Disease Prediction System using Data Mining Techniques based on Classification Mechanism: Survey Study

The widespread dissemination and accessibility of information have led to unprecedented amounts of information. A huge part of this information is random and untapped, while very little of it is regulated. This has led to the urgent need to regulate this impressive volume of information for use in many tasks such as corporate decision-making, increasing their competitiveness, etc. This has led to creating and developing algorithms with the ability to classify and organize data and extract knowledge from it. This facilitates the process of predicting, detecting, or preventing diseases, thereby preserving human capital, reducing expenditures, and keeping society healthy. This technology is a promising opportunity for investment and growth in various fields. The latest statistics of the Saudi General Authority for Statistics were in its report on the results of the survey [1]. (Chronic diagnosed diseases among the Kingdom's population are among those aged 15 and over, at 16.4%. Chronic diseases increase significantly as age increases. The prevalence of chronic diseases among older persons in the 65-year age group is 7.7% higher than among young age groups (15-34 years, at 4.4%). This is what motivated the researcher to find the best means of predicting diseases and how they might help us understand the initial signals of diseases and avoid them. The study aims to review and analyze classification applications, clarify their uses and important features in the field of disease detection and the future of these classifications in the Kingdom, and compile the latest studies in disease prediction using classification algorithms. The researcher used a survey approach to answer the question of this research. This survey includes a review of previous studies from 2018 to 2021 in the Disease Prediction System using classification techniques. These studies have reached many of the results that we list as follows: center dot Proper selection of attributes plays an important role in enhancing and increasing the accuracy of classification systems, especially as the same classifications have been used to determine disease-specific attributes. center dot Guidance to the need for researchers to choose the right classification for their study where it gives faster and more accurate results. center dot The use of macro-learning for big data, exploration and automation using classification techniques gives more accurate and sensitive results. Based on the foregoing, we found that not all data will serve any purpose without discovering knowledge and that data mining helps to shape the perception of hidden patterns and trends in data sets for diseases that may not have been known before, as the results of the survey showed that the classification techniques used to predict heart disease were as follows: (Naif Bays 84%, hybrid classification techniques using a total of 87.4% classification techniques, then the random forest 88.7%, and finally the most accurate percentage by study is the decision tree technique that gave 99.2%), and the classification techniques with diabetes disease came as follows: (The synthetic neural network is the most accurate and sensitive 98.4%, followed by the closest neighbors, support vectors, Nayef Baez, and finally the decision tree), and classification techniques to predict diseases in general: On the other hand, (the neural network had an accuracy of 84.5%. The most commonly used classifications were support vectors, followed by Naif Bays technology). These classifications lead to clear and correct decisions that benefit the economy, health, and all areas of service. Indeed, in light of its blessed Vision 2030, the Kingdom of Saudi Arabia has achieved qualitative leaps in the use of classification applications. Sehaty, Tawakkolna, and Taba'd from the first apps that have been creative in predicting diseases through the use of mobile phones loaded with those smart apps.