Machine learning in the prediction of cancer therapy

Drug discovery and mechanism prediction with explainable graph neural networks

Apprehension of drug action mechanism is paramount for drug response prediction and precision medicine. The unprecedented development of machine learning and deep learning algorithms has expedited the drug response prediction research. However, existing methods mainly focus on forward encoding of drugs, which is to obtain an accurate prediction of the response levels, but omitted to decipher the reaction mechanism between drug molecules and genes. We propose the eXplainable Graph-based Drug response Prediction (XGDP) approach that achieves a precise drug response prediction and reveals the comprehensive mechanism of action between drugs and their targets. XGDP represents drugs with molecular graphs, which naturally preserve the structural information of molecules and a Graph Neural Network module is applied to learn the latent features of molecules. Gene expression data from cancer cell lines are incorporated and processed by a Convolutional Neural Network module. A couple of deep learning attribution algorithms are leveraged to interpret interactions between drug molecular features and genes. We demonstrate that XGDP not only enhances the prediction accuracy compared to pioneering works but is also capable of capturing the salient functional groups of drugs and interactions with significant genes of cancer cells.

Factors predicting treatment response to biological and targeted synthetic disease-modifying antirheumatic drugs in psoriatic arthritis - a systematic review and meta-analysis

The therapeutic response of patients with psoriatic arthritis (PsA) varies greatly and is often unsatisfactory. Accordingly, it is essential to individualise treatment selection to minimise long-term complications. This study aimed to identify factors that might predict treatment response to biological and targeted synthetic disease-modifying antirheumatic drugs (bDMARDs and tsDMARDs) in patients with PsA and to outline their potential application using artificial intelligence (AI). Five electronic databases were screened to identify relevant studies. A random-effects meta-analysis was performed for factors that were investigated in at least four studies. Finally, 37 studies with a total of 17,042 patients were included. The most frequently investigated predictors in these studies were sex, age, C-reactive protein (CRP), the Health Assessment Questionnaire (HAQ), BMI, and disease duration. The meta-analysis revealed that male sex (odds ratio (OR) = 2.188, 95% confidence interval (CI) = 1.912-2.503) and higher baseline CRP (1.537, 1.111-2.125) were associated with greater treatment response. Older age (0.982, 0.975-0.99), higher baseline HAQ score (0.483, 0.336-0.696), higher baseline DAPSA score (0.789, 0.663-0.938), and higher baseline tender joint count (TJC) (0.97, 0.945-0.996) were negatively correlated with the response to therapy. The other factors were not statistically significant but might be of clinical importance in the context of a complex AI test battery. Further studies are needed to validate these findings and identify novel factors that could guide personalised treatment decisions for PsA patients, in particular in developing AI applications. In accordance with the latest medical developments, decision-support tools based on supervised learning algorithms have been proposed as a clinical application of these predictors. Key messages • Given the often unsatisfactory and unpredictable therapeutic response in patients with Psoriatic Arthritis (PsA), treatment selection must be highly individualized. • A systematic literature review was conducted to identify the most reliable predictors of treatment response to biologic and targeted synthetic disease-modifying antirheumatic drugs in PsA patients. • The potential integration of these predictors into AI tools for routine clinical practice is discussed.

Identifying feature genes of chickens with different feather pecking tendencies based on three machine learning algorithms and WGCNA

Feather pecking (FP) is a significant welfare concern in poultry, which can result in reduced egg production, deterioration of feather condition, and an increase in mortality rate. This can harm the health of birds and the economic benefits of breeders. FP, as a complex trait, is regulated by multiple factors, and so far, no one has been able to elucidate its exact mechanism. In order to delve deeper into the genetic mechanism of FP, we acquired the expression matrix of dataset GSE36559. We analyzed the gene modules associated with the trait through WGCNA (Weighted correlation network analysis), and then used KEGG and GO to identify the biological pathways enriched by the modules using KEGG and GO. Subsequently, we analyzed the module with the highest correlation (0.99) using three machine learning (ML) algorithms to identify the feature genes that they collectively recognized. In this study, five feature genes, NUFIP2, ST14, OVM, GLULD1, and LOC424943, were identified. Finally, the discriminant value of the feature genes was evaluated by manipulating the receiver operating curve (ROC) in the external dataset GSE10380.

Evaluation of glucocorticoid-related genes reveals GPD1 as a therapeutic target and regulator of sphingosine 1-phosphate metabolism in CRPC

Prostate cancer (PCa) is an androgen-dependent disease, with castration-resistant prostate cancer (CRPC) being an advanced stage that no longer responds to androgen deprivation therapy (ADT). Mounting evidence suggests that glucocorticoid receptors (GR) confer resistance to ADT in CRPC patients by bypassing androgen receptor (AR) blockade. GR, as a novel therapeutic target in CRPC, has attracted substantial attention worldwide. This study utilized bioinformatic analysis of publicly available CRPC single-cell data to develop a consensus glucocorticoid-related signature (Glu-sig) that can serve as an independent predictor for relapse-free survival. Our results revealed that the signature demonstrated consistent and robust performance across seven publicly accessible datasets and an internal cohort. Furthermore, our findings demonstrated that glycerol-3-phosphate dehydrogenase 1 (GPD1) in Glu-sig can significantly promote CRPC progression by mediating the cell cycle pathway. Additionally, GPD1 was shown to be regulated by GR, with the GR antagonist mifepristone enhancing the anti-tumorigenic effects of GPD1 in CRPC cells. Mechanistically, targeting GPD1 induced the production of sphingosine 1-phosphate (S1P) and enhanced histone acetylation, thereby inducing the transcription of p21 that involved in cell cycle regulation. In conclusion, Glu-sig could serve as a robust and promising tool to improve the clinical outcomes of PCa patients, and modulating the GR/GPD1 axis that promotes tumor growth may be a promising approach for delaying CRPC progression.

Artificial Intelligence-Driven Precision Medicine: Multi-Omics and Spatial Multi-Omics Approaches in Diffuse Large B-Cell Lymphoma (DLBCL)

In this comprehensive review, we delve into the transformative role of artificial intelligence (AI) in refining the application of multi-omics and spatial multi-omics within the realm of diffuse large B-cell lymphoma (DLBCL) research. We scrutinized the current landscape of multi-omics and spatial multi-omics technologies, accentuating their combined potential with AI to provide unparalleled insights into the molecular intricacies and spatial heterogeneity inherent to DLBCL. Despite current progress, we acknowledge the hurdles that impede the full utilization of these technologies, such as the integration and sophisticated analysis of complex datasets, the necessity for standardized protocols, the reproducibility of findings, and the interpretation of their biological significance. We proceeded to pinpoint crucial research voids and advocated for a trajectory that incorporates the development of advanced AI-driven data integration and analytical frameworks. The evolution of these technologies is crucial for enhancing resolution and depth in multi-omics studies. We also emphasized the importance of amassing extensive, meticulously annotated multi-omics datasets and fostering translational research efforts to connect laboratory discoveries with clinical applications seamlessly. Our review concluded that the synergistic integration of multi-omics, spatial multi-omics, and AI holds immense promise for propelling precision medicine forward in DLBCL. By surmounting the present challenges and steering towards the outlined futuristic pathways, we can harness these potent investigative tools to decipher the molecular and spatial conundrums of DLBCL. This will pave the way for refined diagnostic precision, nuanced risk stratification, and individualized therapeutic regimens, ushering in a new era of patient-centric oncology care.

Construction of a risk prediction model for postoperative deep vein thrombosis in colorectal cancer patients based on machine learning algorithms

Background

Colorectal cancer is a prevalent malignancy of the digestive system, with an increasing incidence. Lower extremity deep vein thrombosis (DVT) is a frequent postoperative complication, occurring in up to 40% of cases.

Objective

This research aims to develop and validate a machine learning model (ML) to predict the risk of lower limb deep vein thrombosis in patients with colorectal cancer, facilitating preventive and therapeutic measures to enhance recovery and ensure safety.

Methods

In this retrospective cohort study, we collected data from 429 colorectal cancer patients from January 2021 to January 2024. The medical records included age, blood test results, body mass index, underlying diseases, clinical staging, histological typing, surgical methods, and postoperative complications. We employed the Synthetic Minority Oversampling Technique to address imbalanced data and split the dataset into training and validation sets in a 7:3 ratio. Feature selection was performed using Random Forest (RF), XGBoost, and Least Absolute Shrinkage and Selection Operator algorithms (LASSO). We then trained six machine learning models: Logistic Regression (LR), Naive Bayes (NB), Gaussian Process (GP), Random Forest, XGBoost, and Multilayer Perceptron (MLP). The model's performance was evaluated using metrics such as area under the Receiver Operating Characteristic curve, accuracy, sensitivity, specificity, F1 score, and confusion matrix. Additionally, SHAP and LIME were used to enhance the interpretability of the results.

Results

The study combined Random Forest, XGBoost algorithms, and LASSO regression with univariate regression analysis to identify significant predictive factors, including age, preoperative prealbumin, preoperative albumin, preoperative hemoglobin, operation time, PIKVA2, CEA, and preoperative neutrophil count. The XGBoost model outperformed other ML algorithms, achieving an AUC of 0.996, an accuracy of 0.9636, a specificity of 0.9778, and an F1 score of 0.9576. Moreover, the SHAP method identified age and preoperative prealbumin as the primary determinants influencing ML model predictions. Finally, the study employed LIME for more precise prediction and interpretation of individual predictions.

Conclusion

The machine learning algorithms effectively predicted postoperative lower limb deep vein thrombosis in colorectal cancer patients. The XGBoost model demonstrated strong potential for improving early detection and treatment in clinical settings.

Radial Data Visualization-Based Step-by-Step Eliminative Algorithm to Predict Colorectal Cancer Patients' Response to FOLFOX Therapy

Application of the FOLFOX scheme to colorectal cancer (CRC) patients often results in the development of chemo-resistance, leading to therapy failure. This study aimed to develop a functional and easy-to-use algorithm to predict patients' response to FOLFOX treatment. Transcriptomic data of CRC patient's samples treated with FOLFOX were downloaded from the Gene Expression Omnibus database (GSE83129, GSE28702, GSE69657, GSE19860 and GSE41568). Comparing the expression of top up- and downregulated genes in FOLFOX responder and non-responder patients' groups, we selected 30 potential markers that were used to create a step-by-step eliminative procedure based on modified radial data visualization, which depicts the interplay between the expression level of chosen attributes (genes) to locate data points in low-dimensional space. Our analysis proved that FOLFOX-resistant CRC samples are predominantly characterized by upregulated expression levels of TMEM182 and MCM9 and downregulated LRRFIP1. Additionally, the procedure developed based on expression levels of TMEM182, MCM9, LRRFIP1, LAMP1, FAM161A, KLHL36, ETV5, RNF168, SRSF11, NCKAP5, CRTAP, VAMP2, ZBTB49 and RIMBP2 proved to be capable in predicting FOLFOX therapy response. In conclusion, our approach can give a unique insight into clinical decision-making regarding therapy scheme administration, potentially increasing patients' survival and, consequently, medical futility due to incorrect therapy application.

A Combinatorial Functional Precision Medicine Platform for Rapid Therapeutic Response Prediction in AML

BACKGROUND

Despite advances made in targeted biomarker-based therapy for acute myeloid leukemia (AML) treatment, remission is often short and followed by relapse and acquired resistance. Functional precision medicine (FPM) efforts have been shown to improve therapy selection guidance by incorporating comprehensive biological data to tailor individual treatment. However, effectively managing complex biological data, while also ensuring rapid conversion of actionable insights into clinical utility remains challenging.

METHODS

We have evaluated the clinical applicability of quadratic phenotypic optimization platform (QPOP), to predict clinical response to combination therapies in AML and reveal patient-centric insights into combination therapy sensitivities. In this prospective study, 51 primary samples from newly diagnosed (ND) or refractory/relapsed (R/R) AML patients were evaluated by QPOP following ex vivo drug testing.

RESULTS

Individualized drug sensitivity reports were generated in 55/63 (87.3%) patient samples with a median turnaround time of 5 (4-10) days from sample collection to report generation. To evaluate clinical feasibility, QPOP-predicted response was compared to clinical treatment outcomes and indicated concordant results with 83.3% sensitivity and 90.9% specificity and an overall 86.2% accuracy. Serial QPOP analysis in a FLT3-mutant patient sample indicated decreased FLT3 inhibitor (FLT3i) sensitivity, which is concordant with increasing FLT3 allelic burden and drug resistance development. Forkhead box M1 (FOXM1)-AKT signaling was subsequently identified to contribute to resistance to FLT3i.

CONCLUSION

Overall, this study demonstrates the feasibility of applying QPOP as a functional combinatorial precision medicine platform to predict therapeutic sensitivities in AML and provides the basis for prospective clinical trials evaluating ex vivo-guided combination therapy.

Integrating machine learning and single-cell transcriptomic analysis to identify potential biomarkers and analyze immune features of ischemic stroke

This study employs machine learning and single-cell transcriptome sequencing (scRNA-seq) analysis to unearth novel biomarkers and delineate the immune characteristics of ischemic stroke (IS), thereby contributing fresh insights into IS treatment strategies.Our research leverages gene expression data sourced from the GEO database. We undertake weighted gene co-expression network analysis (WGCNA) to filter pertinent genes and subsequently employ machine learning algorithms for the identification of feature genes. Concurrently, we rigorously execute quality control measures, dimensionality reduction techniques, and cell annotation on the scRNA-seq data to pinpoint differentially expressed genes (DEGs). The identification of core genes, denoted as Hub genes, among the feature genes and DEGs, is achieved through meticulous overlapping analysis. We illuminate the immune characteristics of these Hub genes using a suite of analytical tools, encompassing CIBERSORT, MCPcounter, and pseudotemporal analysis, all based on immune cell annotations and single-cell transcriptome data.Subsequently, we harness the CMap database to prognosticate potential therapeutic drugs and scrutinize their associations with the identified Hub genes. Our findings unveil robust linkages between three pivotal Hub genes-namely, RNF13, VASP, and CD163-and specific immune cell types such as T cells and neutrophils. These Hub genes predominantly manifest in macrophages and microglial cells within the scRNA-seq immune cell population, exhibiting variances across different stages of cellular differentiation. In conclusion, this study unearths highly pertinent biomarkers for IS diagnosis and elucidates IS-induced immune infiltration characteristics, thus providing a firm foundation for a comprehensive exploration of potential immune mechanisms and the identification of novel therapeutic targets for IS.

Model ensembling as a tool to form interpretable multi-omic predictors of cancer pharmacosensitivity

Stratification of patients diagnosed with cancer has become a major goal in personalized oncology. One important aspect is the accurate prediction of the response to various drugs. It is expected that the molecular characteristics of the cancer cells contain enough information to retrieve specific signatures, allowing for accurate predictions based solely on these multi-omic data. Ideally, these predictions should be explainable to clinicians, in order to be integrated in the patients care. We propose a machine-learning framework based on ensemble learning to integrate multi-omic data and predict sensitivity to an array of commonly used and experimental compounds, including chemotoxic compounds and targeted kinase inhibitors. We trained a set of classifiers on the different parts of our dataset to produce omic-specific signatures, then trained a random forest classifier on these signatures to predict drug responsiveness. We used the Cancer Cell Line Encyclopedia dataset, comprising multi-omic and drug sensitivity measurements for hundreds of cell lines, to build the predictive models, and validated the results using nested cross-validation. Our results show good performance for several compounds (Area under the Receiver-Operating Curve >79%) across the most frequent cancer types. Furthermore, the simplicity of our approach allows to examine which omic layers have a greater importance in the models and identify new putative markers of drug responsiveness. We propose several models based on small subsets of transcriptional markers with the potential to become useful tools in personalized oncology, paving the way for clinicians to use the molecular characteristics of the tumors to predict sensitivity to therapeutic compounds.

The Use of Artificial Intelligence in Predicting Chemotherapy-Induced Toxicities in Metastatic Colorectal Cancer: A Data-Driven Approach for Personalized Oncology

BACKGROUND

Machine learning models learn about general behavior from data by finding the relationships between features. Our purpose was to develop a predictive model to identify and predict which subset of colorectal cancer patients are more likely to experience chemotherapy-induced toxicity and to determine the specific attributes that influence the presence of treatment-related side effects.

METHODS

The predictor was general toxicity, and for the construction of our data training, we selected 95 characteristics that represent the health state of 74 patients prior to their first round of chemotherapy. After the data were processed, Random Forest models were trained to offer an optimal balance between accuracy and interpretability.

RESULTS

We constructed a machine learning predictor with an emphasis on assessing the importance of numerical and categorical variables in relation to toxicity.

CONCLUSIONS

The incorporation of artificial intelligence in personalizing colorectal cancer management by anticipating and overseeing toxicities more effectively illustrates a pivotal shift towards more personalized and precise medical care.

Cancer pharmacoinformatics: Databases and analytical tools

Cancer is a subject of extensive investigation, and the utilization of omics technology has resulted in the generation of substantial volumes of big data in cancer research. Numerous databases are being developed to manage and organize this data effectively. These databases encompass various domains such as genomics, transcriptomics, proteomics, metabolomics, immunology, and drug discovery. The application of computational tools into various core components of pharmaceutical sciences constitutes "Pharmacoinformatics", an emerging paradigm in rational drug discovery. The three major features of pharmacoinformatics include (i) Structure modelling of putative drugs and targets, (ii) Compilation of databases and analysis using statistical approaches, and (iii) Employing artificial intelligence/machine learning algorithms for the discovery of novel therapeutic molecules. The development, updating, and analysis of databases using statistical approaches play a pivotal role in pharmacoinformatics. Multiple software tools are associated with oncoinformatics research. This review catalogs the databases and computational tools related to cancer drug discovery and highlights their potential implications in the pharmacoinformatics of cancer.

Identification and validation of Rab GTPases RAB13 as biomarkers for peritoneal metastasis and immune cell infiltration in colorectal cancer patients

Background

As one of the most common cancer, colorectal cancer (CRC) is with high morbidity and mortality. Peritoneal metastasis (PM) is a fatal state of CRC, and few patients may benefit from traditional therapies. There is a complex interaction between PM and immune cell infiltration. Therefore, we aimed to determine biomarkers associated with colorectal cancer peritoneal metastasis (CRCPM) and their relationship with immune cell infiltration.

Methods

By informatic analysis, differently expressed genes (DEGs) were selected and hub genes were screened out. RAB13, one of the hub genes, was identificated from public databases and validated in CRC tissues. The ESTIMATE, CEBERSORT and TIMER algorithms were applied to analyze the correlation between RAB13 and immune infiltration in CRC. RAB13's expression in different cells were analyzed at the single-cell level in scRNA-Seq. The Gene Set Enrichment Analysis (GSEA) was performed for RAB13 enrichment and further confirmed. Using oncoPredict algorithm, RAB13's impact on drug sensitivity was evaluated.

Results

High RAB13 expression was identified in public databases and led to a poor prognosis. RAB13 was found to be positively correlated with the macrophages and other immune cells infiltration and from scRNA-Seq, RAB13 was found to be located in CRC cells and macrophages. GSEA revealed that high RAB13 expression enriched in a various of biological signaling, and oncoPredict algorithm showed that RAB13 expression was correlated with paclitaxel sensitivity.

Conclusion

Our study indicated clinical role of RAB13 in CRC-PM, suggesting its potential as a therapeutic target in the future.

Modeling and predicting meat yield and growth performance using morphological features of narrow-clawed crayfish with machine learning techniques

In recent studies, artificial intelligence and machine learning methods give higher accuracy than other prediction methods in large data sets with complex structures. Instead of statistical methods, artificial intelligence, and machine learning are used due to the difficulty of constructing mathematical models in multi-parameter and multivariate problems. In this study, predictions of length-weight relationships and meat productivity were generated by machine learning models using measurement data of male and female crayfish in the narrow-clawed crayfish population living in Apolyont Lake. The data set was created using the growth performance and morphometric characters from 1416 crayfish in different years to determine the length-weight relationship and length-meat yield. Statistical methods, artificial intelligence, and machine learning are used due to the difficulty of constructing mathematical models in multi-parameter and multivariate problems. The analysis results show that most models designed as an alternative to traditional estimation methods in future planning studies in sustainable fisheries, aquaculture, and natural sources management are valid for machine learning and artificial intelligence. Seven different machine learning algorithms were applied to the data set and the length-weight relationships and length-meat yields were evaluated for both male and female individuals. Support vector regression (SVR) has achieved the best prediction performance accuracy with 0.996 and 0.992 values for the length-weight of males and females, with 0.996 and 0.995 values for the length-meat yield of males and females. The results showed that the SVR outperforms the others for all scenarios regarding the accuracy, sensitivity, and specificity metrics.

Evaluating accuracy and fairness of clinical decision support algorithms when health care resources are limited

OBJECTIVE

Guidance on how to evaluate accuracy and algorithmic fairness across subgroups is missing for clinical models that flag patients for an intervention but when health care resources to administer that intervention are limited. We aimed to propose a framework of metrics that would fit this specific use case.

METHODS

We evaluated the following metrics and applied them to a Veterans Health Administration clinical model that flags patients for intervention who are at risk of overdose or a suicidal event among outpatients who were prescribed opioids (N = 405,817): Receiver - Operating Characteristic and area under the curve, precision - recall curve, calibration - reliability curve, false positive rate, false negative rate, and false omission rate. In addition, we developed a new approach to visualize false positives and false negatives that we named 'per true positive bars.' We demonstrate the utility of these metrics to our use case for three cohorts of patients at the highest risk (top 0.5 %, 1.0 %, and 5.0 %) by evaluating algorithmic fairness across the following age groups: <=30, 31-50, 51-65, and >65 years old.

RESULTS

Metrics that allowed us to assess group differences more clearly were the false positive rate, false negative rate, false omission rate, and the new 'per true positive bars'. Metrics with limited utility to our use case were the Receiver - Operating Characteristic and area under the curve, the calibration - reliability curve, and the precision - recall curve.

CONCLUSION

There is no "one size fits all" approach to model performance monitoring and bias analysis. Our work informs future researchers and clinicians who seek to evaluate accuracy and fairness of predictive models that identify patients to intervene on in the context of limited health care resources. In terms of ease of interpretation and utility for our use case, the new 'per true positive bars' may be the most intuitive to a range of stakeholders and facilitates choosing a threshold that allows weighing false positives against false negatives, which is especially important when predicting severe adverse events.

DeepDRA: Drug repurposing using multi-omics data integration with autoencoders

Cancer treatment has become one of the biggest challenges in the world today. Different treatments are used against cancer; drug-based treatments have shown better results. On the other hand, designing new drugs for cancer is costly and time-consuming. Some computational methods, such as machine learning and deep learning, have been suggested to solve these challenges using drug repurposing. Despite the promise of classical machine-learning methods in repurposing cancer drugs and predicting responses, deep-learning methods performed better. This study aims to develop a deep-learning model that predicts cancer drug response based on multi-omics data, drug descriptors, and drug fingerprints and facilitates the repurposing of drugs based on those responses. To reduce multi-omics data's dimensionality, we use autoencoders. As a multi-task learning model, autoencoders are connected to MLPs. We extensively tested our model using three primary datasets: GDSC, CTRP, and CCLE to determine its efficacy. In multiple experiments, our model consistently outperforms existing state-of-the-art methods. Compared to state-of-the-art models, our model achieves an impressive AUPRC of 0.99. Furthermore, in a cross-dataset evaluation, where the model is trained on GDSC and tested on CCLE, it surpasses the performance of three previous works, achieving an AUPRC of 0.72. In conclusion, we presented a deep learning model that outperforms the current state-of-the-art regarding generalization. Using this model, we could assess drug responses and explore drug repurposing, leading to the discovery of novel cancer drugs. Our study highlights the potential for advanced deep learning to advance cancer therapeutic precision.

A deep learning model of tumor cell architecture elucidates response and resistance to CDK4/6 inhibitors

Cyclin-dependent kinase 4 and 6 inhibitors (CDK4/6is) have revolutionized breast cancer therapy. However, <50% of patients have an objective response, and nearly all patients develop resistance during therapy. To elucidate the underlying mechanisms, we constructed an interpretable deep learning model of the response to palbociclib, a CDK4/6i, based on a reference map of multiprotein assemblies in cancer. The model identifies eight core assemblies that integrate rare and common alterations across 90 genes to stratify palbociclib-sensitive versus palbociclib-resistant cell lines. Predictions translate to patients and patient-derived xenografts, whereas single-gene biomarkers do not. Most predictive assemblies can be shown by CRISPR-Cas9 genetic disruption to regulate the CDK4/6i response. Validated assemblies relate to cell-cycle control, growth factor signaling and a histone regulatory complex that we show promotes S-phase entry through the activation of the histone modifiers KAT6A and TBL1XR1 and the transcription factor RUNX1. This study enables an integrated assessment of how a tumor's genetic profile modulates CDK4/6i resistance.

Reliable anti-cancer drug sensitivity prediction and prioritization

The application of machine learning (ML) to solve real-world problems does not only bear great potential but also high risk. One fundamental challenge in risk mitigation is to ensure the reliability of the ML predictions, i.e., the model error should be minimized, and the prediction uncertainty should be estimated. Especially for medical applications, the importance of reliable predictions can not be understated. Here, we address this challenge for anti-cancer drug sensitivity prediction and prioritization. To this end, we present a novel drug sensitivity prediction and prioritization approach guaranteeing user-specified certainty levels. The developed conformal prediction approach is applicable to classification, regression, and simultaneous regression and classification. Additionally, we propose a novel drug sensitivity measure that is based on clinically relevant drug concentrations and enables a straightforward prioritization of drugs for a given cancer sample.

Explainable cancer factors discovery: Shapley additive explanation for machine learning models demonstrates the best practices in the case of pancreatic cancer

Pancreatic cancer is one of digestive tract cancers with high mortality rate. Despite the wide range of available treatments and improvements in surgery, chemotherapy, and radiation therapy, the five-year prognosis for individuals diagnosed pancreatic cancer remains poor. There is still research to be done to see if immunotherapy may be used to treat pancreatic cancer. The goals of our research were to comprehend the tumor microenvironment of pancreatic cancer, found a useful biomarker to assess the prognosis of patients, and investigated its biological relevance. In this paper, machine learning methods such as random forest were fused with weighted gene co-expression networks for screening hub immune-related genes (hub-IRGs). LASSO regression model was used to further work. Thus, we got eight hub-IRGs. Based on hub-IRGs, we created a prognosis risk prediction model for PAAD that can stratify accurately and produce a prognostic risk score (IRG_Score) for each patient. In the raw data set and the validation data set, the five-year area under the curve (AUC) for this model was 0.9 and 0.7, respectively. And shapley additive explanation (SHAP) portrayed the importance of prognostic risk prediction influencing factors from a machine learning perspective to obtain the most influential certain gene (or clinical factor). The five most important factors were TRIM67, CORT, PSPN, SCAMP5, RFXAP, all of which are genes. In summary, the eight hub-IRGs had accurate risk prediction performance and biological significance, which was validated in other cancers. The result of SHAP helped to understand the molecular mechanism of pancreatic cancer.

Overcoming limitations in current measures of drug response may enable AI-driven precision oncology

Machine learning (ML) models of drug sensitivity prediction are becoming increasingly popular in precision oncology. Here, we identify a fundamental limitation in standard measures of drug sensitivity that hinders the development of personalized prediction models - they focus on absolute effects but do not capture relative differences between cancer subtypes. Our work suggests that using z-scored drug response measures mitigates these limitations and leads to meaningful predictions, opening the door for sophisticated ML precision oncology models.

Development and validation of a machine learning-based postoperative prognostic model for plasma cell neoplasia with spinal lesions as initial clinical manifestations: a single-center cohort study

BACKGROUND

Spinal multiple myeloma (MM) and solitary plasmacytoma of bone (SPB), both plasma cell neoplasms, greatly affect patients' quality of life due to spinal involvement. Accurate prediction of surgical outcomes is crucial for personalized patient care, but systematic treatment guidelines and predictive models are lacking.

OBJECTIVE

This study aimed to develop and validate a machine learning (ML)-based model to predict postoperative outcomes and identify prognostic factors for patients with spinal MM and SPB.

METHODS

A retrospective analysis was conducted on patients diagnosed with MM or SPB from 2011 to 2015, followed by prospective data collection from 2016 to 2017. Patient demographics, tumor characteristics, clinical treatments, and laboratory results were analyzed as input features. Four types of ML algorithms were employed for model development. The performance was assessed using discrimination and calibration measures, and the Shapley Additive exPlanations (SHAP) method was applied for model interpretation.

RESULTS

A total of 169 patients were included, with 119 for model training and 50 for validation. The Gaussian Naïve Bayes (GNB) model exhibited superior predictive accuracy and stability. Prospective validation on the 50 patients revealed an area under the curve (AUC) of 0.863, effectively distinguishing between 5-year survivors and non-survivors. Key prognostic factors identified included International Staging System (ISS) stage, Durie-Salmon (DS) stage, targeted therapy, and age.

CONCLUSIONS

The GNB model has the best performance and high reliability in predicting postoperative outcomes. Variables such as ISS stage and DS stage were significant in influencing patient prognosis. This study enhances the ability to identify patients at risk of poor outcomes, thereby aiding clinical decision-making.

Bioinformatics analysis identifies a key gene HLA_DPA1 in severe influenza-associated immune infiltration

BACKGROUND

Severe influenza is a serious global health issue that leads to prolonged hospitalization and mortality on a significant scale. The pathogenesis of this infectious disease is poorly understood. Therefore, this study aimed to identify the key genes associated with severe influenza patients necessitating invasive mechanical ventilation.

METHODS

The current study utilized two publicly accessible gene expression profiles (GSE111368 and GSE21802) from the Gene Expression Omnibus database. The research focused on identifying the genes exhibiting differential expression between severe and non-severe influenza patients. We employed three machine learning algorithms, namely the Least Absolute Shrinkage and Selection Operator regression model, Random Forest, and Support Vector Machine-Recursive Feature Elimination, to detect potential key genes. The key gene was further selected based on the diagnostic performance of the target genes substantiated in the dataset GSE101702. A single-sample gene set enrichment analysis algorithm was applied to evaluate the participation of immune cell infiltration and their associations with key genes.

RESULTS

A total of 44 differentially expressed genes were recognized; among them, we focused on 10 common genes, namely PCOLCE2, HLA_DPA1, LOC653061, TDRD9, MPO, HLA_DQA1, MAOA, S100P, RAP1GAP, and CA1. To ensure the robustness of our findings, we employed overlapping LASSO regression, Random Forest, and SVM-RFE algorithms. By utilizing these algorithms, we were able to pinpoint the aforementioned 10 genes as potential biomarkers for distinguishing between both cases of influenza (severe and non-severe). However, the gene HLA_DPA1 has been recognized as a crucial factor in the pathological condition of severe influenza. Notably, the validation dataset revealed that this gene exhibited the highest area under the receiver operating characteristic curve, with a value of 0.891. The use of single-sample gene set enrichment analysis has provided valuable insights into the immune responses of patients afflicted with severe influenza that have further revealed a categorical correlation between the expression of HLA_DPA1 and lymphocytes.

CONCLUSION

The findings indicated that the HLA_DPA1 gene may play a crucial role in the immune-pathological condition of severe influenza and could serve as a promising therapeutic target for patients infected with severe influenza.

Modeling 5-FU-Induced Chemotherapy Selection of a Drug-Resistant Cancer Stem Cell Subpopulation

(1) Background: Cancer stem cells (CSCs) are a subpopulation of cells in a tumor that can self-regenerate and produce different types of cells with the ability to initiate tumor growth and dissemination. Chemotherapy resistance, caused by numerous mechanisms by which tumor tissue manages to overcome the effects of drugs, remains the main problem in cancer treatment. The identification of markers on the cell surface specific to CSCs is important for understanding this phenomenon. (2) Methods: The expression of markers CD24, CD44, ALDH1, and ABCG2 was analyzed on the surface of CSCs in two cancer cell lines, MDA-MB-231 and HCT-116, after treatment with 5-fluorouracil (5-FU) using flow cytometry analysis. A machine learning model (ML)-genetic algorithm (GA) was used for the in silico simulation of drug resistance. (3) Results: As evaluated through the use of flow cytometry, the percentage of CD24-CD44+ MDA-MB-231 and CD44, ALDH1 and ABCG2 HCT-116 in a group treated with 5-FU was significantly increased compared to untreated cells. The CSC population was enriched after treatment with chemotherapy, suggesting that these cells have enhanced drug resistance mechanisms. (4) Conclusions: Each individual GA prediction model achieved high accuracy in estimating the expression rate of CSC markers on cancer cells treated with 5-FU. Artificial intelligence can be used as a powerful tool for predicting drug resistance.

SynerGNet: A Graph Neural Network Model to Predict Anticancer Drug Synergy

Drug combination therapy shows promise in cancer treatment by addressing drug resistance, reducing toxicity, and enhancing therapeutic efficacy. However, the intricate and dynamic nature of biological systems makes identifying potential synergistic drugs a costly and time-consuming endeavor. To facilitate the development of combination therapy, techniques employing artificial intelligence have emerged as a transformative solution, providing a sophisticated avenue for advancing existing therapeutic approaches. In this study, we developed SynerGNet, a graph neural network model designed to accurately predict the synergistic effect of drug pairs against cancer cell lines. SynerGNet utilizes cancer-specific featured graphs created by integrating heterogeneous biological features into the human protein-protein interaction network, followed by a reduction process to enhance topological diversity. Leveraging synergy data provided by AZ-DREAM Challenges, the model yields a balanced accuracy of 0.68, significantly outperforming traditional machine learning. Encouragingly, augmenting the training data with carefully constructed synthetic instances improved the balanced accuracy of SynerGNet to 0.73. Finally, the results of an independent validation conducted against DrugCombDB demonstrated that it exhibits a strong performance when applied to unseen data. SynerGNet shows a great potential in detecting drug synergy, positioning itself as a valuable tool that could contribute to the advancement of combination therapy for cancer treatment.

Machine Learning-Assisted Optimization of Drug Combinations in Zeolite-Based Delivery Systems for Melanoma Therapy

Two independent artificial neural network (ANN) models were used to determine the optimal drug combination of zeolite-based delivery systems (ZDS) for cancer therapy. The systems were based on the NaY zeolite using silver (Ag+) and 5-fluorouracil (5-FU) as antimicrobial and antineoplastic agents. Different ZDS samples were prepared, and their characterization indicates the successful incorporation of both pharmacologically active species without any relevant changes to the zeolite structure. Silver acts as a counterion of the negative framework, and 5-FU retains its molecular integrity. The data from the A375 cell viability assays, involving ZDS samples (solid phase), 5-FU, and Ag+ aqueous solutions (liquid phase), were used to train two independent machine learning (ML) models. Both models exhibited a high level of accuracy in predicting the experimental cell viability results, allowing the development of a novel protocol for virtual cell viability assays. The findings suggest that the incorporation of both Ag and 5-FU into the zeolite structure significantly potentiates their anticancer activity when compared to that of the liquid phase. Additionally, two optimal AgY/5-FU@Y ratios were proposed to achieve the best cell viability outcomes. The ZDS also exhibited significant efficacy against Escherichia coli (E. coli) and Staphylococcus aureus (S. aureus); the predicted combination ratio is also effective against S. aureus, underscoring the potential of this approach as a therapeutic option for cancer-associated bacterial infections.

Machine Learning as a Tool for Early Detection: A Focus on Late-Stage Colorectal Cancer across Socioeconomic Spectrums

PURPOSE

To assess the efficacy of various machine learning (ML) algorithms in predicting late-stage colorectal cancer (CRC) diagnoses against the backdrop of socio-economic and regional healthcare disparities.

METHODS

An innovative theoretical framework was developed to integrate individual- and census tract-level social determinants of health (SDOH) with sociodemographic factors. A comparative analysis of the ML models was conducted using key performance metrics such as AUC-ROC to evaluate their predictive accuracy. Spatio-temporal analysis was used to identify disparities in late-stage CRC diagnosis probabilities.

RESULTS

Gradient boosting emerged as the superior model, with the top predictors for late-stage CRC diagnosis being anatomic site, year of diagnosis, age, proximity to superfund sites, and primary payer. Spatio-temporal clusters highlighted geographic areas with a statistically significant high probability of late-stage diagnoses, emphasizing the need for targeted healthcare interventions.

CONCLUSIONS

This research underlines the potential of ML in enhancing the prognostic predictions in oncology, particularly in CRC. The gradient boosting model, with its robust performance, holds promise for deployment in healthcare systems to aid early detection and formulate localized cancer prevention strategies. The study's methodology demonstrates a significant step toward utilizing AI in public health to mitigate disparities and improve cancer care outcomes.

Augmented drug combination dataset to improve the performance of machine learning models predicting synergistic anticancer effects

Combination therapy has gained popularity in cancer treatment as it enhances the treatment efficacy and overcomes drug resistance. Although machine learning (ML) techniques have become an indispensable tool for discovering new drug combinations, the data on drug combination therapy currently available may be insufficient to build high-precision models. We developed a data augmentation protocol to unbiasedly scale up the existing anti-cancer drug synergy dataset. Using a new drug similarity metric, we augmented the synergy data by substituting a compound in a drug combination instance with another molecule that exhibits highly similar pharmacological effects. Using this protocol, we were able to upscale the AZ-DREAM Challenges dataset from 8798 to 6,016,697 drug combinations. Comprehensive performance evaluations show that ML models trained on the augmented data consistently achieve higher accuracy than those trained solely on the original dataset. Our data augmentation protocol provides a systematic and unbiased approach to generating more diverse and larger-scale drug combination datasets, enabling the development of more precise and effective ML models. The protocol presented in this study could serve as a foundation for future research aimed at discovering novel and effective drug combinations for cancer treatment.

AlphaML: A clear, legible, explainable, transparent, and elucidative binary classification platform for tabular data

Leveraging the potential of machine learning and recognizing the broad applications of binary classification, it becomes essential to develop platforms that are not only powerful but also transparent, interpretable, and user friendly. We introduce alphaML, a user-friendly platform that provides clear, legible, explainable, transparent, and elucidative (CLETE) binary classification models with comprehensive customization options. AlphaML offers feature selection, hyperparameter search, sampling, and normalization methods, along with 15 machine learning algorithms with global and local interpretation. We have integrated a custom metric for hyperparameter search that considers both training and validation scores, safeguarding against under- or overfitting. Additionally, we employ the NegLog2RMSL scoring method, which uses both training and test scores for a thorough model evaluation. The platform has been tested using datasets from multiple domains and offers a graphical interface, removing the need for programming expertise. Consequently, alphaML exhibits versatility, demonstrating promising applicability across a broad spectrum of tabular data configurations.

Preclinical Anticipation of On- and Off-Target Resistance Mechanisms to Anti-Cancer Drugs: A Systematic Review

The advent of targeted therapies has led to tremendous improvements in treatment options and their outcomes in the field of oncology. Yet, many cancers outsmart precision drugs by developing on-target or off-target resistance mechanisms. Gaining the ability to resist treatment is the rule rather than the exception in tumors, and it remains a major healthcare challenge to achieve long-lasting remission in most cancer patients. Here, we discuss emerging strategies that take advantage of innovative high-throughput screening technologies to anticipate on- and off-target resistance mechanisms before they occur in treated cancer patients. We divide the methods into non-systematic approaches, such as random mutagenesis or long-term drug treatment, and systematic approaches, relying on the clustered regularly interspaced short palindromic repeats (CRISPR) system, saturated mutagenesis, or computational methods. All these new developments, especially genome-wide CRISPR-based screening platforms, have significantly accelerated the processes for identification of the mechanisms responsible for cancer drug resistance and opened up new avenues for future treatments.

Feasibility of machine learning-based modeling and prediction using multiple centers data to assess intrahepatic cholangiocarcinoma outcomes

BACKGROUND AND AIMS

Currently, there are still no definitive consensus in the treatment of intrahepatic cholangiocarcinoma (iCCA). This study aimed to build a clinical decision support tool based on machine learning using the Surveillance, Epidemiology, and End Results (SEER) database and the data from the Fifth Medical Center of the PLA General Hospital in China.

METHODS

4,398 eligible patients from the SEER database and 504 eligible patients from the hospital data, who presented with histologically proven iCCA, were enrolled for modeling by cross-validation based on machine learning. All the models were trained using the open-source Python library scikit-survival version 0.16.0. Shapley additive explanations method was used to help clinicians better understand the obtained results. Permutation importance was calculated using library ELI5.

RESULTS

All involved treatment modalities could contribute to a better prognosis. Three models were derived and tested using different data sources, with concordance indices of 0.67, 0.69, and 0.73, respectively. The prediction results were consistent with those under actual situations involving randomly selected patients. Model 2, trained using the hospital data, was selected to develop an online tool, due to its advantage in predicting short-term prognosis.

CONCLUSION

The prediction model and tool established in this study can be applied to predict the prognosis of iCCA after treatment by inputting the patient's clinical parameters or TNM stages and treatment options, thus contributing to optimal clinical decisions.KEY MESSAGESA prognostic model related to disease staging and treatment mode was conducted using the method of machine learning, based on the big data of multi centers.The online calculator can predict the short-term survival prognosis of intrahepatic cholangiocarcinoma, thus, help to make the best clinical decision.The online calculator built to calculate the mortality risk and overall survival can be easily obtained and applied.

A hybrid machine learning feature selection model-HMLFSM to enhance gene classification applied to multiple colon cancers dataset

Colon cancer is a significant global health problem, and early detection is critical for improving survival rates. Traditional detection methods, such as colonoscopies, can be invasive and uncomfortable for patients. Machine Learning (ML) algorithms have emerged as a promising approach for non-invasive colon cancer classification using genetic data or patient demographics and medical history. One approach is to use ML to analyse genetic data, or patient demographics and medical history, to predict the likelihood of colon cancer. However, due to the challenges imposed by variable gene expression and the high dimensionality of cancer-related datasets, traditional transductive ML applications have limited accuracy and risk overfitting. In this paper, we propose a new hybrid feature selection model called HMLFSM-Hybrid Machine Learning Feature Selection Model to improve colon cancer gene classification. We developed a multifilter hybrid model including a two-phase feature selection approach, combining Information Gain (IG) and Genetic Algorithms (GA), and minimum Redundancy Maximum Relevance (mRMR) coupling with Particle Swarm Optimization (PSO). We critically tested our model on three colon cancer genetic datasets and found that the new framework outperformed other models with significant accuracy improvements (95%, ~97%, and ~94% accuracies for datasets 1, 2, and 3 respectively). The results show that our approach improves the classification accuracy of colon cancer detection by highlighting important and relevant genes, eliminating irrelevant ones, and revealing the genes that have a direct influence on the classification process. For colon cancer gene analysis, and along with our experiments and literature review, we found that selective input feature extraction prior to feature selection is essential for improving predictive performance.

Predicting line of therapy transition via similar patient augmentation

Analyzing large EHR databases to predict cancer progression and treatments has become a hot trend in recent years. An increasing number of modern deep learning models have been proposed to find the milestones of essential patient medical journey characteristics to predict their disease status and give healthcare professionals valuable insights. However, most of the existing methods are lack of consideration for the inter-relationship among different patients. We believe that more valuable information can be extracted, especially when patients with similar disease statuses visit the same doctors. Towards this end, a similar patient augmentation-based approach named SimPA is proposed to enhance the learning of patient representations and further predict lines of therapy transition. Our experiment results on a real-world multiple myeloma dataset show that our proposed approach outperforms state-of-the-art baseline approaches in terms of standard evaluation metrics for classification tasks.

Augmented drug combination dataset to improve the performance of machine learning models predicting synergistic anticancer effects

Combination therapy has gained popularity in cancer treatment as it enhances the treatment efficacy and overcomes drug resistance. Although machine learning (ML) techniques have become an indispensable tool for discovering new drug combinations, the data on drug combination therapy currently available may be insufficient to build high-precision models. We developed a data augmentation protocol to unbiasedly scale up the existing anti-cancer drug synergy dataset. Using a new drug similarity metric, we augmented the synergy data by substituting a compound in a drug combination instance with another molecule that exhibits highly similar pharmacological effects. Using this protocol, we were able to upscale the AZ-DREAM Challenges dataset from 8,798 to 6,016,697 drug combinations. Comprehensive performance evaluations show that Random Forest and Gradient Boosting Trees models trained on the augmented data achieve higher accuracy than those trained solely on the original dataset. Our data augmentation protocol provides a systematic and unbiased approach to generating more diverse and larger-scale drug combination datasets, enabling the development of more precise and effective ML models. The protocol presented in this study could serve as a foundation for future research aimed at discovering novel and effective drug combinations for cancer treatment.

Study of prevalence and risk factors of chemotherapy-induced mucositis in gastrointestinal cancer using machine learning models

Objective

Chemotherapy-induced mucositis (CIM) significantly impacts clinical outcomes and diminishes the quality of life in patients with gastrointestinal cancer. This study aims to prospectively determine the incidence, severity, and underlying risk factors associated with CIM in this patient population.

Methods

To achieve this objective, we introduce a novel Machine Learning-based Toxicity Prediction Model (ML-TPM) designed to analyze the risk factors contributing to CIM development in gastrointestinal cancer patients. Within the winter season spanning from December 15th, 2018 to January 14th, 2019, we conducted in-person interviews with patients undergoing chemotherapy for gastrointestinal cancer. These interviews encompassed comprehensive questionnaires pertaining to patient demographics, CIM incidence, severity, and any supplementary prophylactic measures employed.

Results

The study encompassed a cohort of 447 participating patients who provided complete questionnaire responses (100%). Of these, 328 patients (73.4%) reported experiencing CIM during the course of their treatment. Notably, CIM-induced complications led to treatment discontinuation in 14 patients (3%). The most frequently encountered CIM symptoms were diarrhea (41.6%), followed by nausea (37.8%), vomiting (25.1%), abdominal pain (21%), gastritis (10.5%), and oral pain (10.3%). Supplementary prophylaxis was administered to approximately 62% of the patients. The analysis revealed significant correlations between the overall incidence of CIM and gender (p=0.015), number of chemotherapy cycles exceeding one (p=0.039), utilization of platinum-based regimens (p=0.039), and administration of irinotecan (p=0.003). Specifically, the incidence of diarrhea exhibited positive correlations with prior surgical history (p=0.037), irinotecan treatment (p=0.021), and probiotics usage (p=0.035). Conversely, diarrhea incidence demonstrated an adverse correlation with platinum-based treatment (p=0.026).

Conclusion

In conclusion, this study demonstrates the successful implementation of the ML-TPM model for automating toxicity prediction with accuracy comparable to conventional physical analyses. Our findings provide valuable insights into the identification of CIM risk factors among gastrointestinal cancer patients undergoing chemotherapy. Furthermore, the results underscore the potential of machine learning in enhancing our understanding of chemotherapy-induced mucositis and advancing personalized patient care strategies.

PLK1 as a cooperating partner for BCL2-mediated antiapoptotic program in leukemia

The deregulation of BCL2 family proteins plays a crucial role in leukemia development. Therefore, pharmacological inhibition of this family of proteins is becoming a prevalent treatment method. However, due to the emergence of primary and acquired resistance, efficacy is compromised in clinical or preclinical settings. We developed a drug sensitivity prediction model utilizing a deep tabular learning algorithm for the assessment of venetoclax sensitivity in T-cell acute lymphoblastic leukemia (T-ALL) patient samples. Through analysis of predicted venetoclax-sensitive and resistant samples, PLK1 was identified as a cooperating partner for the BCL2-mediated antiapoptotic program. This finding was substantiated by additional data obtained through phosphoproteomics and high-throughput kinase screening. Concurrent treatment using venetoclax with PLK1-specific inhibitors and PLK1 knockdown demonstrated a greater therapeutic effect on T-ALL cell lines, patient-derived xenografts, and engrafted mice compared with using each treatment separately. Mechanistically, the attenuation of PLK1 enhanced BCL2 inhibitor sensitivity through upregulation of BCL2L13 and PMAIP1 expression. Collectively, these findings underscore the dependency of T-ALL on PLK1 and postulate a plausible regulatory mechanism.

Computational Advancements in Cancer Combination Therapy Prediction

Given the high attrition rate of de novo drug discovery and limited efficacy of single-agent therapies in cancer treatment, combination therapy prediction through in silico drug repurposing has risen as a time- and cost-effective alternative for identifying novel and potentially efficacious therapies for cancer. The purpose of this review is to provide an introduction to computational methods for cancer combination therapy prediction and to summarize recent studies that implement each of these methods. A systematic search of the PubMed database was performed, focusing on studies published within the past 10 years. Our search included reviews and articles of ongoing and retrospective studies. We prioritized articles with findings that suggest considerations for improving combination therapy prediction methods over providing a meta-analysis of all currently available cancer combination therapy prediction methods. Computational methods used for drug combination therapy prediction in cancer research include networks, regression-based machine learning, classifier machine learning models, and deep learning approaches. Each method class has its own advantages and disadvantages, so careful consideration is needed to determine the most suitable class when designing a combination therapy prediction method. Future directions to improve current combination therapy prediction technology include incorporation of disease pathobiology, drug characteristics, patient multiomics data, and drug-drug interactions to determine maximally efficacious and tolerable drug regimens for cancer. As computational methods improve in their capability to integrate patient, drug, and disease data, more comprehensive models can be developed to more accurately predict safe and efficacious combination drug therapies for cancer and other complex diseases.

Machine Learning-Based Virtual Screening and Molecular Simulation Approaches Identified Novel Potential Inhibitors for Cancer Therapy

Cyclin-dependent kinase 2 (CDK2) is a promising target for cancer treatment, developing new effective CDK2 inhibitors is of great significance in anticancer therapy. The involvement of CDK2 in tumorigenesis has been debated, but recent evidence suggests that specifically inhibiting CDK2 could be beneficial in treating certain tumors. This approach remains attractive in the development of anticancer drugs. Several small-molecule inhibitors targeting CDK2 have reached clinical trials, but a selective inhibitor for CDK2 is yet to be discovered. In this study, we conducted machine learning-based drug designing to search for a drug candidate for CDK2. Machine learning models, including k-NN, SVM, RF, and GNB, were created to detect active and inactive inhibitors for a CDK2 drug target. The models were assessed using 10-fold cross-validation to ensure their accuracy and reliability. These methods are highly suitable for classifying compounds as either active or inactive through the virtual screening of extensive compound libraries. Subsequently, machine learning techniques were employed to analyze the test dataset obtained from the zinc database. A total of 25 compounds with 98% accuracy were predicted as active against CDK2. These compounds were docked into CDK2's active site. Finally, three compounds were selected based on good docking score, and, along with a reference compound, underwent MD simulation. The Gaussian naïve Bayes model yielded superior results compared to other models. The top three hits exhibited enhanced stability and compactness compared to the reference compound. In conclusion, our study provides valuable insights for identifying and refining lead compounds as CDK2 inhibitors.

Interpretable deep learning for improving cancer patient survival based on personal transcriptomes

Precision medicine chooses the optimal drug for a patient by considering individual differences. With the tremendous amount of data accumulated for cancers, we develop an interpretable neural network to predict cancer patient survival based on drug prescriptions and personal transcriptomes (CancerIDP). The deep learning model achieves 96% classification accuracy in distinguishing short-lived from long-lived patients. The Pearson correlation between predicted and actual months-to-death values is as high as 0.937. About 27.4% of patients may survive longer with an alternative medicine chosen by our deep learning model. The median survival time of all patients can increase by 3.9 months. Our interpretable neural network model reveals the most discriminating pathways in the decision-making process, which will further facilitate mechanistic studies of drug development for cancers.

The covert symphony: cellular and molecular accomplices in breast cancer metastasis

Breast cancer has emerged as the most commonly diagnosed cancer and primary cause of cancer-related deaths among women worldwide. Although significant progress has been made in targeting the primary tumor, the effectiveness of systemic treatments to prevent metastasis remains limited. Metastatic disease continues to be the predominant factor leading to fatality in the majority of breast cancer patients. The existence of a prolonged latency period between initial treatment and eventual recurrence in certain patients indicates that tumors can both adapt to and interact with the systemic environment of the host, facilitating and sustaining the progression of the disease. In order to identify potential therapeutic interventions for metastasis, it will be crucial to gain a comprehensive framework surrounding the mechanisms driving the growth, survival, and spread of tumor cells, as well as their interaction with supporting cells of the microenvironment. This review aims to consolidate recent discoveries concerning critical aspects of breast cancer metastasis, encompassing the intricate network of cells, molecules, and physical factors that contribute to metastasis, as well as the molecular mechanisms governing cancer dormancy.

Prognostic models for upper urinary tract urothelial carcinoma patients after radical nephroureterectomy based on a novel systemic immune-inflammation score with machine learning

PURPOSE

This study aimed to evaluate the clinical significance of a novel systemic immune-inflammation score (SIIS) to predict oncological outcomes in upper urinary tract urothelial carcinoma(UTUC) after radical nephroureterectomy(RNU).

METHOD

The clinical data of 483 patients with nonmetastatic UTUC underwent surgery in our center were analyzed. Five inflammation-related biomarkers were screened in the Lasso-Cox model and then aggregated to generate the SIIS based on the regression coefficients. Overall survival (OS) was assessed using Kaplan-Meier analyses. The Cox proportional hazards regression and random survival forest model were adopted to build the prognostic model. Then we established an effective nomogram for UTUC after RNU based on SIIS. The discrimination and calibration of the nomogram were evaluated using the concordance index (C-index), area under the time-dependent receiver operating characteristic curve (time-dependent AUC), and calibration curves. Decision curve analysis (DCA) was used to assess the net benefits of the nomogram at different threshold probabilities.

RESULT

According to the median value SIIS computed by the lasso Cox model, the high-risk group had worse OS (p<0.0001) than low risk-group. Variables with a minimum depth greater than the depth threshold or negative variable importance were excluded, and the remaining six variables were included in the model. The area under the ROC curve (AUROC) of the Cox and random survival forest models were 0.801 and 0.872 for OS at five years, respectively. Multivariate Cox analysis showed that elevated SIIS was significantly associated with poorer OS (p<0.001). In terms of predicting overall survival, a nomogram that considered the SIIS and clinical prognostic factors performed better than the AJCC staging.

CONCLUSION

The pretreatment levels of SIIS were an independent predictor of prognosis in upper urinary tract urothelial carcinoma after RNU. Therefore, incorporating SIIS into currently available clinical parameters helps predict the long-term survival of UTUC.

Integrative toxicogenomics: Advancing precision medicine and toxicology through artificial intelligence and OMICs technology

More information about a person's genetic makeup, drug response, multi-omics response, and genomic response is now available leading to a gradual shift towards personalized treatment. Additionally, the promotion of non-animal testing has fueled the computational toxicogenomics as a pivotal part of the next-gen risk assessment paradigm. Artificial Intelligence (AI) has the potential to provid new ways analyzing the patient data and making predictions about treatment outcomes or toxicity. As personalized medicine and toxicogenomics involve huge data processing, AI can expedite this process by providing powerful data processing, analysis, and interpretation algorithms. AI can process and integrate a multitude of data including genome data, patient records, clinical data and identify patterns to derive predictive models anticipating clinical outcomes and assessing the risk of any personalized medicine approaches. In this article, we have studied the current trends and future perspectives in personalized medicine & toxicology, the role of toxicogenomics in connecting the two fields, and the impact of AI on personalized medicine & toxicology. In this work, we also study the key challenges and limitations in personalized medicine, toxicogenomics, and AI in order to fully realize their potential.

Roadmap for precision preclinical x-ray radiation studies

This Roadmap paper covers the field of precision preclinical x-ray radiation studies in animal models. It is mostly focused on models for cancer and normal tissue response to radiation, but also discusses other disease models. The recent technological evolution in imaging, irradiation, dosimetry and monitoring that have empowered these kinds of studies is discussed, and many developments in the near future are outlined. Finally, clinical translation and reverse translation are discussed.

Deep learning methods for drug response prediction in cancer: Predominant and emerging trends

Cancer claims millions of lives yearly worldwide. While many therapies have been made available in recent years, by in large cancer remains unsolved. Exploiting computational predictive models to study and treat cancer holds great promise in improving drug development and personalized design of treatment plans, ultimately suppressing tumors, alleviating suffering, and prolonging lives of patients. A wave of recent papers demonstrates promising results in predicting cancer response to drug treatments while utilizing deep learning methods. These papers investigate diverse data representations, neural network architectures, learning methodologies, and evaluations schemes. However, deciphering promising predominant and emerging trends is difficult due to the variety of explored methods and lack of standardized framework for comparing drug response prediction models. To obtain a comprehensive landscape of deep learning methods, we conducted an extensive search and analysis of deep learning models that predict the response to single drug treatments. A total of 61 deep learning-based models have been curated, and summary plots were generated. Based on the analysis, observable patterns and prevalence of methods have been revealed. This review allows to better understand the current state of the field and identify major challenges and promising solution paths.

A Receptor Tyrosine Kinase Inhibitor Sensitivity Prediction Model Identifies AXL Dependency in Leukemia

Despite incredible progress in cancer treatment, therapy resistance remains the leading limiting factor for long-term survival. During drug treatment, several genes are transcriptionally upregulated to mediate drug tolerance. Using highly variable genes and pharmacogenomic data for acute myeloid leukemia (AML), we developed a drug sensitivity prediction model for the receptor tyrosine kinase inhibitor sorafenib and achieved more than 80% prediction accuracy. Furthermore, by using Shapley additive explanations for determining leading features, we identified AXL as an important feature for drug resistance. Drug-resistant patient samples displayed enrichment of protein kinase C (PKC) signaling, which was also identified in sorafenib-treated FLT3-ITD-dependent AML cell lines by a peptide-based kinase profiling assay. Finally, we show that pharmacological inhibition of tyrosine kinase activity enhances AXL expression, phosphorylation of the PKC-substrate cyclic AMP response element binding (CREB) protein, and displays synergy with AXL and PKC inhibitors. Collectively, our data suggest an involvement of AXL in tyrosine kinase inhibitor resistance and link PKC activation as a possible signaling mediator.

Machine learning facilitated structural activity relationship approach for the discovery of novel inhibitors targeting EGFR

This research manuscript aims to find the most effective epidermal growth factor receptor (EGFR) inhibitors from millions of in house compounds through Machine Learning (ML) techniques. ML-based structure activity relationship (SAR) models were validated to predict biological activity of untested novel molecules. Six ML algorithms, including k nearest neighbour (KNN), decision tree (DT), Logistic Regression, support vector machine (SVM), multilinear regression (MLR), and random forest (RF), were used to build for activity prediction. Among these, RF classifier (accuracy for train and test set is 90% and 81%) and RF regressor (R2 and MSE for trainset is 0.83 and 0.29 and for test set, 0.69 and 0.46) showed good predictive performance. Also, the six most essential features that affect the biological activity parameter and highly contribute to model development were successfully selected by the variable importance technique. RF regression model was used to predict the biological activity expressed as pIC50 of nearly ten million molecules while RF classification model classifies those molecules into active, moderately active, and least active according to their predicted pIC50. Based on two models, thousand molecules from million molecules with higher predicted pIC50 values and classified as active were selected for molecular docking. Based on the docking scores, predicted pIC50, and binding interactions with MET769 residue, compounds, i.e., Zinc257233137, Zinc257232249, and Zinc101379788, were identified as potential EGFR inhibitors with predicted pIC50 7.72, 7.85, and 7.70. Dynamics studies were also performed on Zinc257233137 to illustrate that it has good binding free energy and stable hydrogen bonding interactions with EGFR. These molecules can be used for further research and proved to be the novel drugs for EGFR in cancer treatment.Communicated by Ramaswamy H. Sarma.

A deep tabular data learning model predicting cisplatin sensitivity identifies BCL2L1 dependency in cancer

Cisplatin, a platinum-based chemotherapeutic agent, is widely used as a front-line treatment for several malignancies. However, treatment outcomes vary widely due to intrinsic and acquired resistance. In this study, cisplatin-perturbed gene expression and pathway enrichment were used to define a gene signature, which was further utilized to develop a cisplatin sensitivity prediction model using the TabNet algorithm. The TabNet model performed better (>80 % accuracy) than all other machine learning models when compared to a wide range of machine learning algorithms. Moreover, by using feature importance and comparing predicted ovarian cancer patient samples, BCL2L1 was identified as an important gene contributing to cisplatin resistance. Furthermore, the pharmacological inhibition of BCL2L1 was found to synergistically increase cisplatin efficacy. Collectively, this study developed a tool to predict cisplatin sensitivity using cisplatin-perturbed gene expression and pathway enrichment knowledge and identified BCL2L1 as an important gene in this setting.

Classification and diagnostic prediction of breast cancer metastasis on clinical data using machine learning algorithms

Metastatic Breast Cancer (MBC) is one of the primary causes of cancer-related deaths in women. Despite several limitations, histopathological information about the malignancy is used for the classification of cancer. The objective of our study is to develop a non-invasive breast cancer classification system for the diagnosis of cancer metastases. The anaconda-Jupyter notebook is used to develop various python programming modules for text mining, data processing, and Machine Learning (ML) methods. Utilizing classification model cross-validation criteria, including accuracy, AUC, and ROC, the prediction performance of the ML models is assessed. Welch Unpaired t-test was used to ascertain the statistical significance of the datasets. Text mining framework from the Electronic Medical Records (EMR) made it easier to separate the blood profile data and identify MBC patients. Monocytes revealed a noticeable mean difference between MBC patients as compared to healthy individuals. The accuracy of ML models was dramatically improved by removing outliers from the blood profile data. A Decision Tree (DT) classifier displayed an accuracy of 83% with an AUC of 0.87. Next, we deployed DT classifiers using Flask to create a web application for robust diagnosis of MBC patients. Taken together, we conclude that ML models based on blood profile data may assist physicians in selecting intensive-care MBC patients to enhance the overall survival outcome.

Identification of the diagnostic genes and immune cell infiltration characteristics of gastric cancer using bioinformatics analysis and machine learning

Background: Finding reliable diagnostic markers for gastric cancer (GC) is important. This work uses machine learning (ML) to identify GC diagnostic genes and investigate their connection with immune cell infiltration. Methods: We downloaded eight GC-related datasets from GEO, TCGA, and GTEx. GSE13911, GSE15459, GSE19826, GSE54129, and GSE79973 were used as the training set, GSE66229 as the validation set A, and TCGA & GTEx as the validation set B. First, the training set screened differentially expressed genes (DEGs), and gene ontology (GO), kyoto encyclopedia of genes and genomes (KEGG), disease Ontology (DO), and gene set enrichment analysis (GSEA) analyses were performed. Then, the candidate diagnostic genes were screened by LASSO and SVM-RFE algorithms, and receiver operating characteristic (ROC) curves evaluated the diagnostic efficacy. Then, the infiltration characteristics of immune cells in GC samples were analyzed by CIBERSORT, and correlation analysis was performed. Finally, mutation and survival analyses were performed for diagnostic genes. Results: We found 207 up-regulated genes and 349 down-regulated genes among 556 DEGs. gene ontology analysis significantly enriched 413 functional annotations, including 310 biological processes, 23 cellular components, and 80 molecular functions. Six of these biological processes are closely related to immunity. KEGG analysis significantly enriched 11 signaling pathways. 244 diseases were closely related to Ontology analysis. Multiple entries of the gene set enrichment analysis analysis were closely related to immunity. Machine learning screened eight candidate diagnostic genes and further validated them to identify ABCA8, COL4A1, FAP, LY6E, MAMDC2, and TMEM100 as diagnostic genes. Six diagnostic genes were mutated to some extent in GC. ABCA8, COL4A1, LY6E, MAMDC2, TMEM100 had prognostic value. Conclusion: We screened six diagnostic genes for gastric cancer through bioinformatic analysis and machine learning, which are intimately related to immune cell infiltration and have a definite prognostic value.