our proposed methodology demonstrates the improved generalization for real-time ASD detection during clinical trials.

Benchmarking feature selection and feature extraction methods to improve the performances of machine-learning algorithms for patient classification using metabolomics biomedical data

Objective

Classification tasks are an open challenge in the field of biomedicine. While several machine-learning techniques exist to accomplish this objective, several peculiarities associated with biomedical data, especially when it comes to omics measurements, prevent their use or good performance achievements. Omics approaches aim to understand a complex biological system through systematic analysis of its content at the molecular level. On the other hand, omics data are heterogeneous, sparse and affected by the classical "curse of dimensionality" problem, i.e. having much fewer observation, samples (n) than omics features (p). Furthermore, a major problem with multi-omics data is the imbalance either at the class or feature level. The objective of this work is to study whether feature extraction and/or feature selection techniques can improve the performances of classification machine-learning algorithms on omics measurements.

Methods

Among all omics, metabolomics has emerged as a powerful tool in cancer research, facilitating a deeper understanding of the complex metabolic landscape associated with tumorigenesis and tumor progression. Thus, we selected three publicly available metabolomics datasets, and we applied several feature extraction techniques both linear and non-linear, coupled or not with feature selection methods, and evaluated the performances regarding patient classification in the different configurations for the three datasets.

Results

We provide general workflow and guidelines on when to use those techniques depending on the characteristics of the data available. To further test the extension of our approach to other omics data, we have included a transcriptomics and a proteomics data. Overall, for all datasets, we showed that applying supervised feature selection improves the performances of feature extraction methods for classification purposes. Scripts used to perform all analyses are available at: https://github.com/Plant-Net/Metabolomic_project/.

Identifying factors controlling cellular uptake of gold nanoparticles by machine learning

There is strong interest to improve the therapeutic potential of gold nanoparticles (GNPs) while ensuring their safe development. The utility of GNPs in medicine requires a molecular-level understanding of how GNPs interact with biological systems. Despite considerable research efforts devoted to monitoring the internalisation of GNPs, there is still insufficient understanding of the factors responsible for the variability in GNP uptake in different cell types. Data-driven models are useful for identifying the sources of this variability. Here, we trained multiple machine learning models on 2077 data points for 193 individual nanoparticles from 59 independent studies to predict cellular uptake level of GNPs and compared different algorithms for their efficacies of prediction. The five ensemble learners (Xgboost, random forest, bootstrap aggregation, gradient boosting, light gradient boosting machine) made the best predictions of GNP uptake, accounting for 80-90% of the variance in the test data. The models identified particle size, zeta potential, GNP concentration and exposure duration as the most important drivers of cellular uptake. We expect this proof-of-concept study will foster the more effective use of accumulated cellular uptake data for GNPs and minimise any methodological bias in individual studies that may lead to under- or over-estimation of cellular internalisation rates.

Joining forces for pathology diagnostics with AI assistance: The EMPAIA initiative

Over the past decade, artificial intelligence (AI) methods in pathology have advanced substantially. However, integration into routine clinical practice has been slow due to numerous challenges, including technical and regulatory hurdles in translating research results into clinical diagnostic products and the lack of standardized interfaces. The open and vendor-neutral EMPAIA initiative addresses these challenges. Here, we provide an overview of EMPAIA's achievements and lessons learned. EMPAIA integrates various stakeholders of the pathology AI ecosystem, i.e., pathologists, computer scientists, and industry. In close collaboration, we developed technical interoperability standards, recommendations for AI testing and product development, and explainability methods. We implemented the modular and open-source EMPAIA Platform and successfully integrated 14 AI-based image analysis apps from eight different vendors, demonstrating how different apps can use a single standardized interface. We prioritized requirements and evaluated the use of AI in real clinical settings with 14 different pathology laboratories in Europe and Asia. In addition to technical developments, we created a forum for all stakeholders to share information and experiences on digital pathology and AI. Commercial, clinical, and academic stakeholders can now adopt EMPAIA's common open-source interfaces, providing a unique opportunity for large-scale standardization and streamlining of processes. Further efforts are needed to effectively and broadly establish AI assistance in routine laboratory use. To this end, a sustainable infrastructure, the non-profit association EMPAIA International, has been established to continue standardization and support broad implementation and advocacy for an AI-assisted digital pathology future.

A review on trends in development and translation of omics signatures in cancer

The field of cancer genomics and transcriptomics has evolved from targeted profiling to swift sequencing of individual tumor genome and transcriptome. The steady growth in genome, epigenome, and transcriptome datasets on a genome-wide scale has significantly increased our capability in capturing signatures that represent both the intrinsic and extrinsic biological features of tumors. These biological differences can help in precise molecular subtyping of cancer, predicting tumor progression, metastatic potential, and resistance to therapeutic agents. In this review, we summarized the current development of genomic, methylomic, transcriptomic, proteomic and metabolic signatures in the field of cancer research and highlighted their potentials in clinical applications to improve diagnosis, prognosis, and treatment decision in cancer patients.

Using staged tree models for health data: Investigating invasive fungal infections by aspergillus and other filamentous fungi

Machine learning models are increasingly used in the medical domain to study the association between risk factors and diseases to support practitioners in understanding health outcomes. In this paper, we showcase the use of machine-learned staged tree models for investigating complex asymmetric dependence structures in health data. Staged trees are a specific class of generative, probabilistic graphical models that formally model asymmetric conditional independence and non-regular sample spaces. An investigation of the risk factors in invasive fungal infections demonstrates the insights staged trees provide to support medical decision-making.

A class-imbalanced hybrid learning strategy based on Raman spectroscopy of serum samples for the diagnosis of hepatitis B, hepatitis A, and thyroid dysfunction

Computer-aided vibrational spectroscopy detection technology has achieved promising results in the field of early disease diagnosis. Yet limited by factors such as the number of actual samples and the cost of spectral acquisition in clinical medicine, the data available for model training are insufficient, and the amount of data varies greatly between different diseases, which constrain the performance optimization and enhancement of the diagnostic model. In this study, vibrational spectroscopy data of three common diseases are selected as research objects, and experimental research is conducted around the class imbalance situation that exists in medical data. When dealing with the challenge of class imbalance in medical vibrational spectroscopy research, it no longer relies on some kind of independent and single method, but considers the combined effect of multiple strategies. SVM, K-Nearest Neighbor (KNN), and Decision Tree (DT) are used as baseline comparison models on Raman spectroscopy medical datasets with different imbalance rates. The performance of the three strategies, Ensemble Learning, Feature Extraction, and Resampling, is verified on the class imbalance dataset by G-mean and AUC metrics, respectively. The results show that all the above three methods mitigate the negative impact caused by unbalanced learning. Based on this, we propose a hybrid ensemble classifier (HEC) that integrates resampling, feature extraction, and ensemble learning to verify the effectiveness of the hybrid learning strategy in solving the class imbalance problem. The G-mean and AUC values of the HEC method are 82.7 % and 83.12 % for the HBV dataset, is 2.02 % and 1.98 % higher than the optimal strategy; 83.62 % and 83.76 % for the HCV dataset, is 9.79 % and 8.47 % higher than the optimal strategy; while for the thyroid dysfunction dataset are 77.56 % and 77.85 %, is 6.92 % and 6.36 % higher than that of the optimal strategy, respectively. The experimental results show that the G-mean and AUC metrics of the HEC method are higher than those of the baseline classifier as well as the optimal combination using separate strategies. It can be seen that the HEC method can effectively counteract the unfavorable effects of imbalance learning and is expected to be applied to a wider range of imbalance scenarios.

Lipids balance as a spectroscopy marker of diabetes. Analysis of FTIR spectra by 2D correlation and machine learning analyses

The number of people suffering from type 2 diabetes has rapidly increased. Taking into account, that elevated intracellular lipid concentrations, as well as their metabolism, are correlated with diminished insulin sensitivity, in this study we would like to show lipids spectroscopy markers of diabetes. For this purpose, serum collected from rats (animal model of diabetes) was analyzed using Fourier Transformed Infrared-Attenuated Total Reflection (FTIR-ATR) spectroscopy. Analyzed spectra showed that rats with diabetes presented higher concentration of phospholipids and cholesterol in comparison with non-diabetic rats. Moreover, the analysis of second (IInd) derivative spectra showed no structural changes in lipids. Machine learning methods showed higher accuracy for IInd derivative spectra (from 65 % to 89 %) than for absorbance FTIR spectra (53-65 %). Moreover, it was possible to identify significant wavelength intervals from IInd derivative spectra using random forest-based feature selection algorithm, which further increased the accuracy of the classification (up to 92 % for phospholipid region). Moreover decision tree based on the selected features showed, that peaks at 1016 cm-1 and 2936 cm-1 can be good candidates of lipids marker of diabetes.

CVAE-DF: A hybrid deep learning framework for fertilization status detection of pre-incubation duck eggs based on VIS/NIR spectroscopy

Unfertilized duck eggs not removed prior to incubation will deteriorate quickly, posing a risk of contaminating the normally fertilized duck eggs. Thus, detecting the fertilization status of breeding duck eggs as early as possible is a meaningful and challenging task. Most existing work usually focus on the characteristics of chicken eggs during mid-term hatching. However, little attention has been paid to the detection for duck eggs prior to incubation. In this paper, we present a novel hybrid deep learning detection framework for the fertilization status of pre-incubation duck eggs, termed CVAE-DF, based on visible/near-infrared (VIS/NIR) transmittance spectroscopy. The framework comprises the encoder of a convolutional variational autoencoder (CVAE) and an improved deep forest (DF) model. More specifically, we first collected transmittance spectral data (400-1000 nm) of 255 duck eggs before hatching. The multiplicative scatter correction (MSC) method was then used to eliminate noise and extraneous information of the raw spectral data. Two efficient data augmentation methods were adopted to provide sufficient data. After that, CVAE was applied to extract representative features and reduce the feature dimension for the detection task. Finally, an improved DF model was employed to build the classification model on the enhanced feature set. The CVAE-DF model achieved an overall accuracy of 95.94 % on the test dataset. These experimental results in terms of four metrics demonstrate that our CVAE-DF method outperforms the traditional methods by a significant margin. Furthermore, the results also indicate that CVAE holds great promise as a novel feature extraction method for the VIS/NIR spectral analysis of other agricultural products. It is extremely beneficial to practical engineering.

Advancing toxicity studies of per- and poly-fluoroalkyl substances (pfass) through machine learning: Models, mechanisms, and future directions

Perfluorinated and perfluoroalkyl substances (PFASs), encompassing a vast array of isomeric chemicals, are recognized as typical emerging contaminants with direct or potential impacts on human health and the ecological environment. With the complex and elusive toxicological profiles of PFASs, machine learning (ML) has been increasingly employed in their toxicity studies due to its proficiency in prediction and data analytics. This integration is poised to become a predominant trend in environmental toxicology, propelled by the swift advancements in computational technology. This review diligently examines the literature to encapsulate the varied objectives of employing ML in the toxicity studies of PFASs: (1) Utilizing ML to establish Quantitative Structure-Activity Relationship (QSAR) models for PFASs with diverse toxicity endpoints, facilitating the targeted toxicity prediction of unidentified PFASs; (2) Investigating and substantiating the Adverse Outcome Pathway (AOP) through the synergy of ML and traditional toxicological methods, with this refining the toxicity assessment framework for PFASs; (3) Dissecting and elucidating the features of established ML models to advance Open Research into the toxicity of PFASs, with a primary focus on determinants and mechanisms. The discourse extends to an in-depth examination of ML studies, segregating findings based on their distinct application trajectories. Given that ML represents a nascent paradigm within PFASs research, this review delineates the collective challenges encountered in the ML-mediated study of PFAS toxicity and proffers strategic guidance for ensuing investigations.

Automated machine learning-aided prediction and interpretation of gaseous by-products from the hydrothermal liquefaction of biomass

Hydrothermal liquefaction (HTL) is a thermochemical conversion technology that produces bio-oil from wet biomass without drying. However, by-product gases will inevitably be produced, and their formation is unclear. Therefore, an automated machine learning (AutoML) approach, automatically training without human intervention, was used to aid in predicting gaseous production and interpreting the formation mechanisms of four gases (CO2, CH4, CO, and H2). Specifically, four accurate optimal single-target models based on AutoML were developed with elemental compositions and HTL conditions as inputs for four gases. Herein, the gradient boosting machine (GBM) performed excellently with train R2 ≥ 0.99 and test R2 ≥ 0.80. Then, the screened GBM algorithm-based ML multi-target models (maximum average test R2 = 0.89 and RMSE = 0.39) were built to predict four gases simultaneously. Results indicated that biomass carbon, solid content, pressure, and biomass hydrogen were the top four factors for gas production from HTL of biomass. This study proposed an AutoML-aided prediction and interpretation framework, which could provide new insight for rapid prediction and revelation of gaseous compositions from the HTL process.

Revealing the nature of cardiovascular disease using DERGA, a novel data ensemble refinement greedy algorithm

BACKGROUND

The study aimed to determine the most crucial parameters associated with CVD and employ a novel data ensemble refinement procedure to uncover the optimal pattern of these parameters that can result in a high prediction accuracy.

METHODS AND RESULTS

Data were collected from 369 patients in total, 281 patients with CVD or at risk of developing it, compared to 88 otherwise healthy individuals. Within the group of 281 CVD or at-risk patients, 53 were diagnosed with coronary artery disease (CAD), 16 with end-stage renal disease, 47 newly diagnosed with diabetes mellitus 2 and 92 with chronic inflammatory disorders (21 rheumatoid arthritis, 41 psoriasis, 30 angiitis). The data were analyzed using an artificial intelligence-based algorithm with the primary objective of identifying the optimal pattern of parameters that define CVD. The study highlights the effectiveness of a six-parameter combination in discerning the likelihood of cardiovascular disease using DERGA and Extra Trees algorithms. These parameters, ranked in order of importance, include Platelet-derived Microvesicles (PMV), hypertension, age, smoking, dyslipidemia, and Body Mass Index (BMI). Endothelial and erythrocyte MVs, along with diabetes were the least important predictors. In addition, the highest prediction accuracy achieved is 98.64%. Notably, using PMVs alone yields a 91.32% accuracy, while the optimal model employing all ten parameters, yields a prediction accuracy of 0.9783 (97.83%).

CONCLUSIONS

Our research showcases the efficacy of DERGA, an innovative data ensemble refinement greedy algorithm. DERGA accelerates the assessment of an individual's risk of developing CVD, allowing for early diagnosis, significantly reduces the number of required lab tests and optimizes resource utilization. Additionally, it assists in identifying the optimal parameters critical for assessing CVD susceptibility, thereby enhancing our understanding of the underlying mechanisms.

Magnetic resonance imaging-based radiomics model for preoperative assessment of risk stratification in endometrial cancer

BACKGROUND

Preoperative risk stratification is significant for the management of endometrial cancer (EC) patients. Radiomics based on magnetic resonance imaging (MRI) in combination with clinical features may be useful to predict the risk grade of EC.

AIM

To construct machine learning models to predict preoperative risk stratification of patients with EC based on radiomics features extracted from MRI.

METHODS

The study comprised 112 EC patients. The participants were randomly separated into training and validation groups with a 7:3 ratio. Logistic regression analysis was applied to uncover independent clinical predictors. These predictors were then used to create a clinical nomogram. Extracted radiomics features from the T2-weighted imaging and diffusion weighted imaging sequences of MRI images, the Mann-Whitney U test, Pearson test, and least absolute shrinkage and selection operator analysis were employed to evaluate the relevant radiomic features, which were subsequently utilized to generate a radiomic signature. Seven machine learning strategies were used to construct radiomic models that relied on the screening features. The logistic regression method was used to construct a composite nomogram that incorporated both the radiomic signature and clinical independent risk indicators.

RESULTS

Having an accuracy of 0.82 along with an area under the curve (AUC) of 0.915 [95% confidence interval (CI): 0.806-0.986], the random forest method trained on radiomics characteristics performed better than expected. The predictive accuracy of radiomics prediction models surpassed that of both the clinical nomogram (AUC: 0.75, 95%CI: 0.611-0.899) and the combined nomogram (AUC: 0.869, 95%CI: 0.702-0.986) that integrated clinical parameters and radiomic signature.

CONCLUSION

The MRI-based radiomics model may be an effective tool for preoperative risk grade prediction in EC patients.

Machine learning applied to the prediction of relapse, hospitalization, and suicide in bipolar disorder using neuroimaging and clinical data: A systematic review

BACKGROUND

Bipolar disorder (BD) is associated with increased morbidity/mortality. Adverse outcome prediction might help with the management of patients with BD.

METHODS

We systematically reviewed the performance of machine learning (ML) studies in predicting adverse outcomes (relapse or recurrence, hospital admission, and suicide-related events) in patients with BD. Demographic, clinical, and neuroimaging-related poor outcome predictors were also reviewed. Three databases (PubMed, Scopus, and Web of Science) were explored from inception to July 2023.

RESULTS

Eighteen studies, accounting for >30,000 patients, were included. Support vector machine, decision trees, random forest, and logistic regression were the most frequently used ML algorithms. ML models' area under the receiver operating characteristic (ROC) curve (AUC), sensitivity, and specificity ranged from 0.71 to 0.98, 72.7-92.8 %, and 59.0-95.2 % for relapse/recurrence prediction (4 studies (3 on relapses and 1 on recurrences). The corresponding values were 0.78-0.88, 21.4-100 %, and 77.0-99.7 % for hospital admissions (3 studies, 21,266 patients), and 0.71-0.99, 44.4-97.9 %, and 38.9-95.0 % for suicide-related events (10 studies, 5558 patients). Also, one study addressed a combination of the interest outcomes. Adverse outcome predictors included early onset BD, BD type I, comorbid psychiatric or substance use disorder, circadian rhythm disruption, hospitalization characteristics, and neuroimaging parameters, including increased dynamic amplitude of low-frequency fluctuation, decreased frontolimbic functional connectivity and aberrant dynamic functional connectivity in corticostriatal circuitry.

CONCLUSIONS

ML models can predict adverse outcomes of BD with relatively acceptable performance measures. Future studies with larger samples and nested cross-validation validation should be conducted to reach more reliable results.

On formal limitations of causal ecological networks

Causal multivariate time-series analysis, combined with network theory, provide a powerful tool for studying complex ecological interactions. However, these methods have limitations often underestimated when used in graphical modelling of ecological systems. In this opinion article, I examine the relationship between formal logic methods used to describe causal networks and their inherent statistical and epistemological limitations. I argue that while these methods offer valuable insights, they are restricted by axiomatic assumptions, statistical constraints and the incompleteness of our knowledge. To prove that, I first consider causal networks as formal systems, define causality and formalize their axioms in terms of modal logic and use ecological counterexamples to question the axioms. I also highlight the statistical limitations when using multivariate time-series analysis and Granger causality to develop ecological networks, including the potential for spurious correlations among other data characteristics. Finally, I draw upon Gödel's incompleteness theorems to highlight the inherent limits of fully understanding complex networks as formal systems and conclude that causal ecological networks are subject to initial rules and data characteristics and, as any formal system, will never fully capture the intricate complexities of the systems they represent. This article is part of the theme issue 'Connected interactions: enriching food web research by spatial and social interactions'.

PepExplainer: An explainable deep learning model for selection-based macrocyclic peptide bioactivity prediction and optimization

Macrocyclic peptides possess unique features, making them highly promising as a drug modality. However, evaluating their bioactivity through wet lab experiments is generally resource-intensive and time-consuming. Despite advancements in artificial intelligence (AI) for bioactivity prediction, challenges remain due to limited data availability and the interpretability issues in deep learning models, often leading to less-than-ideal predictions. To address these challenges, we developed PepExplainer, an explainable graph neural network based on substructure mask explanation (SME). This model excels at deciphering amino acid substructures, translating macrocyclic peptides into detailed molecular graphs at the atomic level, and efficiently handling non-canonical amino acids and complex macrocyclic peptide structures. PepExplainer's effectiveness is enhanced by utilizing the correlation between peptide enrichment data from selection-based focused library and bioactivity data, and employing transfer learning to improve bioactivity predictions of macrocyclic peptides against IL-17C/IL-17 RE interaction. Additionally, PepExplainer underwent further validation for bioactivity prediction using an additional set of thirteen newly synthesized macrocyclic peptides. Moreover, it enabled the optimization of the IC50 of a macrocyclic peptide, reducing it from 15 nM to 5.6 nM based on the contribution score provided by PepExplainer. This achievement underscores PepExplainer's skill in deciphering complex molecular patterns, highlighting its potential to accelerate the discovery and optimization of macrocyclic peptides.

The statistics of cognitive variability: Explaining common patterns in individuals, groups and financial markets

Psychological variability (i.e., "noise") displays interesting structure which is hidden by the common practice of averaging over trials. Interesting noise structure, termed 'stylized facts', is observed in financial markets (i.e., behaviors from many thousands of traders). Here we investigate the parallels between psychological and financial time series. In a series of three experiments (total N = 202), we successively simplified a market-based price prediction task by first removing external information, and then removing any interaction between participants. Finally, we removed any resemblance to an asset market by asking individual participants to simply reproduce temporal intervals. All three experiments reproduced the main stylized facts found in financial markets, and the robustness of the results suggests that a common cognitive-level mechanism can produce them. We identify one potential model based on mental sampling algorithms, showing how this general-purpose model might account for behavior across these very different tasks.

Optimizing hybrid ensemble feature selection strategies for transcriptomic biomarker discovery in complex diseases

Biomedical research takes advantage of omic data, such as transcriptomics, to unravel the complexity of diseases. A conventional strategy identifies transcriptomic biomarkers characterized by expression patterns associated with a phenotype by relying on feature selection approaches. Hybrid ensemble feature selection (HEFS) has become increasingly popular as it ensures robustness of the selected features by performing data and functional perturbations. However, it remains difficult to make the best suited choices at each step when designing such approaches. We conducted an extensive analysis of four possible HEFS scenarios for the identification of Stage IV colorectal, Stage I kidney and lung and Stage III endometrial cancer biomarkers from transcriptomic data. These scenarios investigate the use of two types of feature reduction by filters (differentially expressed genes and variance) conjointly with two types of resampling strategies (repeated holdout by distribution-balanced stratified and random stratified) for downstream feature selection through an aggregation of thousands of wrapped machine learning models. Based on our results, we emphasize the advantages of using HEFS approaches to identify complex disease biomarkers, given their ability to produce generalizable and stable results to both data and functional perturbations. Finally, we highlight critical issues that need to be considered in the design of such strategies.

Performance of Automated Machine Learning in Predicting Outcomes of Pneumatic Retinopexy

Purpose

Automated machine learning (AutoML) has emerged as a novel tool for medical professionals lacking coding experience, enabling them to develop predictive models for treatment outcomes. This study evaluated the performance of AutoML tools in developing models predicting the success of pneumatic retinopexy (PR) in treatment of rhegmatogenous retinal detachment (RRD). These models were then compared with custom models created by machine learning (ML) experts.

Design

Retrospective multicenter study.

Participants

Five hundred and thirty nine consecutive patients with primary RRD that underwent PR by a vitreoretinal fellow at 6 training hospitals between 2002 and 2022.

Methods

We used 2 AutoML platforms: MATLAB Classification Learner and Google Cloud AutoML. Additional models were developed by computer scientists. We included patient demographics and baseline characteristics, including lens and macula status, RRD size, number and location of breaks, presence of vitreous hemorrhage and lattice degeneration, and physicians' experience. The dataset was split into a training (n = 483) and test set (n = 56). The training set, with a 2:1 success-to-failure ratio, was used to train the MATLAB models. Because Google Cloud AutoML requires a minimum of 1000 samples, the training set was tripled to create a new set with 1449 datapoints. Additionally, balanced datasets with a 1:1 success-to-failure ratio were created using Python.

Main Outcome Measures

Single-procedure anatomic success rate, as predicted by the ML models. F2 scores and area under the receiver operating curve (AUROC) were used as primary metrics to compare models.

Results

The best performing AutoML model (F2 score: 0.85; AUROC: 0.90; MATLAB), showed comparable performance to the custom model (0.92, 0.86) when trained on the balanced datasets. However, training the AutoML model with imbalanced data yielded misleadingly high AUROC (0.81) despite low F2-score (0.2) and sensitivity (0.17).

Conclusions

We demonstrated the feasibility of using AutoML as an accessible tool for medical professionals to develop models from clinical data. Such models can ultimately aid in the clinical decision-making, contributing to better patient outcomes. However, outcomes can be misleading or unreliable if used naively. Limitations exist, particularly if datasets contain missing variables or are highly imbalanced. Proper model selection and data preprocessing can improve the reliability of AutoML tools.

Financial Disclosures

Proprietary or commercial disclosure may be found in the Footnotes and Disclosures at the end of this article.

Incremental Confidence Sampling with Optimal Transport for Domain Adaptation

Domain adaptation is a subfield of statistical learning theory that takes into account the shift between the distribution of training and test data, typically known as source and target domains, respectively. In this context, this paper presents an incremental approach to tackle the intricate challenge of unsupervised domain adaptation, where labeled data within the target domain is unavailable. The proposed approach, OTP-DA, endeavors to learn a sequence of joint subspaces from both the source and target domains using Linear Discriminant Analysis (LDA), such that the projected data into these subspaces are domain-invariant and well-separated. Nonetheless, the necessity of labeled data for LDA to derive the projection matrix presents a substantial impediment, given the absence of labels within the target domain in the setting of unsupervised domain adaptation. To circumvent this limitation, we introduce a selective label propagation technique grounded on optimal transport (OTP), to generate pseudo-labels for target data, which serve as surrogates for the unknown labels. We anticipate that the process of inferring labels for target data will be substantially streamlined within the acquired latent subspaces, thereby facilitating a self-training mechanism. Furthermore, our paper provides a rigorous theoretical analysis of OTP-DA, underpinned by the concept of weak domain adaptation learners, thereby elucidating the requisite conditions for the proposed approach to solve the problem of unsupervised domain adaptation efficiently. Experimentation across a spectrum of visual domain adaptation problems suggests that OTP-DA exhibits promising efficacy and robustness, positioning it favorably compared to several state-of-the-art methods.

Wavelet Transform, Reconstructed Phase Space, and Deep Learning Neural Networks for EEG-Based Schizophrenia Detection

This study proposes an innovative expert system that uses exclusively EEG signals to diagnose schizophrenia in its early stages. For diagnosing psychiatric/neurological disorders, electroencephalogram (EEG) testing is considered a financially viable, safe, and reliable alternative. Using the reconstructed phase space (RPS) and the continuous wavelet transform, the researchers maximized the differences between the EEG nonstationary signals of normal and schizophrenia individuals, which cannot be observed in the time, frequency, or time-frequency domains. This reveals significant information, highlighting more distinguishable features. Then, a deep learning network was trained to enhance the accuracy of the resulting image classification. The algorithm's efficacy was confirmed through three distinct methods: employing 70% of the dataset for training, 15% for validation, and the remaining 15% for testing. This was followed by a 5-fold cross-validation technique and a leave-one-out classification approach. Each method was iterated 100 times to ascertain the algorithm's robustness. The performance metrics derived from these tests - accuracy, precision, sensitivity, F1 score, Matthews correlation coefficient, and Kappa - indicated remarkable outcomes. The algorithm demonstrated steady performance across all evaluation strategies, underscoring its relevance and reliability. The outcomes validate the system's accuracy, precision, sensitivity, and robustness by showcasing its capability to autonomously differentiate individuals diagnosed with schizophrenia from those in a state of normal health.

Modeling Bellman-error with logistic distribution with applications in reinforcement learning

In modern Reinforcement Learning (RL) approaches, optimizing the Bellman error is a critical element across various algorithms, notably in deep Q-Learning and related methodologies. Traditional approaches predominantly employ the mean-squared Bellman error (MSELoss) as the standard loss function. However, the assumption of Bellman errors following the Gaussian distribution may oversimplify the nuanced characteristics of RL applications. In this work, we revisit the distribution of Bellman error in RL training, demonstrating that it tends to follow the Logistic distribution rather than the commonly assumed Normal distribution. We propose replacing MSELoss with a Logistic maximum likelihood function (LLoss) and rigorously test this hypothesis through extensive numerical experiments across diverse online and offline RL environments. Our findings consistently show that integrating the Logistic correction into the loss functions of various baseline RL methods leads to superior performance compared to their MSE counterparts. Additionally, we employ Kolmogorov-Smirnov tests to substantiate that the Logistic distribution offers a more accurate fit for approximating Bellman errors. This study also offers a novel theoretical contribution by establishing a clear connection between the distribution of Bellman error and the practice of proportional reward scaling, a common technique for performance enhancement in RL. Moreover, we explore the sample-accuracy trade-off involved in approximating the Logistic distribution, leveraging the Bias-Variance decomposition to mitigate excessive computational resources. The theoretical and empirical insights presented in this study lay a significant foundation for future research, potentially advancing methodologies, and understanding in RL, particularly in the distribution-based optimization of Bellman error.

Machine learning for the advancement of genome-scale metabolic modeling

Constraint-based modeling (CBM) has evolved as the core systems biology tool to map the interrelations between genotype, phenotype, and external environment. The recent advancement of high-throughput experimental approaches and multi-omics strategies has generated a plethora of new and precise information from wide-ranging biological domains. On the other hand, the continuously growing field of machine learning (ML) and its specialized branch of deep learning (DL) provide essential computational architectures for decoding complex and heterogeneous biological data. In recent years, both multi-omics and ML have assisted in the escalation of CBM. Condition-specific omics data, such as transcriptomics and proteomics, helped contextualize the model prediction while analyzing a particular phenotypic signature. At the same time, the advanced ML tools have eased the model reconstruction and analysis to increase the accuracy and prediction power. However, the development of these multi-disciplinary methodological frameworks mainly occurs independently, which limits the concatenation of biological knowledge from different domains. Hence, we have reviewed the potential of integrating multi-disciplinary tools and strategies from various fields, such as synthetic biology, CBM, omics, and ML, to explore the biochemical phenomenon beyond the conventional biological dogma. How the integrative knowledge of these intersected domains has improved bioengineering and biomedical applications has also been highlighted. We categorically explained the conventional genome-scale metabolic model (GEM) reconstruction tools and their improvement strategies through ML paradigms. Further, the crucial role of ML and DL in omics data restructuring for GEM development has also been briefly discussed. Finally, the case-study-based assessment of the state-of-the-art method for improving biomedical and metabolic engineering strategies has been elaborated. Therefore, this review demonstrates how integrating experimental and in silico strategies can help map the ever-expanding knowledge of biological systems driven by condition-specific cellular information. This multiview approach will elevate the application of ML-based CBM in the biomedical and bioengineering fields for the betterment of society and the environment.

Stacking and ridge regression-based spectral ensemble preprocessing method and its application in near-infrared spectral analysis

Spectral preprocessing techniques can, to a certain extent, eliminate irrelevant information, such as current noise and stray light from spectral data, thereby enhancing the performance of prediction models. However, current preprocessing techniques mostly attempt to find the best single preprocessing method or their combination, overlooking the complementary information among different preprocessing methods. These preprocessing techniques fail to maximize the utilization of useful information in spectral data and restrict the performance of prediction models. This study proposed a spectral ensemble preprocessing method based on the rapidly developing ensemble learning methods in recent years and the ridge regression (RR) model, named stacking preprocessing ridge regression (SPRR), to address the aforementioned issues. Different from conventional ensemble learning methods, the proposed SPRR method applied multiple different preprocessing techniques to the original spectral data, generating multiple preprocessed datasets. These datasets were then individually inputted into RR base models for training. Ultimately, RR still served as the meta-model, integrating the output results of each RR base model through stacking. This approach not only produced diversity in base models but also achieved higher accuracy and lower computational complexity by using a single type of base model. On the apple spectral dataset collected by our team, correlation analysis showed significant complementary information among the data produced by different preprocessing techniques. This provided robust theoretical support for the proposed SPRR method. By introducing the currently popular averaging ensemble preprocessing method in a comparative experiment, the results of applying the proposed SPRR method to six datasets (apple, meat, wheat, olive oil, tablet, and corn) demonstrated that compared to the single preprocessing method and averaging ensemble preprocessing method, SPRR yielded the best accuracy and reliability for all six datasets. Furthermore, under the same conditions of the training and test datasets, the proposed SPRR method demonstrated better performance than the four commonly used ensemble preprocessing methods.

EResNet-SVM: an overfitting-relieved deep learning model for recognition of plant diseases and pests

BACKGROUND

The accurate recognition and early warning for plant diseases and pests are a prerequisite of intelligent prevention and control for plant diseases and pests. As a result of the phenotype similarity of the hazarded plant after plant diseases and pests occur, as well as the interference of the external environment, traditional deep learning models often face the overfitting problem in phenotype recognition of plant diseases and pests, which leads to not only the slow convergence speed of the network, but also low recognition accuracy.

RESULTS

Motivated by the above problems, the present study proposes a deep learning model EResNet-support vector machine (SVM) to alleviate the overfitting for the recognition and classification of plant diseases and pests. First, the feature extraction capability of the model is improved by increasing feature extraction layers in the convolutional neural network. Second, the order-reduced modules are embedded and a sparsely activated function is introduced to reduce model complexity and alleviate overfitting. Finally, a classifier fused by SVM and fully connected layers are introduced to transforms the original non-linear classification problem into a linear classification problem in high-dimensional space to further alleviate the overfitting and improve the recognition accuracy of plant diseases and pests. The ablation experiments further demonstrate that the fused structure can effectively alleviate the overfitting and improve the recognition accuracy. The experimental recognition results for typical plant diseases and pests show that the proposed EResNet-SVM model has 99.30% test accuracy for eight conditions (seven plant diseases and one normal), which is 5.90% higher than the original ResNet18. Compared with the classic AlexNet, GoogLeNet, Xception, SqueezeNet and DenseNet201 models, the accuracy of the EResNet-SVM model has improved by 5.10%, 7%, 8.10%, 6.20% and 1.90%, respectively. The testing accuracy of the EResNet-SVM model for 6 insect pests is 100%, which is 3.90% higher than that of the original ResNet18 model.

CONCLUSION

This research provides not only useful references for alleviating the overfitting problem in deep learning, but also a theoretical and technical support for the intelligent detection and control of plant diseases and pests. © 2024 Society of Chemical Industry.

Deep reinforcement learning in chemistry: A review

Reinforcement learning (RL) has been applied to various domains in computational chemistry and has found wide-spread success. In this review, we first motivate the application of RL to chemistry and list some broad application domains, for example, molecule generation, geometry optimization, and retrosynthetic pathway search. We set up some of the formalism associated with reinforcement learning that should help the reader translate their chemistry problems into a form where RL can be used to solve them. We then discuss the solution formulations and algorithms proposed in recent literature for these problems, the advantages of one over the other, together with the necessary details of the RL algorithms they employ. This article should help the reader understand the state of RL applications in chemistry, learn about some relevant actively-researched open problems, gain insight into how RL can be used to approach them and hopefully inspire innovative RL applications in Chemistry.

Diagnosis of placenta accreta spectrum using ultrasound texture feature fusion and machine learning

INTRODUCTION

Placenta accreta spectrum (PAS) is an obstetric disorder arising from the abnormal adherence of the placenta to the uterine wall, often leading to life-threatening complications including postpartum hemorrhage. Despite its significance, PAS remains frequently underdiagnosed before delivery. This study delves into the realm of machine learning to enhance the precision of PAS classification. We introduce two distinct models for PAS classification employing ultrasound texture features.

METHODS

The first model leverages machine learning techniques, harnessing texture features extracted from ultrasound scans. The second model adopts a linear classifier, utilizing integrated features derived from 'weighted z-scores'. A novel aspect of our approach is the amalgamation of classical machine learning and statistical-based methods for feature selection. This, coupled with a more transparent classification model based on quantitative image features, results in superior performance compared to conventional machine learning approaches.

RESULTS

Our linear classifier and machine learning models attain test accuracies of 87 % and 92 %, and 5-fold cross validation accuracies of 88.7 (4.4) and 83.0 (5.0), respectively.

CONCLUSIONS

The proposed models illustrate the effectiveness of practical and robust tools for enhanced PAS detection, offering non-invasive and computationally-efficient diagnostic tools. As adjunct methods for prenatal diagnosis, these tools can assist clinicians by reducing the need for unnecessary interventions and enabling earlier planning of management strategies for delivery.

Artificial intelligence and machine learning: Definition of terms and current concepts in critical care research

With increasing computing power, artificial intelligence (AI) and machine learning (ML) have prospered, which facilitate the analysis of large datasets, especially those found in critical care. It is important to define these terminologies, to inform a standardized approach to critical care research. This manuscript hopes to clarify these terms with examples from medical literature. Three major components that are required for a successful ML implementation: (i) reliable dataset, (ii) ML algorithm, and (iii) unbiased model evaluation, are discussed. A reliable dataset can be structured or unstructured with limited noise, outliers, and missing values. ML, a subset of AI, is typically focused on supervised or unsupervised learning tasks in which the output is based on inputs and derived from iterative pattern recognition algorithms, while AI is the overall ability of a machine to "think" or mimic human behavior; and to analyze data free from human influence. Even with successful implementation, advanced AI and ML algorithms have faced challenges in adoption into practice, mainly due to their lack of interpretability, which hinders trust, buy-in, and engagement from clinicians. Consequently, traditional algorithms, such as linear and logistic regression, that may have reduced predictive power but are highly interpretable, continue to be widely used.

Deep Learning-Driven Transformation: A Novel Approach for Mitigating Batch Effects in Diffusion MRI Beyond Traditional Harmonization

BACKGROUND

"Batch effect" in MR images, due to vendor-specific features, MR machine generations, and imaging parameters, challenges image quality and hinders deep learning (DL) model generalizability.

PURPOSE

We aim to develop a DL model using contrast adjustment and super-resolution to reduce diffusion-weighted images (DWIs) diversity across magnetic field strengths and imaging parameters.

STUDY TYPE

Retrospective.

SUBJECTS

The DL model was built using an open dataset from one individual. The MR machine identification model was trained and validated on a dataset of 1134 adults (54% females, 46% males), with 1050 subjects showing no DWI abnormalities and 84 with conditions like stroke and tumors. The 21,000 images were divided into 80% for training, 20% for validation, and 3500 for testing.

FIELD STRENGTH/SEQUENCE

Seven MR scanners from four manufacturers with 1.5 T and 3 T magnetic field strengths. DWIs were acquired using spin-echo sequences and high-resolution T2WIs using the T2-SPACE sequence.

ASSESSMENT

An experienced, board-certified radiologist evaluated the effectiveness of restoring high-resolution T2WI and harmonizing diverse DWI with metrics such as PSNR and SSIM, and the texture and frequency attributes were further analyzed using gray-level co-occurrence matrix and 1-dimensional power spectral density. The model's impact on machine-specific characteristics was gauged through the performance metrics of a ResNet-50 model. Comprehensive statistical tests were employed for statistical robustness, including McNemar's test and the Dice index.

RESULTS

Our DL protocol reduced DWI contrast and resolution variation. ResNet-50 model's accuracy decreased from 0.9443 to 0.5786, precision from 0.9442 to 0.6494, recall from 0.9443 to 0.5786, and F1 score from 0.9438 to 0.5587. The t-SNE visualization indicated more consistent image features across multiple MR devices. Autoencoder halved learning iterations; Dice coefficient >0.74 confirmed signal reproducibility in 84 lesions.

CONCLUSION

This study presents a DL strategy to mitigate batch effects in diffusion MR images, improving their quality and generalizability.

EVIDENCE LEVEL

3 TECHNICAL EFFICACY: Stage 1.

Machine learning prediction of hospitalization costs for coronary artery bypass grafting operations

BACKGROUND

With the steady rise in health care expenditures, the examination of factors that may influence the costs of care has garnered much attention. Although machine learning models have previously been applied in health economics, their application within cardiac surgery remains limited. We evaluated several machine learning algorithms to model hospitalization costs for coronary artery bypass grafting.

METHODS

All adult hospitalizations for isolated coronary artery bypass grafting were identified in the 2016 to 2020 Nationwide Readmissions Database. Machine learning models were trained to predict expenditures and compared with traditional linear regression. Given the significance of postoperative length of stay, we additionally developed models excluding postoperative length of stay to uncover other drivers of costs. To facilitate comparison, machine learning classification models were also trained to predict patients in the highest decile of costs. Significant factors associated with high cost were identified using SHapley Additive exPlanations beeswarm plots.

RESULTS

Among 444,740 hospitalizations included for analysis, the median cost of hospitalization in coronary artery bypass grafting patients was $43,103. eXtreme Gradient Boosting most accurately predicted hospitalization costs, with R2 = 0.519 over the validation set. The top predictive features in the eXtreme Gradient Boosting model included elective procedure status, prolonged mechanical ventilation, new-onset respiratory failure or myocardial infarction, and postoperative length of stay. After removing postoperative length of stay, eXtreme Gradient Boosting remained the most accurate model (R2 = 0.38). Prolonged ventilation, respiratory failure, and elective status remained important predictive parameters.

CONCLUSION

Machine learning models appear to accurately model total hospitalization costs for coronary artery bypass grafting. Future work is warranted to uncover other drivers of costs and improve the value of care in cardiac surgery.

A Comprehensive Survey on Source-Free Domain Adaptation

Over the past decade, domain adaptation has become a widely studied branch of transfer learning which aims to improve performance on target domains by leveraging knowledge from the source domain. Conventional domain adaptation methods often assume access to both source and target domain data simultaneously, which may not be feasible in real-world scenarios due to privacy and confidentiality concerns. As a result, the research of Source-Free Domain Adaptation (SFDA) has drawn growing attention in recent years, which only utilizes the source-trained model and unlabeled target data to adapt to the target domain. Despite the rapid explosion of SFDA work, there has been no timely and comprehensive survey in the field. To fill this gap, we provide a comprehensive survey of recent advances in SFDA and organize them into a unified categorization scheme based on the framework of transfer learning. Instead of presenting each approach independently, we modularize several components of each method to more clearly illustrate their relationships and mechanisms in light of the composite properties of each method. Furthermore, we compare the results of more than 30 representative SFDA methods on three popular classification benchmarks, namely Office-31, Office-home, and VisDA, to explore the effectiveness of various technical routes and the combination effects among them. Additionally, we briefly introduce the applications of SFDA and related fields. Drawing on our analysis of the challenges confronting SFDA, we offer some insights into future research directions and potential settings.

EFG-CS: Predicting chemical shifts from amino acid sequences with protein structure prediction using machine learning and deep learning models

Nuclear magnetic resonance (NMR) crystallography is one of the main methods in structural biology for analyzing protein stereochemistry and structure. The chemical shift of the resonance frequency reflects the effect of the protons in a molecule producing distinct NMR signals in different chemical environments. Apprehending chemical shifts from NMR signals can be challenging since having an NMR structure does not necessarily provide all the required chemical shift information, making predictive models essential for accurately deducing chemical shifts, either from protein structures or, more ideally, directly from amino acid sequences. Here, we present EFG-CS, a web server that specializes in chemical shift prediction. EFG-CS employs a machine learning-based transfer prediction model for backbone atom chemical shift prediction, using ESMFold-predicted protein structures. Additionally, ESG-CS incorporates a graph neural network-based model to provide comprehensive side-chain atom chemical shift predictions. Our method demonstrated reliable performance in backbone atom prediction, achieving comparable accuracy levels with root mean square errors (RMSE) of 0.30 ppm for H, 0.22 ppm for Hα, 0.89 ppm for C, 0.89 ppm for Cα, 0.84 ppm for Cβ, and 1.69 ppm for N. Moreover, our approach also showed predictive capabilities in side-chain atom chemical shift prediction achieving RMSE values of 0.71 ppm for Hβ, 0.74-1.15 ppm for Hδ, and 0.58-0.94 ppm for Hγ, solely utilizing amino acid sequences without homology or feature curation. This work shows for the first time that generative AI protein models can predict NMR shifts nearly comparable to experimental models. This web server is freely available at https://biosig.lab.uq.edu.au/efg_cs, and the chemical shift prediction results can be downloaded in tabular format and visualized in 3D format.

Construction of lung cancer serum markers based on ReliefF feature selection

Serum miRNAs are available clinical samples for cancer screening. Identifying early serum markers in lung cancer (LC) is essential for patients' early diagnosis and clinical treatment. Expression data of serum miRNAs of lung adenocarcinoma (LUAD) patients and healthy individuals were downloaded from the Gene Expression Omnibus (GEO). These data were normalized and subjected to differential expression analysis to obtain differentially expressed miRNAs (DEmiRNAs). The DEmiRNAs were subsequently subjected to ReliefF feature selection, and subsets closely related to cancer were screened as candidate feature miRNAs. Thereafter, a Gaussian Naive Bayes (NB), Support Vector Machine (SVM), and Random Forest (RF) classifier were constructed based on these candidate feature miRNAs. Then the best diagnostic signature was constructed through NB combined with incremental feature selection (IFS). Thereafter, these samples were subjected to principal component analysis (PCA) based on miRNAs with optimal predictive performance. Finally, the peripheral serum miRNAs of 64 LUAD patients and 59 normal individuals were extracted for qRT-PCR analysis to validate the performance of the diagnostic model in respect of clinical detection. Finally, according to area under the curve (AUC) and accuracy values, the NB classifier composed of miR-5100 and miR-663a manifested the most outstanding diagnostic performance. The PCA results also revealed that the 2-miRNA diagnostic signature could effectively distinguish cancer patients from healthy individuals. Finally, qRT-PCR results of clinical serum samples revealed that miR-5100 and miR-663a expression in tumor samples was remarkably higher than that in normal samples. The AUC of the 2-miRNA diagnostic signature was 0.968. In summary, we identified markers (miR-5100 and miR-663a) in serum for early LUAD screening, providing ideas for developing early LUAD diagnostic models.

Skin-CAD: Explainable deep learning classification of skin cancer from dermoscopic images by feature selection of dual high-level CNNs features and transfer learning

Skin cancer (SC) significantly impacts many individuals' health all over the globe. Hence, it is imperative to promptly identify and diagnose such conditions at their earliest stages using dermoscopic imaging. Computer-aided diagnosis (CAD) methods relying on deep learning techniques especially convolutional neural networks (CNN) can effectively address this issue with outstanding outcomes. Nevertheless, such black box methodologies lead to a deficiency in confidence as dermatologists are incapable of comprehending and verifying the predictions that were made by these models. This article presents an advanced an explainable artificial intelligence (XAI) based CAD system named "Skin-CAD" which is utilized for the classification of dermoscopic photographs of SC. The system accurately categorises the photographs into two categories: benign or malignant, and further classifies them into seven subclasses of SC. Skin-CAD employs four CNNs of different topologies and deep layers. It gathers features out of a pair of deep layers of every CNN, particularly the final pooling and fully connected layers, rather than merely depending on attributes from a single deep layer. Skin-CAD applies the principal component analysis (PCA) dimensionality reduction approach to minimise the dimensions of pooling layer features. This also reduces the complexity of the training procedure compared to using deep features from a CNN that has a substantial size. Furthermore, it combines the reduced pooling features with the fully connected features of each CNN. Additionally, Skin-CAD integrates the dual-layer features of the four CNNs instead of entirely depending on the features of a single CNN architecture. In the end, it utilizes a feature selection step to determine the most important deep attributes. This helps to decrease the general size of the feature set and streamline the classification process. Predictions are analysed in more depth using the local interpretable model-agnostic explanations (LIME) approach. This method is used to create visual interpretations that align with an already existing viewpoint and adhere to recommended standards for general clarifications. Two benchmark datasets are employed to validate the efficiency of Skin-CAD which are the Skin Cancer: Malignant vs. Benign and HAM10000 datasets. The maximum accuracy achieved using Skin-CAD is 97.2 % and 96.5 % for the Skin Cancer: Malignant vs. Benign and HAM10000 datasets respectively. The findings of Skin-CAD demonstrate its potential to assist professional dermatologists in detecting and classifying SC precisely and quickly.

Uncertainty-guided label correction with wavelet-transformed discriminative representation enhancement

Label noises, categorized into closed-set noise and open-set noise, are prevalent in real-world scenarios and can seriously hinder the generalization ability of models. Identifying noise is challenging because noisy samples closely resemble true positives. Existing approaches often assume a single noise source, oversimplify closed-set noise, or treat open-set noise as toxic and eliminate it, resulting in limited practical effects. To address these issues, we present a novel approach named uncertainty-guided label correction with wavelet-transformed discriminative representation enhancement (Ultra), designed to mitigate the effects of mixed noise. Specifically, our approach considers a more practical noise setting. To achieve robust mixed-noise identification, we initially look into a learnable wavelet filter for obtaining discriminative features and filtering spurious cues automatically at the representation level. Subsequently, we introduce a two-fold uncertainty estimation to stably locate noise within the corrupted supervised signal level. These insights pave the way for a simple yet potent label correction technique, enabling comprehensive utilization of open-set noise, which can be rendered non-toxic in a specific manner, in contrast to harmful closed-set noise. Experimental validation on datasets with synthetic mixed noise, web noise corruption, and a real-world dataset confirms the effectiveness and generality of Ultra. Furthermore, our approach enhances the application of efficient techniques (e.g., supervised contrastive learning) within label noise scenarios.

Meta-heuristic optimization algorithms based feature selection for joint moment prediction of sit-to-stand movement using machine learning algorithms

The sit-to-stand (STS) movement is fundamental in daily activities, involving coordinated motion of the lower extremities and trunk, which leads to the generation of joint moments based on joint angles and limb properties. Traditional methods for determining joint moments often involve sensors or complex mathematical approaches, posing limitations in terms of movement restrictions or expertise requirements. Machine learning (ML) algorithms have emerged as promising tools for joint moment estimation, but the challenge lies in efficiently selecting relevant features from diverse datasets, especially in clinical research settings. This study aims to address this challenge by leveraging metaheuristic optimization algorithms to predict joint moments during STS using minimal input data. Motion analysis data from 20 participants with varied mass and inertia properties are utilized, and joint angles are computed alongside simulations of joint moments. Feature selection is performed using the Manta Ray Foraging Optimization (MRFO), Marine Predators Algorithm (MPA), and Equilibrium Optimizer (EO) algorithms. Subsequently, Decision Tree Regression (DTR), Random Forest Regression (RFR), Extra Tree Regression (ETR), and eXtreme Gradient Boosting Regression (XGBoost Regression) ML algorithms are deployed for joint moment prediction. The results reveal EO-ETR as the most effective algorithm for ankle, knee, and neck joint moment prediction, while MPA-ETR exhibits superior performance for hip joint prediction. This approach demonstrates potential for enhancing accuracy in joint moment estimation with minimal feature input, offering implications for biomechanical research and clinical applications.

The systematic analysis of adults' environmental sensory tendencies dataset

This study was conducted to investigate the profound impact of human activities on the environment, based on scientific data, recognizing the potential of environmental problems to turn into devastating crises if appropriate measures are not taken. It emphasizes the important role of education in developing environmental awareness, knowledge and sensitivity to counter adverse environmental consequences. For this purpose, a dataset was created for the emotional tendencies of university students, who represent a demographic that has the potential to influence the sustainable future of the world. A survey data including 34 different variables was collected from 388 university students in Turkey. Environmental Sensory Tendencies Dataset is intended to provide valuable guidance for the development of effective environmental education programs and policies aimed at increasing university students' awareness and participation in environmental issues. Our research underlines the vital importance of developing responsible attitudes and behaviors to effectively address environmental challenges and thereby contribute to a healthier and more sustainable global ecosystem. This study will make a significant contribution to the literature and highlight the interconnection between human actions and environmental well-being.

Towards equitable AI in oncology

Artificial intelligence (AI) stands at the threshold of revolutionizing clinical oncology, with considerable potential to improve early cancer detection and risk assessment, and to enable more accurate personalized treatment recommendations. However, a notable imbalance exists in the distribution of the benefits of AI, which disproportionately favour those living in specific geographical locations and in specific populations. In this Perspective, we discuss the need to foster the development of equitable AI tools that are both accurate in and accessible to a diverse range of patient populations, including those in low-income to middle-income countries. We also discuss some of the challenges and potential solutions in attaining equitable AI, including addressing the historically limited representation of diverse populations in existing clinical datasets and the use of inadequate clinical validation methods. Additionally, we focus on extant sources of inequity including the type of model approach (such as deep learning, and feature engineering-based methods), the implications of dataset curation strategies, the need for rigorous validation across a variety of populations and settings, and the risk of introducing contextual bias that comes with developing tools predominantly in high-income countries.

Czech medical coding assistant based on transformer networks

The International Classification of Diseases (ICD) hierarchical taxonomy is used for so-called clinical coding of medical reports, typically presented in unstructured text. In the Czech Republic, it is currently carried out manually by a so-called clinical coder. However, due to the human factor, this process is error-prone and expensive. The coder needs to be properly trained and spends significant effort on each report, leading to occasional mistakes. The main goal of this paper is to propose and implement a system that serves as an assistant to the coder and automatically predicts diagnosis codes. These predictions are then presented to the coder for approval or correction, aiming to enhance efficiency and accuracy. We consider two classification tasks: main (principal) diagnosis; and all diagnoses. Crucial requirements for the implementation include minimal memory consumption, generality, ease of portability, and sustainability. The main contribution lies in the proposal and evaluation of ICD classification models for the Czech language with relatively few training parameters, allowing swift utilisation on the prevalent computer systems within Czech hospitals and enabling easy retraining or fine-tuning with newly available data. First, we introduce a small transformer-based model for each task followed by the design of a transformer-based "Four-headed" model incorporating four distinct classification heads. This model achieves comparable, sometimes even better results, against four individual models. Moreover this novel model significantly economises memory usage and learning time. We also show that our models achieve comparable results against state-of-the-art English models on the Mimic IV dataset even though our models are significantly smaller.

Metabolomic and lipidomic fingerprints in inflammatory skin diseases - Systemic illumination of atopic dermatitis, hidradenitis suppurativa and plaque psoriasis

Auto-inflammatory skin diseases place considerable symptomatic and emotional burden on the affected and put pressure on healthcare expenditures. Although most apparent symptoms manifest on the skin, the systemic inflammation merits a deeper analysis beyond the surface. We set out to identify systemic commonalities, as well as differences in the metabolome and lipidome when comparing between diseases and healthy controls. Lipidomic and metabolomic LC-MS profiling was applied, using plasma samples collected from patients suffering from atopic dermatitis, plaque-type psoriasis or hidradenitis suppurativa or healthy controls. Plasma profiles revealed a notable shift in the non-enzymatic anti-oxidant defense in all three inflammatory disorders, placing cysteine metabolism at the center of potential dysregulation. Lipid network enrichment additionally indicated the disease-specific provision of lipid mediators associated with key roles in inflammation signaling. These findings will help to disentangle the systemic components of autoimmune dermatological diseases, paving the way to individualized therapy and improved prognosis.

Patient stratification based on the risk of severe illness in emergency departments through collaborative machine learning models

OBJECTIVES

Emergency department (ED) overcrowding presents a global challenge that inhibits prompt care for critically ill patients. Traditional 5-level triage system that heavily rely on the judgment of the triage staff could fail to detect subtle symptoms in critical patients, thus leading to delayed treatment. Unlike previous rivalry-focused approaches, our study aimed to establish a collaborative machine learning (ML) model that renders risk scores for severe illness, which may assist the triage staff to provide a better patient stratification for timely critical cares.

METHODS

This retrospective study was conducted at a tertiary teaching hospital. Data were collected from January 2015 to October 2022. Demographic and clinical information were collected at triage. The study focused on severe illness as the outcome. We developed artificial neural network (ANN) models, with or without utilizing the Taiwan Triage and Acuity Scale (TTAS) score as one of the predictors. The model using the TTAS score is termed a machine-human collaborative model (ANN-MH), while the model without it is referred to as a machine-only model (ANN-MO). The predictive power of these models was assessed using the area under the receiver-operating-characteristic (AUROC) and the precision-recall curves (AUPRC); their sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and F1 score were compared.

RESULTS

The study analyzed 668,602 ED visits from 2015 to 2022. Among them, 278,724 visits from 2015 to 2018 were used for model training and validation, while 320,201 visits from 2019 to 2022 were for testing model performance. Approximately 2.6% of visits were by severely ill patients, whose TTAS scores ranged from 1 to 5. The ANN-MH model achieved a testing AUROC of 0.918 and AUPRC of 0.369, while for the ANN-MO model the AUROC and AUPRC were 0.909 and 0.339, respectively. Based on these metrics, the ANN-MH model outperformed the ANN-MO model, and both surpassed human triage classification. Subgroup analyses further highlighted the models' capability to identify higher-risk patients within the same triage level.

CONCLUSIONS

The traditional 5-level triage system often falls short, leading to under-triage of critical patients. Our models include a score-based differentiation within a triage level to offer advanced risk stratification, thereby promoting patient safety.

Evidential Multi-Source-Free Unsupervised Domain Adaptation

Multi-Source-Free Unsupervised Domain Adaptation (MSFUDA) requires aggregating knowledge from multiple source models and adapting it to the target domain. Two challenges remain: 1) suboptimal coarse-grained (domain-level) aggregation of multiple source models, and 2) risky semantics propagation based on local structures. In this article, we propose an evidential learning method for MSFUDA, where we formulate two uncertainties, i.e. Evidential Prediction Uncertainty (EPU) and Evidential Adjacency-Consistent Uncertainty (EAU), respectively for addressing the two challenges. The former, EPU, captures the uncertainty of a sample fitted to a source model, which can suggest the preferences of target samples for different source models. Based on this, we develop an EPU-Based Multi-Source Aggregation module to achieve fine-grained, instance-level source knowledge aggregation. The latter, EAU, provides a robust measure of consistency among adjacent samples in the target domain. Utilizing this, we develop an EAU-Guided Local Structure Mining module to ensure the trustworthy propagation of semantics. The two modules are integrated into the Evidential Aggregation and Adaptation Framework (EAAF), and we demonstrated that this framework achieves state-of-the-art performances on three MSFUDA benchmarks.

A deep reinforcement learning algorithm framework for solving multi-objective traveling salesman problem based on feature transformation

As a special type of multi-objective combinatorial optimization problems (MOCOPs), the multi-objective traveling salesman problem (MOTSP) plays an important role in practical fields such as transportation and robot control. However, due to the complexity of its solution space and the conflicts between different objectives, it is difficult to obtain satisfactory solutions in a short time. This paper proposes an end-to-end algorithm framework for solving MOTSP based on deep reinforcement learning (DRL). By decomposing strategies, solving MOTSP is transformed into solving multiple single-objective optimization subproblems. Through linear transformation, the features of the MOTSP are combined with the weights of the objective function. Subsequently, a modified graph pointer network (GPN) model is used to solve the decomposed subproblems. Compared with the previous DRL model, the proposed algorithm can solve all the subproblems using only one model without adding weight information as input features. Furthermore, our algorithm can output a corresponding solution for each weight, which increases the diversity of solutions. In order to verify the performance of our proposed algorithm, it is compared with four classical evolutionary algorithms and two DRL algorithms on several MOTSP instances. The comparison shows that our proposed algorithm outperforms the compared algorithms both in terms of training time and the quality of the resulting solutions.

Exploring the potential of representation and transfer learning for anatomical neuroimaging: Application to psychiatry

The perspective of personalized medicine for brain disorders requires efficient learning models for anatomical neuroimaging-based prediction of clinical conditions. There is now a consensus on the benefit of deep learning (DL) in addressing many medical imaging tasks, such as image segmentation. However, for single-subject prediction problems, recent studies yielded contradictory results when comparing DL with Standard Machine Learning (SML) on top of classical feature extraction. Most existing comparative studies were limited in predicting phenotypes of little clinical interest, such as sex and age, and using a single dataset. Moreover, they conducted a limited analysis of the employed image pre-processing and feature selection strategies. This paper extensively compares DL and SML prediction capacity on five multi-site problems, including three increasingly complex clinical applications in psychiatry namely schizophrenia, bipolar disorder, and Autism Spectrum Disorder (ASD) diagnosis. To compensate for the relative scarcity of neuroimaging data on these clinical datasets, we also evaluate three pre-training strategies for transfer learning from brain imaging of the general healthy population: self-supervised learning, generative modeling and supervised learning with age. Overall, we find similar performance between randomly initialized DL and SML for the three clinical tasks and a similar scaling trend for sex prediction. This was replicated on an external dataset. We also show highly correlated discriminative brain regions between DL and linear ML models in all problems. Nonetheless, we demonstrate that self-supervised pre-training on large-scale healthy population imaging datasets (N≈10k), along with Deep Ensemble, allows DL to learn robust and transferable representations to smaller-scale clinical datasets (N≤1k). It largely outperforms SML on 2 out of 3 clinical tasks both in internal and external test sets. These findings suggest that the improvement of DL over SML in anatomical neuroimaging mainly comes from its capacity to learn meaningful and useful abstract representations of the brain anatomy, and it sheds light on the potential of transfer learning for personalized medicine in psychiatry.

Learn to optimize-a brief overview

Most optimization problems of practical significance are typically solved by highly configurable parameterized algorithms. To achieve the best performance on a problem instance, a trial-and-error configuration process is required, which is very costly and even prohibitive for problems that are already computationally intensive, e.g. optimization problems associated with machine learning tasks. In the past decades, many studies have been conducted to accelerate the tedious configuration process by learning from a set of training instances. This article refers to these studies as learn to optimize and reviews the progress achieved.

Interactive Virtual Ankle Movement Controlled by Wrist sEMG Improves Motor Imagery: An Exploratory Study

Virtual reality (VR) techniques can significantly enhance motor imagery training by creating a strong illusion of action for central sensory stimulation. In this article, we establish a precedent by using surface electromyography (sEMG) of contralateral wrist movement to trigger virtual ankle movement through an improved data-driven approach with a continuous sEMG signal for fast and accurate intention recognition. Our developed VR interactive system can provide feedback training for stroke patients in the early stages, even if there is no active ankle movement. Our objectives are to evaluate: 1) the effects of VR immersion mode on body illusion, kinesthetic illusion, and motor imagery performance in stroke patients; 2) the effects of motivation and attention when utilizing wrist sEMG as a trigger signal for virtual ankle motion; 3) the acute effects on motor function in stroke patients. Through a series of well-designed experiments, we have found that, compared to the 2D condition, VR significantly increases the degree of kinesthetic illusion and body ownership of the patients, and improves their motor imagery performance and motor memory. When compared to conditions without feedback, using contralateral wrist sEMG signals as trigger signals for virtual ankle movement enhances patients' sustained attention and motivation during repetitive tasks. Furthermore, the combination of VR and feedback has an acute impact on motor function. Our exploratory study suggests that the sEMG-based immersive virtual interactive feedback provides an effective option for active rehabilitation training for severe hemiplegia patients in the early stages, with great potential for clinical application.

TopicRefiner: Coherence-Guided Steerable LDA for Visual Topic Enhancement

This article presents a new Human-steerable Topic Modeling (HSTM) technique. Unlike existing techniques commonly relying on matrix decomposition-based topic models, we extend LDA as the fundamental component for extracting topics. LDA's high popularity and technical characteristics, such as better topic quality and no need to cherry-pick terms to construct the document-term matrix, ensure better applicability. Our research revolves around two inherent limitations of LDA. First, the principle of LDA is complex. Its calculation process is stochastic and difficult to control. We thus give a weighting method to incorporate users' refinements into the Gibbs sampling to control LDA. Second, LDA often runs on a corpus with massive terms and documents, forming a vast search space for users to find semantically relevant or irrelevant objects. We thus design a visual editing framework based on the coherence metric, proven to be the most consistent with human perception in assessing topic quality, to guide users' interactive refinements. Cases on two open real-world datasets, participants' performance in a user study, and quantitative experiment results demonstrate the usability and effectiveness of the proposed technique.