Learning With Multiclass AUC: Theory and Algorithms

Yang, Zhiyong; Xu, Qianqian; Bao, Shanhu; Cao, Xiaochun; Huang, Qingming

doi:10.1109/tpami.2021.3101125

Cited by 49 publications

(20 citation statements)

References 50 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Unlike previous approaches, this model used has readily accessible information to physicians such as donor type, conditioning used, degree, and HLA mismatched, indicating that this model may have greater potential for translation into widespread clinical practice. In this model and similar ensemble models, the reported area under the curve (AUC), -was 61.3-64% [43,59,60]. Studies using penalized logistic regression and decision tree models also achieved similar accuracy [14,21].…”

Section: Predicting the Risk Of Graft-versus-host Diseasementioning

confidence: 74%

Machine learning applications and challenges in graft-versus-host disease: a scoping review

Mushtaq,

Shafqat,

Salah

et al. 2023

Current Opinion in Oncology

View full text Add to dashboard Cite

Purpose of review This review delves into the potential of artificial intelligence (AI), particularly machine learning (ML), in enhancing graft-versus-host disease (GVHD) risk assessment, diagnosis, and personalized treatment. Recent findings Recent studies have demonstrated the superiority of ML algorithms over traditional multivariate statistical models in donor selection for allogeneic hematopoietic stem cell transplantation. ML has recently enabled dynamic risk assessment by modeling time-series data, an upgrade from the static, “snapshot” assessment of patients that conventional statistical models and older ML algorithms offer. Regarding diagnosis, a deep learning model, a subset of ML, can accurately identify skin segments affected with chronic GVHD with satisfactory results. ML methods such as Q-learning and deep reinforcement learning have been utilized to develop adaptive treatment strategies (ATS) for the personalized prevention and treatment of acute and chronic GVHD. Summary To capitalize on these promising advancements, there is a need for large-scale, multicenter collaborations to develop generalizable ML models. Furthermore, addressing pertinent issues such as the implementation of stringent ethical guidelines is crucial before the widespread introduction of AI into GVHD care.

show abstract

Section: Predicting the Risk Of Graft-versus-host Diseasementioning

confidence: 74%

Machine learning applications and challenges in graft-versus-host disease: a scoping review

Mushtaq,

Shafqat,

Salah

et al. 2023

Current Opinion in Oncology

View full text Add to dashboard Cite

show abstract

“…When AUC > 0.5, the closer the AUC is to 1, the better the diagnostic effect. When AUC = 0.5, the diagnostic method is completely ineffective[ 29 ]. In this study, we assessed the specificity and sensitivity of the risk score model in predicting the survival and prognosis of GBM patients through the value of AUC.…”

Section: Methodsmentioning

confidence: 99%

Cell senescence-associated genes predict the malignant characteristics of glioblastoma

et al. 2022

View full text Add to dashboard Cite

Background Glioblastoma (GBM) is the most malignant, aggressive and recurrent primary brain tumor. Cell senescence can cause irreversible cessation of cell division in normally proliferating cells. According to studies, senescence is a primary anti-tumor mechanism that may be seen in a variety of tumor types. It halts the growth and spread of tumors. Tumor suppressive functions held by cellular senescence provide new directions and pathways to promote cancer therapy. Methods We comprehensively analyzed the cell senescence-associated genes expression patterns. The potential molecular subtypes were acquired based on unsupervised cluster analysis. The tumor immune microenvironment (TME) variations, immune cell infiltration, and stemness index between 3 subtypes were analyzed. To identify genes linked with GBM prognosis and build a risk score model, we used weighted gene co-expression network analysis (WGCNA), univariate Cox regression, Least absolute shrinkage and selection operator regression (LASSO), and multivariate Cox regression analysis. And the correlation between risk scores and clinical traits, TME, GBM subtypes, as well as immunotherapy responses were estimated. Immunohistochemistry (IHC) and cellular experiments were performed to evaluate the expression and function of representative genes. Then the 2 risk scoring models were constructed based on the same method of calculation whose samples were acquired from the CGGA dataset and TCGA datasets to verify the rationality and the reliability of the risk scoring model. Finally, we conducted a pan-cancer analysis of the risk score, assessed drug sensitivity based on risk scores, and analyzed the pathways of sensitive drug action. Results The 3 potential molecular subtypes were acquired based on cell senescence-associated genes expression. The Log-rank test showed the difference in GBM patient survival between 3 potential molecular subtypes (P = 0.0027). Then, 11 cell senescence-associated genes were obtained to construct a risk-scoring model, which was systematically randomized to distinguish the train set (n = 293) and the test set (n = 292). The Kaplan-Meier (K-M) analyses indicated that the high-risk score in the train set (P < 0.0001), as well as the test set (P = 0.0053), corresponded with poorer survival. In addition, the high-risk score group showed a poor response to immunotherapy. The reliability and credibility of the risk scoring model were confirmed according to the CGGA dataset, TCGA datasets, and Pan-cancer analysis. According to drug sensitivity analysis, it was discovered that LJI308, a potent selective inhibitor of RSK pathways, has the highest drug sensitivity. Moreover, the GBM patients with higher risk scores may potentially be more beneficial from drugs that target cell cycle, mitosis, microtubule, DNA replication and apoptosis regulation signaling. Conclusion We identified potential associations between clinical characteristics, TME, stemness, subtypes, and immunotherapy, and we clarified the therapeutic usefulness of cell senescence-associated genes.

show abstract

“…The first task in our set of experiments is classifying the different tumor types in the pan-cancer study. The classification performance was measured using five metrics: Accuracy, Macro-averaged F1 score, precision, recall, and ROC-AUC [41]. We also perform a second classification task to validate our findings on a smaller dataset and test the robustness of our method.…”

Section: Test Cases and Datasetsmentioning

confidence: 97%

CustOmics: A versatile deep-learning based strategy for multi-omics integration

Hakim

Pradat²,

Michiels³

et al. 2023

PLoS Comput Biol

View full text Add to dashboard Cite

The availability of patient cohorts with several types of omics data opens new perspectives for exploring the disease’s underlying biological processes and developing predictive models. It also comes with new challenges in computational biology in terms of integrating high-dimensional and heterogeneous data in a fashion that captures the interrelationships between multiple genes and their functions. Deep learning methods offer promising perspectives for integrating multi-omics data. In this paper, we review the existing integration strategies based on autoencoders and propose a new customizable one whose principle relies on a two-phase approach. In the first phase, we adapt the training to each data source independently before learning cross-modality interactions in the second phase. By taking into account each source’s singularity, we show that this approach succeeds at taking advantage of all the sources more efficiently than other strategies. Moreover, by adapting our architecture to the computation of Shapley Additive explanations, our model can provide interpretable results in a multi-source setting. Using multiple omics sources from different TCGA cohorts, we demonstrate the performance of the proposed method for cancer on test cases for several tasks, such as the classification of tumor types and breast cancer subtypes, as well as survival outcome prediction. We show through our experiments the great performances of our architecture on seven different datasets with various sizes and provide some interpretations of the results obtained. Our code is available on (https://github.com/HakimBenkirane/CustOmics).

show abstract

Learning With Multiclass AUC: Theory and Algorithms

Cited by 49 publications

References 50 publications

Machine learning applications and challenges in graft-versus-host disease: a scoping review

Machine learning applications and challenges in graft-versus-host disease: a scoping review

Cell senescence-associated genes predict the malignant characteristics of glioblastoma

CustOmics: A versatile deep-learning based strategy for multi-omics integration

Contact Info

Product

Resources

About