Abstarct: Abstract The increasing use of smartphones baxsed on the Android operating system has made this platform one of the most important targets of malware attacks. Android malware, using complex and variable methods, is not only a serious threat to users' privacy and information security, but also brings extensive economic and social losses. Therefore, identifying and dealing with these threats has become one of the main challenges in the field of information security and requires the development of efficient and accurate methods for their detection. In this research, focusing on the use of feature selection methods, machine learning algorithms and their combination, a new method is presented to improve the accuracy and efficiency in detecting Android malware. The method proposed in this thesis is baxsed on the use of two categories of static and dynamic features in the detection process. In summary, the proposed method first identifies the importance of features using the ANOVA algorithm to eliminate low-impact features and reduce the dimensions of the problem. Then, an improved version of the SFE feature selection algorithm is used to search for optimal subsets in the reduced search space. Finally, a classifier combination method (weighted soft voting) is used so that the output of stronger models has a greater contribution to the final decision. The results obtained from the experiments showed that the proposed method was able to perform better than the conventional algorithms and the methods presented in recent years. In the static features section, the accuracy of the proposed method reached 95.4% and the F1 value reached 94.6%, which was significantly superior to other methods. Also, in the dynamic features section, the combined method managed to achieve an accuracy of 84.4% and an F1 value of 84.6%, which was better than the others. The results of this study show that the intelligent combination of feature selection algorithms and classifiers not only provides higher accuracy, but also increases the stability and generalizability of the proposed method. Overall, the present study has taken an effective step towards improving the security of this platform by presenting a hybrid model baxsed on machine learning and feature selection of Android applications. The results obtained can be a basis for the development of intelligent and real-time systems in the field of malware detection of smart devices and pave the way for future research to improve the security of users of this operating system.