Applications of Artificial Intelligence forPrecision Medicine in AML

Introduction: Acute myeloid leukemia (AML) is the mostcommon myeloid neoplasm and shows naturally high geneticand phenotypic heterogeneity. Machine learning (ML) is anapplication of artificial intelligence (AI) that defines a dataanalysis method that automatically learns how to analyze bigdata. Since AML’s heterogeneous nature and because of somegrowing concerns such as increasing prevalence, ineffectivetreatment and poor prognosis of AML, it represents the goodcandidate for ML. Recent advances in ML algorithms have provideda better understanding of AML biology which may leadto a better prognosis and prediction of this disease. ML applicationsin AML includes automated diagnosis and classificationof AML, predictions about treatment decisions in AML, predictionof drug sensitivities, and image analysis. Also, ML-baseddiagnostic strategies may be helpful to clarify obscure data andsimplify the process of diagnosis. In this review, we intended todemonstrate the application of ML algorithms in AML diagnosisand classification, which is a step forward for personalizedmedicine in AML patients.Methods: In this review, we searched in PubMed, SCOPUSand Google Scholar databases by "Artificial intelligence","Acute myeloid leukemia" and "personalized medicine" keywordsto find related articles.Results: Studies revealed that artificial intelligence and its algorithmscan efficiently classify and diagnose AML. For example,in ۲۰۱۵, Reta et al. showed that SVM and KNN algorithmscan be used for effectively AML classification with ۹۴%accuracy. In ۲۰۲۰, Dasariraju et al. exhibited that reinforcementalgorithm (a types of ML) have efficient role in detectionand classification of immature leukocytes for diagnosis ofAML. They reported that accuracy of detection and classificationwas ۹۲.۹۹% and ۹۳.۴۵%, respectively. In another study in۲۰۲۱, Skead et al. demontrated that deep neural network candiscriminate between AML types with high accuracy. Also, in۲۰۲۲ Arabyarmohammadi et al. indicated in their research thatdeep learning can estimate recurrence of the disease with highaccuracy and sensitivity.Conclusion: So far, researches that have been conducted in thisfield have fascinatingly confirmed the ability of ML to improvethe AML diagnosis and management. By help of the ML algorithms,big data obtained from AML patients can be processedeffectively. This processed data can be very useful for diagnosis,classification, and even targeted and accurate treatment ofAML.

کلیدواژه ها:

Acute Myeloid Leukemia (AML) ، Artificial Intelligence(AI) ، AML Diagnosis ، Personalized Medicine