Проапгрейдил вступление

report.tex

@@ -130,7 +130,17 @@
     
     \section*{Introduction}
     \addcontentsline{toc}{section}{Introduction}
-    Progress has been made in chemotherapy drugs, but drug resistance remains a major challenge in cancer treatment and the main cause of cancer progression and even death. However, there are no clear indicators for predicting the risk of drug resistance in patients. Existing drug sensitivity assessment methods has limitations such as low modeling success rates, high cost, and time-consuming process. Machine learning is both an expanding and evolving field of computing, and it seems that it can significantly help in solving chemotherapy resistance problem. Here we provide an overview of how different studies apply machine learning algorithms to predict and understand chemotherapy resistance in various cancer types. Also we consider the strengths and limitations of each approach and discuss obtained results.
+    Cancer remains one of the leading causes of mortality worldwide, presenting a significant challenge to global health despite considerable advancements in medical research and treatment. According to the data from the International Agency for Research on Cancer (IARC), there were over 19 million new cancer cases and nearly 10 million cancer-related deaths globally in 2020~\cite{cancer}. The high mortality rate is attributed to factors such as late diagnosis, the aggressive nature of certain cancer types, and the complexity of developing effective treatment strategies. The impact of cancer is profound, not only on the patients but also on their families and healthcare systems, underscoring the need for continual improvement in cancer management.
+
+    Chemotherapy is a cornerstone in cancer treatment, widely utilized either as a primary or adjuvant therapy to target and destroy rapidly dividing cancer cells. It is often the first-line treatment for various cancers, including ovarian, lung, and cervical cancers~\cite{therapy}. Standard chemotherapy regimens, such as platinum-based drugs combined with paclitaxel, have significantly improved patient survival rates by inhibiting tumor growth and controlling metastasis. For example, in advanced ovarian cancer, debulking surgery combined with six cycles of platinum and paclitaxel chemotherapy is the standard approach, aiming to reduce tumor burden and eradicate micrometastases~\cite{treateoc}.
+
+    However, a major challenge in chemotherapy is the development of drug resistance, which leads to treatment failure, disease progression, and adversely affects patient survival~\cite{resistance}. Drug resistance can be intrinsic, where tumors inherently do not respond to chemotherapy, or acquired, developing after initial responsiveness due to factors like genetic mutations and cellular adaptations. For example, in epithelial ovarian cancer (EOC), approximately 80\% of patients experience relapse after initial remission because of chemotherapy resistance, making it a significant factor contributing to the ineffectiveness of treatment and the leading cause of death in these cases~\cite{paclitaxel}. Similarly, in cervical cancer, up to 40\% of patients exhibit resistance to platinum-based neoadjuvant chemotherapy, potentially delaying effective surgical interventions and accelerating disease progression~\cite{cervical}. Existing chemotherapy resistance assessment methods often have limitations such as low success rates in modeling, high costs, and time-consuming processes. 
+
+    Machine learning and deep learning has made dramatic breakthroughs in recent years to contribute to the field of medicine~\cite{mlrole}. It has been widely applied to various classification and regression problems, especially in the field of biology where the amount and complexity of data is growing. Machine learning in drug resistance is used for identifying critical genetic, molecular, and morphological features associated with resistance, developing predictive models to classify sensitive and resistant cells, and analyzing complex datasets to uncover patterns. It is also applied to design personalized treatment strategies, predict patient-specific drug responses, and explore mechanisms driving resistance, aiding in the optimization of therapeutic approaches.
+    
+    Machine learning models are often considered "black boxes", as there is typically a limited understanding of how the outputs are produced from the model input. However, in clinical medicine tasks, it is equally important to understand the features that drive a model’s decision-making. To address this, researchers use different tools and approaches to perform feature importance analysis, e.g., SHapley Additive exPlanations (SHAP)~\cite{shap}, least absolute shrinkage and selection operator (LASSO)~\cite{lasso}, DALEX~\cite{dalex}, and other. Such analysis helps to better understand the underlying causes and mechanisms of resistance, providing valuable insights that can guide the development of more effective treatment strategies and targeted therapeutic interventions.
+
+    This article provides a comprehensive overview of how machine learning algorithms have been applied to predict and analyze chemotherapy resistance across various cancer types. We explore diverse approaches employed in the field, ranging from feature extraction and classification to regression and prognostic modeling. The goal of this review is to shed light on the current state of the art in using machine learning for assessing drug resistance and to identify opportunities for future research that can enhance the precision and effectiveness of cancer treatment strategies.
 
     \newpage
 
@@ -146,7 +156,7 @@
 
     The authors of article~\cite{heterogeneity} chose a different approach by applying machine learning algorithms from the specialized software CellProfiler~\cite{cellprofile} to extract quantitative image features. They subsequently used bioinformatics analysis to explore the relationship between these features of intra-tumor heterogeneity (ITH) and drug resistance. Notably, the authors did not aim to train new models but instead utilized pre-trained algorithms from CellProfiler. Unlike studies \cite{paclitaxel}, \cite{mitochondria}, and \cite{sers}, where algorithms were employed for regression and classification tasks, this research focused specifically on extracting quantitative features from images. Based on CellProfiler, the authors constructed a pipeline for the extraction and analysis of these features, which enabled them to draw conclusions regarding the connection between these features and drug resistance in cancer cells.
 
-    In~\cite{platinum}, the authors performed differential protein analysis on the expression profiles of 745 proteins related to platinum-based chemotherapy resistance. They used LASSO regression to select 10 proteins linked to chemotherapy outcomes, followed by univariate logistic regression on nine clinical factors. Variables with p < 0.1 were included in a multivariate logistic regression analysis, resulting in four significant variables: three proteins and one clinical parameter (postoperative residual tumor). This analysis enabled the construction of a predictive machine-learning model for chemotherapy resistance in patients with EOC.
+    In~\cite{platinum}, the authors performed differential protein analysis on the expression profiles of 745 proteins related to platinum-based chemotherapy resistance. They used LASSO regression~\cite{lasso} to select 10 proteins linked to chemotherapy outcomes, followed by univariate logistic regression on nine clinical factors. Variables with p < 0.1 were included in a multivariate logistic regression analysis, resulting in four significant variables: three proteins and one clinical parameter (postoperative residual tumor). This analysis enabled the construction of a predictive machine-learning model for chemotherapy resistance in patients with EOC.
 
     The authors of article~\cite{kras} applied machine learning algorithms for two goals. Firstly, they used algorithms to extract genes highly related with therapy resistance. Each sample of their data contained the expression of 8687 genes and only a small portion was correlated with targeted therapy resistance. To extract highly related genes in this study authors attempted seven algorithms, including Least Absolute Shrinkage and Selection Operator (LASSO), Light Gradient Boosting Machine (LightGBM), Monte Carlo Feature Selection (MCFS), Minimum Redundancy Maximum Relevance (mRMR), Random Forest (RF) -based, Categorical Boosting (CATBoost), and eXtreme Gradient Boosting (XGBoost). Secondly, they selected four algorithms to perform binary classification (resistant vs sensitive) of tumor cells based on extracted features, namely, random forest (RF), support vector machine (SVM), K-Nearest Neighbors (KNN), and decision tree (DT).
 
@@ -189,7 +199,7 @@
        
     Machine learning algorithms are often regarded as black boxes, providing powerful predictions but limited insight into the underlying mechanisms driving those predictions. In fields such as medicine and biology, however, interpretability is not just a desirable feature—it is a necessity. Transparent models and interpretable outputs are critical for ensuring that predictions and recommendations can be explained, validated, and trusted, especially when they influence life-altering decisions. In the context of cancer treatment, this need for interpretability becomes even more pressing. Misguided treatment decisions can delay the administration of effective therapies, significantly increasing the risk of mortality. Beyond guiding clinical decisions, the interpretability of features in machine learning models also offers a unique opportunity to deepen our understanding of drug resistance in cancer. By unraveling the biological and molecular underpinnings of resistance, we can develop more targeted and effective therapeutic strategies.
 
-    The authors of \cite{paclitaxel} applied the SHapley Additive exPlanations (SHAP), a model interpretation framework, to quantify and rank the feature contributions for classification models. After analysing feature importances authors were able to reduce count of the features from 112 to only 25 and even increase the accuracy of their models. Interestingly that their models were able to achieve approximately 78\% even when only the top three features were utilized for classification.
+    The authors of \cite{paclitaxel} applied the SHapley Additive exPlanations (SHAP)~\cite{shap}, a model interpretation framework, to quantify and rank the feature contributions for classification models. After analysing feature importances authors were able to reduce count of the features from 112 to only 25 and even increase the accuracy of their models. Interestingly that their models were able to achieve approximately 78\% even when only the top three features were utilized for classification.
 
     In \cite{tabular} authors also used SHAP to performe feature analysis. In this study, feature importance analysis was utilized to identify key genes associated with cisplatin resistance. By leveraging feature importance derived from multiple predictive models, the authors highlighted BCL2L1 as a critical gene mediating cisplatin resistance. Notably, earlier studies suggested that $\beta$-catenin expression is necessary for maintaining elevated levels of BCL2L1 in cisplatin-resistant cells. Additionally, the analysis revealed the involvement of PLK2, whose expression was linked to $\beta$-catenin activity. The authors demonstrated that lower expression levels of BCL2L1 and PLK2 correlated with improved outcomes in ovarian cancer, and inhibitors targeting these genes showed a synergistic effect when combined with cisplatin. These findings underscore the importance of the $\beta$-catenin/BCL2L1 axis in driving resistance and provide potential targets for enhancing cisplatin efficacy.
 
@@ -304,6 +314,20 @@
         A. Nasimian, M. Ahmed, I. Hedenfalk, and J. U. Kazi, “A deep tabular data learning model predicting cisplatin sensitivity identifies BCL2L1 dependency in cancer,” Computational and Structural Biotechnology Journal, vol. 21, pp. 956–964, doi: 10.1016/j.csbj.2023.01.020.
         \bibitem{deep}
         J. Longden et al., “Deep neural networks identify signaling mechanisms of ErbB-family drug resistance from a continuous cell morphology space,” Cell Reports, vol. 34, no. 3, p. 108657, Jan. 2021, doi: 10.1016/j.celrep.2020.108657.
+        \bibitem{cancer}
+        International Agency for Research on Cancer, F. Bray, IARC, E. Weiderpass, and World Health Organization, “Latest global cancer data: Cancer burden rises to 19.3 million new cases and 10.0 million cancer deaths in 2020,” IARC, Dec. 15, 2020. \url{https://www.iarc.who.int/wp-content/uploads/2020/12/pr292_E.pdf} (accessed Dec. 01, 2024).
+        \bibitem{therapy}
+        Sh. Huang and B. O. Sullivan, “Oral cancer: Current role of radiotherapy and chemotherapy,” Medicina Oral, Patología Oral Y Cirugía Bucal, pp. e233–e240, Jan. 2013, doi: 10.4317/medoral.18772.
+        \bibitem{treateoc}
+        L. Kuroki and S. R. Guntupalli, “Treatment of epithelial ovarian cancer,” BMJ, p. m3773, Nov. 2020, doi: 10.1136/bmj.m3773.
+        \bibitem{resistance}
+        S. W. Johnson, R. F. Ozols, and T. C. Hamilton, “Mechanisms of drug resistance in ovarian cancer,” Cancer, vol. 71, no. S2, pp. 644–649, Aug. 2010, doi: 10.1002/cncr.2820710224.
+        \bibitem{mlrole}
+        Y. Jiang, M. Yang, S. Wang, X. Li, and Y. Sun, “Emerging role of deep learning‐based artificial intelligence in tumor pathology,” Cancer Communications, vol. 40, no. 4, pp. 154–166, Apr. 2020, doi: 10.1002/cac2.12012.
+        \bibitem{shap}
+        S. M. Lundberg and S.-I. Lee, “A unified approach to interpreting model predictions,” arXiv (Cornell University), Jan. 2017, doi: 10.48550/arxiv.1705.07874.
+        \bibitem{lasso}
+        R. Tibshirani, “Regression shrinkage and selection via the lasso,” Journal of the Royal Statistical Society Series B (Statistical Methodology), vol. 58, no. 1, pp. 267–288, Jan. 1996, doi: 10.1111/j.2517-6161.1996.tb02080.x.
         \bibitem{cellprofile}
         C. McQuin et al., “CellProfiler 3.0: Next-generation image processing for biology,” PLoS Biology, vol. 16, no. 7, p. e2005970, Jul. 2018, doi: 10.1371/journal.pbio.2005970.
         \bibitem{tcga}