A guide to machine learning for biologists

Deep learning for the prediction of acute kidney injury after coronary angiography and intervention in patients with chronic kidney disease: a model development and validation study

BACKGROUND

Patients with chronic kidney disease (CKD) are considered the primary population at risk for post-contrast acute kidney injury (PC-AKI), yet there are few predictive tools specifically designed for this vulnerable population.

METHODS

Adult CKD patients undergoing coronary angiography or percutaneous coronary intervention at the Second Xiangya Hospital (2015-2021) were enrolled. The patients were divided into a derivation cohort and a validation cohort based on their admission dates. The primary outcome was the development of PC-AKI. The random forest algorithm was used to identify the most influential predictors of PC-AKI. Six machine learning algorithms were used to construct predictive models for PC-AKI. Model 1 included only preoperative variables, whereas Model 2 included both preoperative and intraoperative variables. The Mehran score was included in the comparison as a classic postoperative predictive model for PC-AKI.

RESULTS

Among the 989 CKD patients enrolled, 125 (12.6%) developed PC-AKI. In the validation cohort, deep neural network (DNN) outperformed other machine learning models with the area under the receiver operating characteristic curve (AUROC) of 0.733 (95% CI 0.654-0.812) for Model 1 and 0.770 (95% CI 0.695-0.845) for Model 2. Furthermore, Model 2 showed better performance compared to the Mehran score (AUROC 0.631, 95% CI 0.538-0.724). The SHapley Additive exPlanations method provided interpretability for the DNN models. A web-based tool was established to help clinicians stratify the risk of PC-AKI (https://xydsbakigroup.streamlit.app/).

CONCLUSION

The explainable DNN models serve as promising tools for predicting PC-AKI in CKD patients undergoing coronary angiography and intervention, which is crucial for risk stratification and clinical descion-making.

Monocyte distribution width (MDW) as a reliable diagnostic biomarker for sepsis in patients with HIV

Sepsis is a leading cause of death among patients with HIV, but early diagnosis remains a challenge. This study evaluates the diagnostic performance of monocyte distribution width (MDW) in detecting sepsis in patients with HIV. A prospective observational study was conducted at Shanghai Public Health Center, involving 488 hospitalized patients with HIV aged 18-65 between December 2022 and August 2023. MDW was measured at admission, and its diagnostic accuracy was compared with Sepsis-3 criteria. Survival rates on day 28 and 90 were also recorded. Additionally, five machine learning (ML) models were tested to enhance diagnostic efficacy. Of 488 subjects, 90 were in the sepsis group and 398 in the control group. MDW showed a diagnostic area under the curve (AUC) of 0.82, comparable to C-reactive protein (CRP) and Procalcitonin (PCT) with AUCs of 0.78 and 0.82, respectively. With a cut-off value of 25.25, MDW had a sensitivity of 0.83 and specificity of 0.76. The positive and negative predictive values were 44% and 95%, respectively. When MDW was combined with platelet count, serum albumin, and hemoglobin in a random forest model, the AUC improved to 0.931. The model achieved a sensitivity of 1.00 and specificity of 0.732. MDW is a useful diagnostic marker for sepsis in patients with HIV, with strong sensitivity and specificity. Combining MDW with other lab markers can further enhance diagnostic accuracy.Trial registration: ClinicalTrials.gov identifier: NCT05036928..

Construction and validation of a predictive model for meningoencephalitis in pediatric scrub typhus based on machine learning algorithms

To retrospectively analyze the clinical characteristics of pediatric scrub typhus (ST) with meningoencephalitis (STME) and to construct and validate predictive models using machine learning.Clinical data were collected from 100 cases of pediatric STME and matched with data from 100 ST cases without meningitis using propensity-score matching. Risk factors for STME in pediatrics were identified through the least absolute shrinkage and selection operator (LASSO) regression analysis. Six predictive models-Logistic Regression, K-Nearest Neighbors, Naive Bayes, Multi-layer Perceptron(MLP), Random Forest, and XGBoost-were constructed using the training set and evaluated for performance, with validation conducted on the test set. The Shapley Additive Explanations (SHAP) method was applied to rank the importance of each variable.All children improved and were discharged following treatment with azithromycin/doxycycline (1/99). Twelve variable features were identified through the LASSO regression. Of the six predictive models developed, the XGBoost model demonstrated the highest performance in the training set (AUC = 0.926), though its performance in the test set was moderate (AUC = 0.740). The MLP model exhibited robust predictive performance in both training and test sets, with AUCs of 0.897 and 0.817, respectively. Clinical decision curve analysis indicated that the MLP and XGBoost models provide significant clinical utility. SHAP analysis identified the most important predictors for STME as ferritin, white blood cell count, edema, prothrombin time, fibrinogen, duration of pre-admission fever, eschar, activated partial thromboplastin time, splenomegaly, and headache. The MLP and XGBoost models showed strong predictive capability for pediatric STME, with favorable outcomes following doxycycline-based therapy.

New perspective: An in vitro study on inferring the post-mortem interval (PMI) of human skeletal muscle based on ATR-FTIR spectroscopy combined with machine learning

The determination of the PMI remains one of the most critical challenges within the field of forensic science. Nonetheless, the estimation of PMI has emerged as one of the most complex and challenging domains of research within the field, primarily due to the absence of precise and dependable methodologies, coupled with the advanced decomposition processes of cadavers. To address these challenges, researchers have tried many methods, such as insects, microorganisms, body fluids, and animal tissues. Nevertheless, there is a paucity of research specifically examining human skeletal muscle. In this study, we collected a substantial number of human skeletal muscle samples and employed ATR-FTIR spectroscopy combined with multiple machine learning to analyze spectral changes in skeletal muscle across different PMIs. These methods primarily included PLS-R, PCR, CLS-R, MLR, SVR, and XGB-R to predict the PMI of human skeletal muscle. The results revealed distinct spectral variations in the protein, carbohydrate, and nucleic acid regions of skeletal muscle across different PMIs. Notably, the absorption peak at the 1540 cm-1 amide II band exhibited a gradual decline over time, demonstrating significant potential for PMI estimation and decomposition tracking. Due to its operational simplicity, rapid analysis, and cost-effectiveness, ATR-FTIR spectroscopy combined with machine learning demonstrates significant practical value and forensic potential for PMI estimation.

Deep learning models link local cellular features with whole-animal growth dynamics in zebrafish

Animal growth is driven by the collective actions of cells, which are reciprocally influenced in real-time by the animal's overall growth state. Whereas cell behavior and animal growth state are expected to be tightly coupled, it is not yet determined whether local cellular features at the micrometer scale might correlate with the body size of an animal at the macroscopic level. By inputting 722 skin cell images and corresponding size data for each zebrafish larva into machine learning models, we determined that the Vision Transformer (ViT) with a random cropping and voting strategy was able to achieve high predictive performance (F-score of 0.91). Remarkably, analyzing as few as 27 skin cells within a single image of 0.01 mm2 was sufficient to predict the individual's overall size, ranging from 0.9 to 3.1 mm2 Using a gradient-weighted class activation map (Grad-CAM), we further identified the cellular features influencing the model's decisions. These findings provide a proof-of-concept that macroscopic organismic information may be de-encrypted from a snapshot of only a few dozen cells using deep learning approaches.

Data integrity of food and machine learning: Strategies, advances and prospective

Data integrity is an emerging concept aimed at recording real food properties in the form of data throughout the food lifecycle. However, due to the one-sided nature of current food control data, the comprehensive implementation of data integrity has not been fully achieved. Cause food data integrity realization is required to establish the connection of data-algorithm-application. Machine learning (ML) provides a possibility for the practical carrier of food data integrity. Despite ML is one of top-trend in food quality and safety, ML applications are floating on the surface. The current review does not reveal the relationships behind different algorithms and data patterns. Similarly, due to the rapid development of ML, the current advanced concepts and data explanation tools have not been systematically reviewed. This paper expounds the feasibility of machine learning to achieve data integrity and looks forward to the future vision brought about by artificial intelligence to data integrity.

Artificial intelligence in bacterial diagnostics and antimicrobial susceptibility testing: Current advances and future prospects

Recently, artificial intelligence (AI) has emerged as a transformative tool, enhancing the speed, accuracy, and scalability of bacterial diagnostics. This review explores the role of AI in revolutionizing bacterial detection and antimicrobial susceptibility testing (AST) by leveraging machine learning models, including Random Forest, Support Vector Machines (SVM), and deep learning architectures such as Convolutional Neural Networks (CNNs) and transformers. The integration of AI into these methods promises to address the current limitations of traditional techniques, offering a path toward more efficient, accessible, and reliable diagnostic solutions. In particular, AI-based approaches have demonstrated significant potential in resource-limited settings by enabling cost-effective and portable diagnostic solutions, reducing dependency on specialized infrastructure, and facilitating remote bacterial detection through smartphone-integrated platforms and telemedicine applications. This review highlights AI's transformative role in automating data analysis, minimizing human error, and delivering real-time diagnostic results, ultimately improving patient outcomes and optimizing healthcare efficiency. In addition, we not only examine the current advances in machine learning and deep learning but also review their applications in plate counting, mass spectrometry, morphology-based and motion-based microscopic detection, holographic microscopy, colorimetric and fluorescence detection, electrochemical sensors, Raman and Surface-Enhanced Raman Spectroscopy (SERS), and Atomic Force Microscopy (AFM) for bacterial diagnostics and AST. Finally, we discuss the future directions and potential advancements in AI-driven bacterial diagnostics.

Artificial intelligence to predict cancer risk, are we there yet? A comprehensive review across cancer types

Cancer remains the second leading cause of death worldwide, representing a substantial challenge to global health. Although traditional risk prediction models have played a crucial role in epidemiology of several cancer types, they have limitations especially in the ability to process complex and multidimensional data. In contrast, artificial intelligence (AI) approaches represent a promising solution to overcome this limitation. AI techniques have the potential to identify complex patterns and relationships in data that traditional methods might overlook, making them especially useful for handling large and heterogeneous datasets analysed in cancer research. This review first examines the current state of the art of AI techniques, highlighting their differences and suitability for various data types. Then, offers a comprehensive analysis of the literature, focusing on the application of AI approaches in nineteen cancer types (bladder cancer, breast cancer, cervical cancer, colorectal cancer, endometrial cancer, esophageal cancer, gastric cancer, gynaecological cancers, head and neck cancer, haematological cancers, kidney cancer, liver cancer, lung cancer, melanoma, ovarian cancer, pancreatic cancer, prostate cancer, thyroid cancer and overall cancer), evaluating the models, metrics, and exposure variables used. Finally, the review discusses the application of AI in the clinical practice, along with an assessment of its potential limitations and future directions.

Non-invasive classification of non-neoplastic and neoplastic gallbladder polyps based on clinical imaging and ultrasound radiomics features: An interpretable machine learning model

BACKGROUND

Gallbladder (GB) adenomas, precancerous lesions for gallbladder carcinoma (GBC), lack reliable non-invasive tools for preoperative differentiation of neoplastic polyps from cholesterol polyps. This study aimed to evaluate an interpretable machine learning (ML) combined model for the precise differentiation of the pathological nature of gallbladder polyps (GPs).

METHODS

This study consecutively enrolled 744 patients from Xi'an Jiaotong University First Affiliated Hospital between January 2017 and December 2023 who were pathologically diagnosed postoperatively with cholesterol polyps, adenomas or T1-stage GBC. Radiomics features were extracted and selected, while clinical variables were subjected to univariate and multivariate logistic regression analyses to identify significant predictors of neoplastic polyps. A optimal ML-based radiomics model was developed, and separate clinical, US and combined models were constructed. Finally, SHapley Additive exPlanations (SHAP) was employed to visualize the classification process.

RESULTS

The areas under the curves (AUCs) of the CatBoost-based radiomics model were 0.852 (95 % CI: 0.818-0.884) and 0.824 (95 % CI: 0.758-0.881) for the training and test sets, respectively. The combined model demonstrated the best performance with an improved AUC of 0.910 (95 % CI: 0.885-0.934) and 0.869 (95 % CI: 0.812-0.919), outperformed the clinical, radiomics, and US model (all P < 0.05), and reduced the rate of unnecessary cholecystectomies. SHAP analysis revealed that the polyp short diameter is a crucial independent risk factor in predicting the nature of the GPs.

CONCLUSION

The ML-based combined model may be an effective non-invasive tool for improving the precision treatment of GPs, utilizing SHAP to visualize the classification process can enhance its clinical application.

Exploring pesticide risk in autism via integrative machine learning and network toxicology

Autism Spectrum Disorder (ASD) is a prevalent neurodevelopmental condition influenced by both genetic and environmental factors, including pesticide exposure. This study aims to investigate the pathogenic mechanisms of ASD and identify potential causative pesticides by integrating bioinformatics, machine learning, network toxicology, and molecular docking approaches. A total of 156 differentially expressed genes (128 upregulated and 28 downregulated) were identified from ASD-related transcriptomic datasets. Using the LASSO algorithm, 23 key targets were initially selected. Each combination of 1-23 targets was used to construct predictive models using eight different machine learning algorithms. The Stochastic Gradient Descent (SGD) model demonstrated the best predictive performance for 20 features, which were defined as hub targets. These targets were subsequently used in a network toxicology framework to screen for associated environmental toxicants. Three pesticide candidates-epoxiconazole, flusilazole, and DEET-were identified as strongly interacting with these core targets. Molecular docking analysis further validated stable binding affinities between these pesticides and the hub targets. Functional enrichment analysis revealed significant involvement of glycosylation-related pathways, including mucin-type O-glycan biosynthesis, implicating potential mechanisms in ASD pathogenesis. Collectively, our findings highlight novel biomolecular links between pesticide exposure and ASD risk, and propose a set of candidate biomarkers and toxicants for further experimental validation and regulatory consideration.

Evolution of gene regulatory networks in insects

Changes in gene regulatory networks (GRNs) underlying the evolution of traits have been intensively studied, with insects providing excellent model cases. In studies using Drosophila, butterflies, and other insects, several well-known cases have shown that changes in the cis-regulatory region of a gene controlling a trait can result in the co-option of the gene for a role different from that in its original developmental context. When the expression of a regulatory gene that controls the expression of multiple downstream genes is altered, the expression of these downstream genes changes accordingly, representing the simplest form of GRN co-option. Many studies have explored the applicability of this model to the acquisition of new traits, yielding substantial insights. However, no study has yet comprehensively elucidated the co-option of a GRN or the evolution of a network architecture, including associated genes and their regulatory relationships. In the near future, the use of single-cell multiomics and machine learning will allow for larger-scale data analysis, leading to a better understanding of the evolution of traits through the evolution of GRNs.

The performance of machine learning for predicting the recurrent stroke: a systematic review and meta-analysis on 24,350 patients

BACKGROUND

Stroke is a leading cause of death and disability worldwide. Approximately one-third of patients with stroke experienced a second stroke. This study investigates the predictive value of machine learning (ML) algorithms for recurrent stroke.

METHOD

This study was prepared according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guideline. PubMed, Scopus, Embase, and Web of Science (WOS) were searched until January 1, 2024. The quality assessment of studies was conducted using the QUADAS-2 tool. The diagnostic meta-analysis was conducted to calculate the pooled sensitivity, specificity, diagnostic accuracy, positive and negative diagnostic likelihood ratio (DLR), diagnostic accuracy, diagnostic odds ratio (DOR), and area under of the curve (AUC) by the MIDAS package in STATA V.17.

RESULTS

Twelve studies, comprising 24,350 individuals, were included. The meta-analysis revealed a sensitivity of 71% (95% CI 0.64-0.78) and a specificity of 88% (95% confidence interval (CI) 0.76-0.95). Positive and negative DLR were 5.93 (95% CI 3.05-11.55) and 0.33 (95% CI 0.28-0.39), respectively. The diagnostic accuracy and DOR was 2.89 (95% CI 2.32-3.46) and 18.04 (95% CI 10.21-31.87), respectively. The summary ROC curve indicated an AUC of 0.82 (95% CI 0.78-0.85).

CONCLUSION

ML demonstrates promise in predicting recurrent strokes, with moderate to high sensitivity and specificity. However, the high heterogeneity observed underscores the need for standardized approaches and further research to enhance the reliability and generalizability of these models. ML-based recurrent stroke prediction can potentially augment clinical decision-making and improve patient outcomes by identifying high-risk patients.

Metabolome guided treasure hunt - learning from metabolic diversity

Metabolomics is a rapidly evolving field focused on the comprehensive identification and quantification of small molecules in biological systems. As the final layer of the biological hierarchy following of the genome, transcriptome and proteome, it presents a dynamic snapshot of phenotype, influenced by genetic, environmental and physiological factors. Whilst the metabolome sits downstream of genes and proteins, there are multiple higher levels-tissues, organs, the entire organism, and interactions with other organisms, which need to be considered in order to fully comprehend organismal biology. Advances in metabolomics continue to expand its applications in plant biology, biotechnology, and natural product discovery unlocking many of nature's most beneficial colors, tastes, nutrients and medicines. Flavonoids and other specialized metabolites are essential for plant defense against oxidative stress and function as key phytonutrients for human health. Recent advancements in gene-editing and metabolic engineering have significantly improved the nutritional value and flavor of crop plants. Here we highlight how advanced metabolic analysis is driving improvements in crops uncovering genes that influence nutrient and flavor profile and plant derived compounds with medicinal potential.

Prediction of endoscopic restenosis after endoscopic balloon dilation in patients with Crohn's disease: a machine learning approach

BACKGROUND

Endoscopic balloon dilation (EBD) is recognized as a minimally invasive and effective procedure for managing intestinal stenosis in patients with Crohn's disease (CD). It offers an alternative to surgery and has been shown to improve the quality of life for these patients by reducing the need for more aggressive interventions. This study aimed to evaluate factors associated with endoscopic restenosis after EBD and construct a prognostic model.

METHODS

We retrospectively collected and analyzed data on patients receiving EBD treatment at the Sixth Affiliated Hospital of Sun Yat-sen University from 2013 to 2024. Seven machine learning (ML) algorithms were used to construct prognostic models. Subsequently, we conducted comparative tests on the performance of the models to ensure accuracy and reliability.

RESULTS

A total of 135 patients were included in the statistical analysis. 53% occurred endoscopic restenosis, with an average restenosis time of 183 days. COX and logistic regression analysis showed that 4 features including ever-use glucocorticoids, stenosis position, technical success, and albumin level were associated with restenosis risk. When comparing different ML models, CoxPH and LASSO models performed better on various evaluation metrics, including C-index which was greater than 0.7 in the train and test set. Based on SHapley Additive exPlanations (SHAP), stenosis position, balloon diameter, and albumin level were identified as the top 3 important features associated with prognosis.

CONCLUSION

The ML-based prognostic model has good predictive performance and can accurately assess the risk of endoscopic restenosis after EBD treatment.

TRAF3 as a potential diagnostic biomarker for recurrent pregnancy loss: insights from single-cell transcriptomics and machine learning

BACKGROUND

Recurrent pregnancy loss (RPL), characterized by multiple miscarriages, remains a condition with unclear etiology, posing significant challenges for affected women and couples. This study aims to explore the underlying mechanisms of RPL, focusing on the role of decidual Natural Killer (dNK) cells and the TNF receptor-associated factor 3 (TRAF3) gene as a potential diagnostic marker and therapeutic target.

METHODS

We used single-cell transcriptomic analysis and machine learning techniques to analyze decidual tissues from RPL patients and normal pregnancy(NP). Weighted Gene Co-expression Network Analysis (WGCNA) was employed to identify key gene clusters. Validation studies included RT-PCR, immunohistochemistry, and molecular docking analyses.

RESULTS

We observed an increased proportion of specific dNK cell subtypes (dNK2 and dNK3) in the RPL group compared to NP, implicating their role in RPL pathology. dNK cells in RPL primarily interacted with monocytes via the Macrophage Migration Inhibitory Factor (MIF) signaling pathway. Our diagnostic model, incorporating TRAF3 and nine other genes, demonstrated high diagnostic efficiency. TRAF3 expression was significantly lower in the decidua of RPL patients, and Diethylstilbestrol and Metformin were identified as potential modulators of TRAF3.

CONCLUSIONS

This study highlights TRAF3 as a promising diagnostic marker and therapeutic target for RPL. The diagnostic model we developed has potential for early detection and personalized treatment strategies for RPL.

Development and Validation of Machine Learning-Based Marker for Early Detection and Prognosis Stratification of Nonalcoholic Fatty Liver Disease

Nonalcoholic fatty liver disease (NAFLD) is the leading cause of chronic liver disease and is considered the hepatic manifestation of metabolic syndrome, triggering out adverse outcomes. A stacked multimodal machine learning model is constructed and validated for early identification and prognosis stratification of NAFLD by integrating genetic and clinical data sourced from 36 490 UK Biobank and 9 007 Nanfang Hospital participants and extracted its probabilities as in-silico scores for NAFLD (ISNLD). The efficacy of ISNLD is evaluated for the early prediction of severe liver disease (SeLD) and analyzed its association with metabolism-related outcomes. The multimodal model performs satisfactorily in classifying individuals into low- and high-risk groups for NAFLD, achieving area under curves (AUCs) of 0.843, 0.840, and 0.872 within training, internal, and external test sets, respectively. Among high-risk group, ISNLD is significantly associated with intrahepatic and metabolism-related complications after lifestyle factors adjustment. Further, ISNLD demonstrates notable capability for early prediction of SeLD and further stratifies high-risk subjects into three risk subgroups of elevated risk for adverse outcomes. The findings emphasize the model's ability to integrate multimodal features to generate ISNLD, enabling early detection and prognostic prediction of NAFLD. This facilitates personalized stratification for NAFLD and metabolism-related outcomes based on digital non-invasive markers, enabling preventive interventions.

Prediction of coronary heart disease based on klotho levels using machine learning

The diagnostic accuracy for coronary heart disease (CHD) needs to be improved. Some studies have indicated that klotho protein levels upon admission comprise an independent risk factor for CHD and have clinical value for predicting CHD. This study aimed to construct a tool to predict CHD risk by analyzing klotho levels and clinically relevant indicators by using a machine learning (ML) method. We randomly assigned the dataset of the National Health and Nutrition Examination Survey (NHANES) 2007-2016 to training and test sets at a ratio of 70:30. We evaluated the ability of five models constructed using logistic regression, neural networks, random forest, support vector machine, and eXtreme Gradient Boosting to predict CHD. We determined their predictive performance using the following parameters: area under the receiver operating characteristic curve, accuracy, precision, recall, F1, and Brier scores. We analyzed data from 11,583 persons in US NHANES and entered 13 potential predictive variables, including klotho and other clinically relevant indicators, into the feature screening process. We established that the five ML models could predict the onset of CHD. The RF model showed the best predictive performance among the five ML models.

Estimation of postmortem interval under different ambient temperatures based on multi-organ metabolomics and machine learning algorithm

In forensic practice, the estimation of postmortem interval has been a persistent challenge. Recently, there has been an increasing utilization of metabolomics techniques combined with machine learning methods for postmortem interval estimation. When examining metabolite changes from a global perspective, rather than relying on specific substance changes, estimating postmortem interval through machine learning methods is more precise and entails fewer errors. Prior studies have investigated the use of metabolomics to estimate postmortem interval. Nevertheless, most of them focused on analyzing the metabolomic properties of a single organ or biofluid concerning a specific temperature. In this study, we employ the GC-MS platform to identify metabolites in the liver, kidney, and quadriceps femoris muscle of mechanically suffocated Sprague Dawley rats at various temperatures. Multivariable statistical analysis was used to determine differential compounds from the original data. The machine learning method was used to establish models for the estimation of postmortem interval under various ambient temperatures. As indicated by the results, liver, kidney, and quadriceps femoris muscle samples were screened for 24, 18, and 19 differential metabolites respectively, associated with postmortem interval under various ambient temperatures. Based on the metabolites listed above, the support vector regression models were established by utilizing single-organ and multi-organ metabolomics data for postmortem interval estimation. The multi-organ model showed a higher estimation accuracy. Also, a comprehensive generalization postmortem interval estimation model was established with multi-organ metabolomics data and temperature variables, which can be used for the postmortem interval estimation within the temperature range of 5-35℃. These results demonstrate that a multi-organ model utilizing metabolomics techniques can accurately estimate the postmortem interval under various ambient temperatures. Meanwhile, this research establishes a strong foundation for the practical application of metabolomics in postmortem interval estimation.

Clinical prediction of pathological complete response in breast cancer: a machine learning study

BACKGROUND

This study aimed to develop and validate machine learning models to predict pathological complete response (pCR) after neoadjuvant therapy in patients with breast cancer patients.

METHODS

Clinical and pathological data from 1143 patients were analyzed, encompassing variables such as age, gender, marital status, histologic grade, T stage, N stage, months from diagnosis to treatment, molecular subtype, and response to neoadjuvant therapy. Seven machine learning models were trained and validated using both internal and external datasets. Model performance was evaluated using multiple metrics, and interpretability analysis was conducted to assess feature importance.

RESULTS

Key variables influencing pCR included grade, N stage, months from diagnosis to treatment, and molecular subtype. The Naive Bayes model emerged as the most effective, with accuracy (0.746), sensitivity (0.699), specificity (0.808), and F1 score (0.759) surpassing other models. Both internal and external validation confirmed the model's robust predictive power. A web tool was developed for clinical use, aiding in personalized treatment planning. Interpretability analysis further elucidated the contribution of features to pCR prediction, enhancing clinical applicability.

CONCLUSION

The Naive Bayes model provides a robust tool for personalized treatment decisions in patients with breast cancer undergoing neoadjuvant therapy. By accurately predicting pCR rates, it enables clinicians to tailor treatment strategies, potentially improving outcomes.

Molecular insights fast-tracked: AI in biosynthetic pathway research

Covering: 2000 to 2025This review explores the potential of artificial intelligence (AI) in addressing challenges and accelerating molecular insights in biosynthetic pathway research, which is crucial for developing bioactive natural products with applications in pharmacology, agriculture, and biotechnology. It provides an overview of various AI techniques relevant to this research field, including machine learning (ML), deep learning (DL), natural language processing, network analysis, and data mining. AI-powered applications across three main areas, namely, pathway discovery and mining, pathway design, and pathway optimization, are discussed, and the benefits and challenges of integrating omics data and AI for enhanced pathway research are also elucidated. This review also addresses the current limitations, future directions, and the importance of synergy between AI and experimental approaches in unlocking rapid advancements in biosynthetic pathway research. The review concludes with an evaluation of AI's current capabilities and future outlook, emphasizing the transformative impact of AI on biosynthetic pathway research and the potential for new opportunities in the discovery and optimization of bioactive natural products.

Digital health technologies in swallowing care from screening to rehabilitation: A narrative review

OBJECTIVES

Digital health technologies (DHTs) have rapidly advanced in the past two decades, through developments in mobile and wearable devices and most recently with the explosion of artificial intelligence (AI) capabilities and subsequent extension into the health space. DHT has myriad potential applications to deglutology, many of which have undergone promising investigations and developments in recent years. We present the first literature review on applications of DHT in swallowing health, from screening to therapeutics. Public health interventions for swallowing care are increasingly needed in the setting of aging populations in the West and East Asia, and DHT may offer a scalable and low-cost solution.

METHODS

A narrative review was performed using PubMed and Google Scholar to identify recent research on applications of AI and digital health in swallow practice. Database searches, conducted in September 2024, included terms such as "digital," "AI," "machine learning," "tools" in combination with "deglutition," "Otolaryngology," "Head and Neck," "speech language pathology," "swallow," and "dysphagia." Primary literature pertaining to digital health in deglutology was included for review.

RESULTS

We review the various applications of DHT in swallowing care, including prevention, screening, diagnosis, treatment planning and rehabilitation.

CONCLUSION

DHT may offer innovative and scalable solutions for swallowing care as public health needs grow and in the setting of limited specialized healthcare workforce. These technological advances are also being explored as time and resource saving solutions at many points of care in swallow practice. DHT could bring affordable and accurate information for self-management of dysphagia to broader patient populations that otherwise lack access to expert providers.

Explicating the transformative role of artificial intelligence in designing targeted nanomedicine

INTRODUCTION

Artificial intelligence (AI) has emerged as a transformative force in nanomedicine, revolutionizing drug delivery, diagnostics, and personalized treatment. While nanomedicine offers precise targeted drug delivery and reduced toxic effects, its clinical translation is hindered by biological complexity, unpredictable in vivo behavior, and inefficient trial-and-error approaches.

AREAS COVERED

This review covers the application of AI and Machine Learning (ML) across the nanomedicine development pipeline, starting from drug and target identification to nanoparticle design, toxicity prediction, and personalized dosing. Different AI/ML models like QSAR, MTK-QSBER, and Alchemite, along with data sources and high-throughput screening methods, have been explored. Real-world applications are critically discussed, including AI-assisted drug repurposing, controlled-release formulations, and cancer-specific delivery systems.

EXPERT OPINION

AI has emerged as an essential component in designing next-generation nanomedicine. Efficiently handling multidimensional datasets, optimizing formulations, and personalizing treatment regimens, it has sped up the innovation process. However, challenges like data heterogeneity, model transparency, and regulatory gaps remain. Addressing these hurdles through interdisciplinary efforts and emerging innovations like explainable AI and federated learning will pave the way for the clinical translation of AI-driven nanomedicine.

GALR1 and PENK serve as potential biomarkers in invasive non-functional pituitary neuroendocrine tumours

BACKGROUND

Some nonfunctioning pituitary neuroendocrine tumor (NFPitNET) can show invasive growth, which increases the difficulty of surgery and indicates a poor prognosis. However, the molecular mechanism related to invasiveness remains to be further studied. This study is to screen and identify the characteristic biomarkers of invasive NFPitNETs.

METHODS

Based on the data of 73 NFPitNETs microarray chips in the GSE169498 dataset, this study used weighted gene co-expression network (WGCNA), differential expression analysis, protein-protein interaction (PPI) network analysis and various machine learning methods (XGBOOST, LASSO regression, random forest, support vector machine) to screen candidate biomarkers for invasive NFPitNET. Then, using gene set enrichment analysis (GSEA) to explore the differences in biological activities and signaling pathways between invasive NFPitNET and non-invasive NFPitNET. Single-sample GSEA (ssGSEA) was used to analyze key biomarkers-related signaling pathways. Finally, the expression and function of the key biomarkers were verified by q-RT PCR, immunohistochemical (IHC) experiments and in vitro experiments.

RESULTS

Combined with WGCNA and differential expression analysis, 128 high-expression and 85 low-expression candidate biomarkers were preliminarily obtained. PPI analysis and four machine learning algorithms further identified GALR1, PENK and HOXD9. The receiver operating characteristic (ROC) curve results showed that the three biomarkers had good predictive ability of invasiveness. After combining the validation set data, GALR1 and PENK were the final key biomarkers. Finally, PCR and IHC results verified the decreased expression of GALR1 and PENK in invasive NFPitNET and promotes proliferation and invasive ablity of pituitary tumor cells.

CONCLUSION

This study confirmed that the reduced expression of GALR1 and PENK is an important molecular feature of invasive NFPitNETs, which may play an important role in inhibiting the development of NFPitNET.

Feasibility of machine learning-based modeling and prediction to assess osteosarcoma outcomes

Osteosarcoma, an aggressive bone malignancy predominantly affecting children and adolescents, is characterized by a poor prognosis and high mortality rates. The development of reliable prognostic tools is critical for advancing personalized treatment strategies. However, identifying robust gene signatures to predict osteosarcoma outcomes remains a significant challenge. In this study, we analyzed gene expression data from 138 osteosarcoma samples across two multicenter cohorts and identified 14 consensus prognosis-associated genes via univariate Cox regression analysis. Using 66 combinations of 10 machine learning (ML) algorithms, we developed a machine learning-derived prognostic signature (MLDPS) optimized by the average C-index across TARGET, GSE21257, and merged cohorts. The MLDPS effectively stratified osteosarcoma patients into high- and low-risk score groups, achieving strong predictive performance for 1-, 3-, and 5-year overall survival (AUC range: 0.852 - 0.963). The MLDPS, comprising seven genes (CTNNBIP1, CORT, DLX2, TERT, BBS4, SLC7A1, NKX2-3), exhibited superior predictive accuracy compared to 10 established gene signatures. The findings of the MLDPS carry significant clinical implications for osteosarcoma treatment. Patients with a high-risk score demonstrated worse prognosis, increased metastasis risk, reduced immune infiltrations, and greater sensitivity to immunotherapy. Conversely, low-risk patients exhibited prolonged survival and distinct drug sensitivities. These findings underscore the potential of MLDPS to guide risk stratification, inform personalized therapeutic strategies, and improve clinical management in osteosarcoma.

Machine learning and single-cell analysis uncover distinctive characteristics of CD300LG within the TNBC immune microenvironment: experimental validation

Investigating the essential function of CD300LG within the tumor microenvironment in triple-negative breast cancer (TNBC). Transcriptomic and single-cell data from TNBC were systematically collected and integrated. Four machine learning algorithms were employed to identify distinct target genes in TNBC patients. Specifically, CIBERSORT and ssGSEA algorithms were utilized to elucidate immune infiltration patterns, whereas TIDE and TCGA algorithms predicted immune-related outcomes. Moreover, single-cell sequencing data were analyzed to investigate the function of CD300LG-positive cells within the tumor microenvironment. Finally, immunofluorescence staining confirmed the significance of CD300LG in tumor phenotyping. After machine learning screening and independent dataset validation, CD300LG was identified as a unique prognostic biomarker for triple-negative breast cancer. Enrichment analysis revealed that CD300LG expression is strongly linked to immune infiltration and inflammation-related pathways, especially those associated with the cell cycle. The presence of CD8+ T cells and M1-type macrophages was elevated in the CD300LG higher group, whereas the abundance of M2-type macrophage infiltration showed a significant decrease. Immunotherapy prediction models indicated that individuals with low CD300LG expression exhibited better responses to PD-1 therapy. Additionally, single-cell RNA sequencing and immunofluorescence analyses uncovered a robust association between CD300LG and genes involved in tumor invasion. CD300LG plays a pivotal role in the tumor microenvironment of TNBC and represents a promising therapeutic target.

Application of the joint clustering algorithm based on Gaussian kernels and differential privacy in lung cancer identification

In the age of big data, privacy, particularly medical data privacy, is becoming increasingly important. Differential privacy (DP) has emerged as a key method for safeguarding privacy during data analysis and publishing. Cancer identification and classification play a vital role in early detection and treatment. This paper introduces a novel algorithm, DPFCM_GK, which combines differential privacy with fuzzy c-means (FCM) clustering using a Gaussian kernel function. The algorithm enhances cancer detection while ensuring data privacy. Three publicly available lung cancer datasets, along with a dataset from our hospital, are used to test and demonstrate the effectiveness of DPFCM_GK. The experimental results show that DPFCM_GK achieves high clustering accuracy and enhanced privacy as the privacy budget (ε) increases. For the UCIML, NLST, and NSCLC datasets, it reaches optimal results at lower ε (1.52, 1.24, and 2.32) compared to DPFCM. In the lung cancer dataset, DPFCM_GK outperforms DPFCM within, 0.05 ≤ ε ≤ 2.5, with significant differences (χ2 = 4.54 ∼ 29.12; P < 0.05), and both methods converge to an accuracy of 94.5% as ε increases. Although differential privacy initially increases iteration counts, DPFCM_GK demonstrates faster convergence and fewer iterations compared to DPFCM, with significant reductions (T= 23.08, 43.47, and 48.93; P<0.05). For the UCIML dataset, DPFCM_GK significantly reduces runtime compared to other models (DPFCM, LDP-SGD, LDP-Fed, LDP-FedSGD, MGM-DPL, LDP-FL) under the same privacy budget. The runtime reduction is statistically significant with T-values of (T = 21.08, 316.24, 102.35, 222.37, 162.23, 159.25; P < 0.05). DPFCM_GK still maintains excellent time efficiency when applied to the NLST and NSCLC datasets(P < 0.05). For the LLCS dataset, For the LLCS dataset, the DPFCM_GK demonstrates significant improvement as the privacy budget increases, especially in low-budget scenarios, where the performance gap is most pronounced (T=4.20, 8.44, 10.92, 3.95, 7.16, 8.51, P < 0.05). These results confirm DPFCM_GK as a practical solution for medical data analysis, balancing accuracy, privacy, and efficiency.

A novel two-stage feature selection method based on random forest and improved genetic algorithm for enhancing classification in machine learning

The data acquisition methods are becoming increasingly diverse and advanced, leading to higher data dimensions, blurred classification boundaries, and overfitting datasets, affecting machine learning models' accuracy. Many studies have sought to improve model performance through feature selection. However, a single feature selection method has incomplete, unstable, or time-consuming shortcomings. Combining the advantages of various feature selection methods can help overcome these defects. This paper proposes a two-stage feature selection method based on random forest and improved genetic algorithm. First, the importance scores of the random forest are calculated and ranked, and the features are preliminarily eliminated according to the scores, reducing the time complexity of the subsequent process. Then, the improved genetic algorithm is used to search for the global optimal feature subset further. This process introduces a multi-objective fitness function to guide the feature subset, minimizing the number of features in the subset while enhancing classification accuracy. This paper also adds an adaptive mechanism and evolution strategy to improve the loss of population diversity and degeneration in the later stages of iteration, thereby enhancing search efficiency. The experimental results on eight UCI datasets show that the proposed method significantly improves classification performance and has excellent feature selection capability.

Identification of hub immune-related genes and construction of predictive models for systemic lupus erythematosus by bioinformatics combined with machine learning

Systemic lupus erythematosus (SLE) is a chronic autoimmune disease that involves multiple systems. SLE is characterized by the production of autoantibodies and inflammatory tissue damage. This study further explored the role of immune-related genes in SLE. We downloaded the expression profiles of GSE50772 using the Gene Expression Omnibus (GEO) database for differentially expressed genes (DEGs) in SLE. The DEGs were also analyzed for Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment. The gene modules most closely associated with SLE were then derived by Weighted Gene Co-expression Network Analysis (WGCNA). Differentially expressed immune-related genes (DE-IRGs) in SLE were obtained by DEGs, key gene modules and IRGs. The protein-protein interaction (PPI) network was constructed through the STRING database. Three machine learning algorithms were applied to DE-IRGs to screen for hub DE-IRGs. Then, we constructed a diagnostic model. The model was validated by external cohort GSE61635 and peripheral blood mononuclear cells (PBMC) from SLE patients. Immune cell abundance assessment was achieved by CIBERSORT. The hub DE-IRGs and miRNA networks were made accessible through the NetworkAnalyst database. We screened 945 DEGs, which are closely related to the type I interferon pathway and NOD-like receptor signaling pathway. Machine learning identified a total of five hub DE-IRGs (CXCL2, CXCL8, FOS, NFKBIA, CXCR2), and validated in GSE61635 and PBMC from SLE patients. Immune cell abundance analysis showed that the hub genes may be involved in the development of SLE by regulating immune cells (especially neutrophils). In this study, we identified five hub DE-IRGs in SLE and constructed an effective predictive model. These hub genes are closely associated with immune cell in SLE. These may provide new insights into the immune-related pathogenesis of SLE.

Personalized prediction of breast cancer candidates for Anti-HER2 therapy using 18F-FDG PET/CT parameters and machine learning: a dual-center study

Background

Accurately evaluating human epidermal growth factor receptor (HER2) expression status in breast cancer enables clinicians to develop individualized treatment plans and improve patient prognosis. The purpose of this study was to assess the performance of a machine learning (ML) model that was developed using 18F-FDG PET/CT parameters and clinicopathological features in distinguishing different levels of HER2 expression in breast cancer.

Methods

This retrospective study enrolled breast cancer patients who underwent 18F-FDG PET/CT scans prior to treatment at Lianyungang First People's Hospital (centre 1, n=157) and the Third Affiliated Hospital of Soochow University (centre 2, n=84). Two classification tasks were analysed: distinguishing HER2-zero expression from HER2-low/positive expression (Task 1) and distinguishing HER2-low expression from HER2-positive expression (Task 2). For each task, patients from Centre 1 were randomly divided into training and internal test sets at a 7:3 ratio, whereas patients from Centre 2 served as an external test set. The prediction models included logistic regression (LR), support vector machine (SVM), extreme gradient boosting (XGBoost) and multilayer perceptron (MLP), and SHAP analysis provided model interpretability. Model performance was evaluated via the area under the receiver operating characteristic curve (AUC), accuracy, sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV).

Results

XGBoost models exhibited the best predictive performance in both tasks. For Task 1, recursive feature elimination (RFE) was used to select 8 features, excluding pathological features, and the XGBoost model achieved AUCs of 0.888, 0.844 and 0.759 for the training, internal and external testing sets, respectively. The top three features according to the SHAP values were the tumour minimum diameter, mean standardized uptake value (SUVmean) and CTmean. For Task 2, 9 features were selected, including progesterone receptor (PR) status as a pathological feature. The XGBoost model achieved AUCs of 0.920, 0.814 and 0.693 for the training, internal and external testing sets, respectively. The top three features according to the SHAP values were the PR status, maximum tumour diameter and metabolic tumour volume (MTV).

Conclusions

ML models that incorporate 18F-FDG PET/CT parameters and clinicopathological features can aid in the prediction of different HER2 expression statuses in breast cancer.

Evaluating the factors influencing accuracy, interpretability, and reproducibility in the use of machine learning classifiers in biology to enable standardization

The complexity and variability of biological data has promoted the increased use of machine learning methods to understand processes and predict outcomes. These same features complicate reliable, reproducible, interpretable, and responsible use of such methods, resulting in questionable relevance of the derived. outcomes. Here we systematically explore challenges associated with applying machine learning to predict and understand biological processes using a well- characterized in vitro experimental system. We evaluated factors that vary while applying machine learning classifers: (1) type of biochemical signature (transcripts vs. proteins), (2) data curation methods (pre- and post-processing), and (3) choice of machine learning classifier. Using accuracy, generalizability, interpretability, and reproducibility as metrics, we found that the above factors significantly mod- ulate outcomes even within a simple model system. Our results caution against the unregulated use of machine learning methods in the biological sciences, and strongly advocate the need for data standards and validation tool-kits for such studies.

Revealing arginine-cysteine and glycine-cysteine NOS linkages by a systematic re-evaluation of protein structures

Nitrogen-oxygen-sulfur (NOS) linkages act as allosteric redox switches, modulating enzymatic activity in response to redox fluctuations. While NOS linkages in proteins were once assumed to occur only between lysine and cysteine, our investigation shows that these bonds extend beyond the well-studied lysine-NOS-cysteine examples. By systematically analyzing over 86,000 high-resolution X-ray protein structures, we uncovered 69 additional NOS bonds, including arginine-NOS-cysteine and glycine-NOS-cysteine. Our pipeline integrates machine learning, quantum-mechanical calculations, and high-resolution X-ray crystallographic data to systematically detect these subtle covalent interactions and identify key predictive descriptors for their formation. The discovery of these previously unrecognized linkages broadens the scope of protein chemistry and may enable targeted modulation in drug design and protein engineering. Although our study focuses on NOS linkages, the flexibility of this methodology allows for the investigation of a wide range of chemical bonds and covalent modifications, including structurally resolvable posttranslational modifications (PTMs). By revisiting and re-examining well-established protein models, this work underscores how systematic data-driven approaches can uncover hidden aspects of protein chemistry and inspire deeper insights into protein function and stability.

A machine learning-based prediction model for colorectal liver metastasis

Colorectal liver metastasis (CRLM) is a primary factor contributing to poor prognosis and metastasis in colorectal cancer (CRC) patients. This study aims to develop and validate a machine learning (ML)-based risk prediction model using conventional clinical data to forecast the occurrence of CRLM. This retrospective study analyzed the clinical data of 865 CRC patients between January 2018 and September 2024. Patients were categorized into non-CRLM and CRLM groups. The least absolute shrinkage and selection operator regression was employed to identify key clinical variables, and five ML algorithms were utilized to develop prediction models. The optimal model was selected based on performance metrics including the receiver operating characteristic curve, precision-recall curve, decision curve analysis, and calibration curve, which collectively evaluated both the predictive accuracy and clinical utility of the model. Among the five ML algorithms evaluated, Random forest demonstrated the best performance. Leveraging the Random forest algorithm, we developed the CRLM-Lab6 prediction model, which incorporates six features: LDH, CA199, ALT, CEA, TBIL, and AGR. This model exhibits robust predictive performance, achieving an area under the curve of 0.94, a sensitivity of 0.88, and a specificity of 0.93. To enhance its practical utility, the model has been integrated into an accessible web application. This study developed a novel risk prediction model by integrating ML algorithms with conventional laboratory test data to evaluate the likelihood of CRLM occurrence. The model demonstrates excellent predictive performance and has significant clinical application potential.

Machine Learning Classification of Chirality and Optical Rotation Using a Simple One-Hot Encoded Cartesian Coordinate Molecular Representation

Absolute stereochemical configurations and optical rotations were computed for 121,416 molecular structures from the QM9 quantum chemistry data set using density functional theory. A representation for the molecules was developed using Cartesian coordinate geometries and encoded atom types to serve as input for various machine learning algorithms. Classifiers were developed and trained to predict the chirality and signs of optical rotations using a variety of machine learning methods. These methods are compared, and the results demonstrate that machine learning is a viable tool for making predictions of the stereochemical properties of molecules.

Application of machine learning for the analysis of peripheral blood biomarkers in oral mucosal diseases: a cross-sectional study

BACKGROUND

Oral mucosal lesions are widespread globally, have a high prevalence in clinical practice, and significantly impact patients' quality of life. However, their pathogenesis remains unclear. Recent evidences suggested that hematological parameters may play a role in their development. Our study investigated the differences in humoral immune indexes, serum vitamin B levels, and micronutrients among patients with oral mucosal lesions and healthy controls. Additionally, it evaluated a Random Forest machine learning model for classifying various oral mucosal diseases based on peripheral blood biomarkers.

METHODS

We recruited 237 patients with recurrent aphthous ulcers (RAU), 35 with oral lichen planus (OLP), 67 with atrophic glossitis (AG), 35 with burning mouth syndrome (BMS), and 82 healthy controls. Clinical data were analyzed by SPSS 24 software. Serum levels of immunoglobulins (IgG, IgA, IgM), complements (C3, C4), vitamin B (VB1, VB2, VB3, VB5), serum zinc (Serum Zn), serum iron (Serum Fe), unsaturated iron-binding capacity (UIBC), total iron-binding capacity (TIBC), and iron saturation (Iron Sat) were measured and compared among groups. A Random Forest model was applied to analyze a dataset comprising 319 samples with eight key biomarkers.

RESULTS

Significant differences were observed between the oral mucosal diseases groups and controls in the serum levels of VB2, VB3, VB5, zinc, iron, TIBC, and Iron Sat. Specifically, serum levels of VB2 and VB3 were significantly higher in patients compared to controls (*p < 0.05), while levels of VB5, Serum Zn, Serum Fe, TIBC, and Iron Sat were significantly lower (*p < 0.05). No significant differences were found for C3, C4, IgG, IgM, IgA, VB1, and UIBC. The optimized Random Forest model demonstrated high performance, and effectively classified different disease groups, though some overlap between groups was noted. Feature importance analysis, based on the Mean Decrease Accuracy and Gini Index, identified VB2, VB3, Serum Fe, TIBC, and Serum Zn as key biomarkers, indicating their potential in distinguishing oral mucosal diseases.

CONCLUSION

Our study identified significant associations between the contents of VB2, VB3, VB5, Serum Fe, Serum Zn, and other micronutrients and oral mucosal lesions. It suggested that regulating these micronutrient levels could be essential for preventing and curing such lesions. The Random Forest model demonstrated high accuracy (94.68%) in classifying disease groups, emphasizing the potential of machine learning to enhance diagnostic precision in oral mucosal diseases. Future research should focus on validating these findings in larger cohorts and exploring alternative machine-learning algorithms to improve diagnostic accuracy further.

Value of Bioinformatics Models for Predicting Translational Control of Angiogenesis

Angiogenesis, the formation of new blood vessels, is a fundamental biological process with implications for both physiological functions and pathological conditions. While the transcriptional regulation of angiogenesis, mediated by factors such as HIF-1α (hypoxia-inducible factor 1-alpha) and VEGF (vascular endothelial growth factor), is well-characterized, the translational regulation of this process remains underexplored. Bioinformatics has emerged as an indispensable tool for advancing our understanding of translational regulation, offering predictive models that leverage large data sets to guide research and optimize experimental approaches. However, a significant gap persists between bioinformatics experts and other researchers, limiting the accessibility and utility of these tools in the broader scientific community. To address this divide, user-friendly bioinformatics platforms are being developed to democratize access to predictive analytics and empower researchers across disciplines. Translational control, compared with transcriptional control, offers a more energy-efficient mechanism that facilitates rapid cellular responses to environmental changes. Furthermore, transcriptional regulators themselves are often subject to translational control, emphasizing the interconnected nature of these regulatory layers. Investigating translational regulation requires advanced, accessible bioinformatics tools to analyze RNA structures, interacting micro-RNAs, long noncoding RNAs, and RBPs (RNA-binding proteins). Predictive platforms such as RNA structure, human internal ribosome entry site Atlas, and RBPSuite enable the study of RNA motifs and RNA-protein interactions, shedding light on these critical regulatory mechanisms. This review highlights the transformative role of bioinformatics using widely accessible user-friendly tools with a Web-browser interface to elucidate translational regulation in angiogenesis. The bioinformatics tools discussed extend beyond angiogenesis, with applications in diverse fields, including clinical care. By integrating predictive models and experimental insights, researchers can streamline hypothesis generation, reduce experimental costs, and find novel translational regulators. By bridging the bioinformatics knowledge gap, this review aims to empower researchers worldwide to adopt bioinformatics tools in their work, fostering innovation and accelerating scientific discovery.

Predicting breast cancer prognosis based on a novel pathomics model through CHEK1 expression analysis using machine learning algorithms

BACKGROUND

Checkpoint kinase 1 (CHEK1) is often overexpressed in solid tumors. Nonetheless, the prognostic significance of CHEK1 in breast cancer (BrC) remains unclear. This study used pathomics leverages machine learning to predict BrC prognosis based on CHEK1 gene expression..

METHODS

Initially, hematoxylin-eosin (H&E)-stained images obtained from The Cancer Genome Atlas Breast Invasive Carcinoma (TCGA-BRCA) were segmented using Otsu's method. Further, the sub-image features were extracted using machine learning algorithms based on PyRadiomics, mRMRe, and Gradient Boosting Machine (GBM). The predicted CHEK1 expression levels were represented as the pathomics score (PS) and validated using the corresponding RNA-seq data. The prognostic significance of both CHEK1 and PS was evaluated using Kaplan-Meier (KM), and univariate and multivariate Cox regression. The model was assessed by comparing CHEK1 expression by immunohistochemistry (IHC) with PS in BrC tissue microarray (TMA).

RESULTS

A 633 × 10 sub-image set was eligible for training and a 158 × 10 set for validation. 1,488 features were extracted and 8 recursive feature elimination (RFE)-screened features were used to generate the model. A high PS was associated with CHEK1 overexpression, significantly correlating with survival outcomes, especially within 96 months post-diagnosis. Further, patients with high PS responded to anti-programmed cell death protein 1 (anti-PD-1) and anti-cytotoxic T lymphocyte antigen-4 (anti-CTLA4) treatments. In TMA validation, the IHC analysis estimated that high PS similarly predicted poorer prognosis and correlated with higher CHEK1 expression.

CONCLUSIONS

The novel pathomics model reliably predicted CHEK1 expression using machine learning algorithms, which might provide potential clinical utility for prognosis and treatment guidance.

Advancements and implications of artificial intelligence for early detection, diagnosis and tailored treatment of cancer

The complexity and heterogeneity of cancer makes early detection and effective treatment crucial to enhance patient survival and quality of life. The intrinsic creative ability of artificial intelligence (AI) offers improvements in patient screening, diagnosis, and individualized care. Advanced technologies, like computer vision, machine learning, deep learning, and natural language processing, can analyze large datasets and identify patterns that permit early cancer detection, diagnosis, management and incorporation of conclusive treatment plans, ensuring improved quality of life for patients by personalizing care and minimizing unnecessary interventions. Genomics, transcriptomics and proteomics data can be combined with AI algorithms to unveil an extensive overview of cancer biology, assisting in its detailed understanding and will help in identifying new drug targets and developing effective therapies. This can also help to identify personalized molecular signatures which can facilitate tailored interventions addressing the unique aspects of each patient. AI-driven transcriptomics, proteomics, and genomes represents a revolutionary strategy to improve patient outcome by offering precise diagnosis and tailored therapy. The inclusion of AI in oncology may boost efficiency, reduce errors, and save costs, but it cannot take the role of medical professionals. While clinicians and doctors have the final say in all matters, it might serve as their faithful assistant.

New Strategies and Artificial Intelligence Methods for the Mitigation of Toxigenic Fungi and Mycotoxins in Foods

The proliferation of toxigenic fungi in food and the subsequent production of mycotoxins constitute a significant concern in the fields of public health and consumer protection. This review highlights recent strategies and emerging methods aimed at preventing fungal growth and mycotoxin contamination in food matrices as opposed to traditional approaches such as chemical fungicides, which may leave toxic residues and pose risks to human and animal health as well as the environment. The novel methodologies discussed include the use of plant-derived compounds such as essential oils, classified as Generally Recognized as Safe (GRAS), polyphenols, lactic acid bacteria, cold plasma technologies, nanoparticles (particularly metal nanoparticles such as silver or zinc nanoparticles), magnetic materials, and ionizing radiation. Among these, essential oils, polyphenols, and lactic acid bacteria offer eco-friendly and non-toxic alternatives to conventional fungicides while demonstrating strong antimicrobial and antifungal properties; essential oils and polyphenols also possess antioxidant activity. Cold plasma and ionizing radiation enable rapid, non-thermal, and chemical-free decontamination processes. Nanoparticles and magnetic materials contribute advantages such as enhanced stability, controlled release, and ease of separation. Furthermore, this review explores recent advancements in the application of artificial intelligence, particularly machine learning methods, for the identification and classification of fungal species as well as for predicting the growth of toxigenic fungi and subsequent mycotoxin production in food products and culture media.

Laboratory measurement and machine learning-based analysis of driving factors for brake wear particle emissions from light-duty electric vehicles and heavy-duty vehicles

This study investigates brake wear particle (BWP) emissions from light-duty electric vehicles (EVs) and heavy-duty vehicles (HDVs) using a self-developed whole-vehicle testing system and a modified brake dynamometer. The results show that regenerative braking significantly reduces emissions: weak and strong regenerative braking modes reduce brake wear PM2.5 by 75 % and 87 %, and brake wear PM10 by 90 % and 95 %, respectively. HDVs with drum brakes produce lower emissions and higher PM2.5/PM10 ratios than those with disc brakes. A machine learning model (XGBoost) was developed to analyze the relationship between BWP emissions and factors (11 for light-duty EVs and 8 for HDVs, based on kinematic, vehicle, and braking parameters). SHapley Additive exPlanations (SHAP) were used for model interpretation. For light-duty EVs, reducing high kinetic energy losses (Ike > 6500 J) and initial speeds (V > 45 km/h) braking events significantly lowers emissions. Additionally, the emission reduction effect of regenerative braking intensity (BI) stabilizes when BI exceeds 900 J. For HDVs, controlling braking temperature (Avg.T < 200°C) and initial speed (V < 50 km/h) effectively reduces emissions. Our findings provide new insights into the emission characteristics and control strategies for BWPs. SYNOPSIS: The construction and interpretation of a machine learning based model of brake wear emissions provides new insights into the refined assessment and control of non-exhaust emissions.

Identification of Endoplasmic Reticulum Stress-Related Genes in Acute Myocardial Infarction: A Bioinformatics Approach with Experimental Validation

Acute myocardial infarction (AMI) continues to pose a substantial risk to human lives worldwide. Endoplasmic reticulum stress (ERS) is increasingly recognized as one of the potential mechanisms of myocardial injury following AMI. The primary goal of this study is to investigate the correlation between ERS and AMI through machine learning-based bioinformatics analysis, explore key genes, and conduct in vivo and in vitro experimental validation. We performed differential analysis and Weighted Gene Co-expression Network Analysis (WGCNA) on gene expression data from the GEO database (GSE62646). The intersection with ERS-related genes (ERSRGs) was taken to obtain AMI-ERS-related genes (MIEGs), and machine learning algorithms were further used to identify key genes (Hubs) from the MIEGs. The validation set GSE59867 was used to assess the expression levels and predictive capabilities of the Hubs for AMI. An AMI rat model was established to detect the mRNA and protein expression levels of the Hubs. The protein inhibitor of the key gene FURIN was used to treat H9C2 cells under oxygen-glucose deprivation (OGD) to explore the effects of FURIN on ERS and apoptosis. Bioinformatics analysis identified 27 MIEGs, and machine learning further determined 5 Hubs highly associated with AMI and ERS: RELA, FURIN, ERGIC3, TPP1, and BGLAP. The expression of these Hubs was significantly elevated in AMI patients within both the training and validation sets, and the area under the curve (AUC) indicated good diagnostic value. Our experiments confirmed that the mRNA levels of Furin and RelA were significantly elevated in AMI rats. Furin protein was increased in AMI rats and OGD H9C2. Furin inhibitor could alleviate OGD-induced ERS and apoptosis in H9C2. Our study demonstrates that Hubs play a pivotal role in myocardial infarction. Notably, Furin and its mediated ERS and apoptosis are significant in the pathogenesis of AMI, potentially serving as target for AMI diagnosis and treatment.

Construction and validation of an interpretable XGBoost machine learning model to predict ESBL positivity rates based on urinalysis data

BACKGROUND

Microbiological culture and drug susceptibility testing of urine samples have lengthy turnaround times, increasing the risk of extended-spectrum β-lactamase (ESBL)-positive urinary tract infection (UTI) patients progressing to sepsis.

OBJECTIVE

To develop an efficient machine learning model for the identification of ESBL positivity in UTI patients.

METHODS

This retrospective study included 528 samples that had undergone drug susceptibility testing, based on inclusion and exclusion criteria. Variables were screened using Lasso regression, with 70% of the samples used to construct nine machine learning models (XGBClassifier, LogisticRegression, LGBMClassifier, AdaBoostClassifier, SVC, MLPClassifier, ComplementNB, GaussianNB, and GradientBoostingClassifier). Model selection was based on criteria including accuracy, sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), F1 score, Kappa score, and Area Under the Curve (AUC). The best model type was identified through ten-fold cross-validation, which was then built using the remaining 30% of the data as a test set. Interpretations of predictive results were provided using the SHAP model, clarifying the impact of each feature on predictions and enhancing model transparency and interpretability.

RESULTS

The variables selected by the Lasso regression model are as follows: gender + urinary protein + urobilinogen + leukocytes + occult blood + age + pH + specific gravity + leukocyte count + erythrocyte count + epithelial cell count + cast count.The XGBoost model outperformed others in ten-fold cross-validation, with scores on the validation set as follows: AUC (95%CI): 0.924 (0.860-0.989); cutoff: 0.664(0.637-0.690); accuracy: 0.862(0.839-0.885); sensitivity: 0.9(0.879-0.920); specificity: 0.725(0.618-0.832); PPV: 0.923(0.896-0.950); NPV: 0.667(0.626-0.707); F1 score: 0.911(0.896-0.925); Kappa: 0.603(0.527-0.679). The final model achieved an AUC of 0.968 and accuracy of 0.943 on the test set.

CONCLUSION

This study developed a rapid and efficient machine learning model capable of identifying ESBL positivity based solely on routine urine test data.

Visit-to-visit blood pressure variability and clinical outcomes in peritoneal dialysis - based on machine learning algorithms

This study aims to investigate the association between visit-to-visit blood pressure variability (VVV) in early stage of continuous ambulatory peritoneal dialysis (CAPD) and long-term clinical outcomes, utilizing machine learning algorithms. Patients who initiated CAPD therapy between January 1, 2006, and December 31, 2009 were enrolled. VVV parameters were collected during the first six months of CAPD therapy. Patient follow-up extended to December 31, 2021, for up to 15.8 years. The primary outcome was the occurrence of a three-point major adverse cardiovascular event (MACE). Four machine learning algorithms and competing risk regression analysis were applied to construct predictive models. A total of 666 participants were included in the analysis with a mean age of 47.9 years. One of the six VVV parameters, standard deviation of diastolic blood pressure (SDDBP), was finally enrolled into the MACE predicting model and mortality predicting model. In the MACE predicting model, higher SDDBP was associated with 99% higher MACE risk. The association between SDDBP and MACE risk was attenuated by better residual renal function (p for interaction <0.001). In the mortality predicting model, higher SDDBP was associated with 46% higher mortality risk. This cohort study discerned that high SDDBP in early stage of CAPD indicated increased long-term MACE and mortality risks.

Machine Learning Analysis of Nutrient Associations with Peripheral Arterial Disease: Insights from NHANES 1999-2004

BACKGROUND

Peripheral arterial disease (PAD) is a common manifestation of atherosclerosis, affecting over 200 million people worldwide. The incidence of PAD is increasing due to the aging population. Common risk factors include smoking, diabetes, and hyperlipidemia, but its exact pathogenesis remains unclear. Nutritional intake is associated with the onset and progression of PAD, although relevant studies remain limited. Some studies suggest that certain nutritional elements may influence the development of PAD. This study aims to explore the relationship between nutrition and PAD using machine learning techniques. Unlike traditional statistical methods, machine learning can effectively capture complex, nonlinear relationships, providing a more comprehensive analysis of PAD risk factor.

METHODS

Data from National Health and Nutrition Examination Survey (NHANES 1999-2004) were analyzed, including demographic, clinical, and dietary information. Nutrient intake was assessed through 24-h dietary recalls using computer-assisted dietary interview system (CADI) and automated multiple pass method (AMPM) methods. PAD was defined as an ankle-brachial index (ABI) < 0.9. Six ML models-extreme gradient boosting (XGBoost), random Forest (RF), naive bayes classifier (NB), support vector machine (SVM), logistic regression (LR), and decision tree (DT)-were trained on a 70/30 train-test split, with missing data imputed and sample imbalance addressed via synthetic minority oversampling technique (SMOTE). Model performance was evaluated using the area under the receiver operating characteristic curve (AUROC), accuracy, sensitivity, specificity, precision, recall, and F1 score. Shapley additive explanations (SHAP) analysis was used to identify key features. In addition, to further enhance the interpretability of the model, we applied SHAP analysis to identify the features that have a significant impact on PAD prediction. This approach allowed us to determine the contribution of different variables to the model's output, providing deeper insights into how each feature influences the prediction of PAD outcomes.

RESULTS

Of 31,126 participants, 4,520 met the inclusion criteria (mean age 61.2 ± 13.5 years; 48.8% male), and 441 (9.7%) had ABI < 0.9. XGBoost outperformed other models, achieving an AUROC of 0.913 (95% CI, 0.891-0.936) and F1 score of 0.932. With SMOTE, its AUROC improved to 0.926 (95% CI, 0.889-0.936) and F1 score to 0.937. SHAP analysis identified vitamin C, saturated fatty acids, selenium, phosphorus, and protein intake as key predictors of PAD.

CONCLUSION

This is the first study to apply ML algorithms to examine nutrient intake and PAD in a general population. Vitamin C and phosphorus showed negative correlations with PAD, while saturated fatty acids, protein, and selenium exhibited positive associations.

Soft Modes as a Predictive Framework for Low-Dimensional Biological Systems Across Scales

All biological systems are subject to perturbations arising from thermal fluctuations, external environments, or mutations. Yet, while biological systems consist of thousands of interacting components, recent high-throughput experiments have shown that their response to perturbations is surprisingly low dimensional: confined to only a few stereotyped changes out of the many possible. In this review, we explore a unifying dynamical systems framework-soft modes-to explain and analyze low dimensionality in biology, from molecules to ecosystems. We argue that this soft mode framework makes nontrivial predictions that generalize classic ideas from developmental biology to disparate systems, namely phenocopying, dual buffering, and global epistasis. While some of these predictions have been borne out in experiments, we discuss how soft modes allow for a surprisingly far-reaching and unifying framework in which to analyze data from protein biophysics to microbial ecology.

A machine-learning approach to optimize nutritional properties and organic wastes recycling efficiency conversed by black soldier fly (Hermetia illucens)

Suboptimal nutrition in organic waste limits the growth of black soldier fly (BSF) larvae, thereby reducing biowaste recycling efficiency. In this study, weight gain data from BSF larvae fed diets with distinct nutrient compositions were used to build a machine learning model. Among the algorithms tested, the XGBoost model demonstrated the best performance in predicting weight gain. The model identified protein as the most critical nutrient factor for larval biomass and was used to determine the optimal diet by calculating the highest weight gain from 30,000 randomly generated nutrient combinations. Supplementing the missing nutrients in organic waste according to the optimal diet improved the weight gain and feed conversion rate of BSF larvae. Feeding larvae a mixture of organic wastes, a cost-effective strategy to meet dietary nutrition requirements, resulted in significant increases in both the bioconversion rate (up to 9.7%) and mass reduction rate (up to 22.8%) of organic waste.

How did we get there? AI applications to biological networks and sequences

The rapidly advancing field of artificial intelligence (AI) has transformed numerous scientific domains, including biology, where a vast and complex volume of data is available for analysis. This paper provides a comprehensive overview of the current state of AI-driven methodologies in genomics, proteomics, and systems biology. We discuss how machine learning algorithms, particularly deep learning models, have enhanced the accuracy and efficiency of embedding sequences, motif discovery, and the prediction of gene expression and protein structure. Additionally, we explore the integration of AI in the embedding and analysis of biological networks, including protein-protein interaction networks and multi-layered networks. By leveraging large-scale biological data, AI techniques have enabled unprecedented insights into complex biological processes and disease mechanisms. This work underlines the potential of applying AI to complex biological data, highlighting current applications and suggesting directions for future research to further explore AI in this rapidly evolving field.

Advances of artificial intelligence in clinical application and scientific research of neuro-oncology: Current knowledge and future perspectives

Brain tumors refer to the abnormal growths that occur within the brain's tissue, comprising both primary neoplasms and metastatic lesions. Timely detection, precise staging, suitable treatment, and standardized management are of significant clinical importance for extending the survival rates of brain tumor patients. Artificial intelligence (AI), a discipline within computer science, is leveraging its robust capacity for information identification and combination to revolutionize traditional paradigms of oncology care, offering substantial potential for precision medicine. This article provides an overview of the current applications of AI in brain tumors, encompassing the primary AI technologies, their working mechanisms and working workflow, the contributions of AI to brain tumor diagnosis and treatment, as well as the role of AI in brain tumor scientific research, particularly in drug innovation and revealing tumor microenvironment. Finally, the paper addresses the existing challenges, potential solutions, and the future application prospects. This review aims to enhance our understanding of the application of AI in brain tumors and provide valuable insights for forthcoming clinical applications and scientific inquiries.

Neural Network-Aided Milk Somatic Cell Count Increase Prediction

Subclinical mastitis (SM) is the most economically damaging yet often visually undetectable disease of dairy cows. Early detection and treatment can reduce the loss caused by the disease; thus, the continuous improvement of SM diagnostic methods is necessary. Although milk's somatic cell count (SCC) is commonly measured for diagnostic purposes, its direct determination is not widely used in everyday practice. The primary objective of our work was to investigate whether the predictive value of SM diagnostics can be improved by training artificial neural networks (ANNs) on data generated using typical conventional milking systems. The best ANN classifier had a sensitivity of 0.54 and a specificity of 0.77, which is comparable to performances of various California Mastitis Tests (CMT) found in the literature. Combining two diagnostic tests, ANN and CMT, we concluded that the positive predictive value could be up to 50% higher than the value provided by the individual CMT. While implementing CMT is a labor-intensive process on herd-level, in milking machines where milk properties or milk yield data can be measured automatically, similar to our work, SCC-increase predictions for all individuals could be obtained daily basis.

Constructing machine learning-based risk prediction model for osteoarthritis in population aged 45 and above: NHANES 2011-2018

Osteoarthritis is a widespread chronic joint disease, becoming increasingly prevalent, particularly among individuals over the age of 45. This condition causes joint pain and dysfunction, significantly disrupting daily life. The objective of this study is to develop an optimal machine learning model for predicting the risk of osteoarthritis in individuals aged 45 and older. This study utilized data from the National Health and Nutrition Examination Survey (NHANES) from 2011 to 2018, which included a total of 2980 individuals. The dataset was randomly divided into a training set (n = 2235) and a validation set (n = 745). Five machine learning algorithms were employed to develop the predictive model for osteoarthritis. The SHapley Additive exPlanation (SHAP) method was used to interpret the machine learning algorithms and identify the most significant features for predicting outcomes. The study involved 2980 participants and focused on predicting the probability of osteoarthritis occurrence using machine learning algorithms. Five algorithms were employed, analyzing 24 features from an average 60-year-old cohort, with 605 osteoarthritis diagnoses. After performing Recursive Feature Elimination (RFE) to select 20 features, the CatBoost model achieved an AUC of 0.8109 and an accuracy rate of 0.7315, making it the most efficient model. The most influential factors in the predictions were Gender, Age, BMI, Waist Circumference, and Race. This study demonstrates that the CatBoost model with 20 features can effectively predict the occurrence of osteoarthritis. This accurate prediction model can help inform early interventions and patient management strategies, potentially improving patient prognosis. Further research will focus on enhancing the model performance, such as incorporating additional relevant features or refining existing ones. Additionally, validating the model in more diverse patient populations, and investigating its potential for real-time implementation in clinical settings would further increase the study's impact and facilitate its translation into clinical practice.

Molecular characteristics of early- and late-onset ovarian cancer: insights from multidimensional evidence

BACKGROUND

Ovarian cancer (OC) is among the most lethal gynecologic malignancies, characterized by poor prognosis. While aging is a well-established risk factor, the underlying mechanisms distinguishing early- and late-onset ovarian cancer remain poorly understood.

METHODS

This study analyzed the global burden and age-related trends of ovarian cancer using the GBD database. A cut-off age of 55 years was used to differentiate between early and late onset ovarian cancer, and a Mendelian randomization method was also used to investigate the causal relationship between aging and ovarian cancer. Machine learning was applied to identify tumor-specific age-associated genes, followed by bioinformatics analyses and single-cell sequencing to explore the roles of these genes and immune profile alterations in ovarian cancer. Additionally, models were constructed, and drug sensitivity analyses performed to evaluate their potential as diagnostic markers or therapeutic targets.

RESULTS

Ovarian cancer incidence and mortality exhibit age-related trends, with telomere length positively associated with increased risk (OR = 1.27, 95% CI: 1.01-1.60, P = 3.90 × 10⁻2). Older patients with OC have a worse prognosis. PRKCD and UCP2 were significantly upregulated in ovarian cancer. PRKCD facilitates epithelial-mesenchymal transition (EMT), contributing to ovarian cancer progression, while UCP2 modulates ROS dynamics, influencing chemoresistance. Immune microenvironment analysis revealed differences between high- and low-expression groups, particularly in T cells, macrophages, and other immune cells. Both genes are sensitive to a varity of drugs, including dasatinib, fluvastatin, highlighting their potential as therapeutic targets.

CONCLUSION

Aging is a significant risk factor for ovarian cancer, with PRKCD and UCP2 closely linked to its onset and progression. These genes show promise as novel biomarkers and therapeutic targets for ovarian cancer.