لخّصلي

Machine learning (ML) is defined as the study of computer programs that leverage algorithms and statistical models to learn through inference and patterns without being explicitly programmed [1]. ML algorithms learn over experience and improve automatically. It finds techniques, trains models, and uses the learned approach to determine the output automatically [2]. Machine learning systems can also adjust themselves to a changing environment.
A model is a machine learning system that has been trained to identify specific types of patterns using an
algorithm in a machine learning system [3]. That means it processes the data and finds out the hidden structures in a dataset [4]. The feature extraction and the known answers of a dataset determine the formula that relies upon the input and output functions and applies it to new data to predict the response [5]. Hence, the model’s algorithm uses a collection of data for training and builds a way to predict the output and saves that procedure for future purposes.
A support vector machine (SVM) is a supervised ma- chine learning model that uses classification algorithms for two-group classification problems. Support vector machine is a fast and dependable classification algorithm that performs very well with a limited amount of data to analyze [6]. SVMs are a group of similar supervised learning techniques that are used for classification and regression problems [7].
The logistic regression model is the appropriate re- gression analysis. Logistic regression is predictive regression analysis [8]. To classify data and to illustrate the relationship between one dependent binary variable and one or more independent nominal, ordinal, interval, or ratio-level vari- ables, logistic regression is used [9].
In a machine learning system, a decision tree algorithm partitions the data into subsets. A decision tree’s purpose is to sum up the training data in the smallest tree possible [10]. The decision tree is a supervised classification method that carries out a split test in its internal node and forecasts an example target class in its leaf node [11]. Decision tree al- gorithms are used to classify the characteristics to be eval- uated at any node to specify the “best” splitting [12]. Decision trees are commonly used in classification problems because of their versatility and consistency.
The random forest is a supervised learning algorithm. Random forest is a versatile, easy-to-use machine learning algorithm that provides, most of the time, a fantastic result even without hyperparameter tuning [13]. Its simple design and variety are also some of the most used algorithms [14].
SVM can be applied to nonlinear problems, whereas lo- gistic regression can only work with linear ones. SVM operates outliers better, as it derives maximum margin solution. De- cision trees are better at dealing with collinearity than logistic regression. For categorical values, decision trees outperform logistic regression. A random forest is a set of decision trees that are randomly generated, and the expected output is chosen by the forest’s majority vote. Decision trees are less reliable and accurate than random forest. SVM solves nonlinear issues using kernel methods, whereas decision trees apply hyper- rectangles in input space to solve the problem. For a classifi- cation problem, SVM performs better than random forest [15].
Machine learning models are now widely used in medical diagnosis [16–19]. This paper compares different machine learning performances to diagnose Alzheimer’s syndrome. Alzheimer’s syndrome is an inherited, irrevers- ible brain condition that steadily affects the ability to per- form the necessary things, memory, and reasoning skills [20]. A massive proportion of neurons stop working in Alzheimer’s disease, losing synaptic connections [21]. Alz- heimer’s diseases are infrequent in people aged between their 30s and mid-60s [22]. Symptoms can include a shift in sleep habits, depression, anxiety, and difficulties doing basic tasks such as reading or writing and aggressive actions, and poor decision-making also happened in Alzheimer’s disease [23]. Alzheimer’s disease and initial changes in the brain begin 10–20 years before the onset of symptoms [24]. It progressively leads to memory damage and decreases thinking abilities [25]. The leading cause of this disease is Dementia. A report shows that around 40–50 million people worldwide are suffering from Dementia, and this number will be increased to around 131.5 million by 2050 [26]. Approximately 70% of people who have Dementia are from low-income countries; see Figure 1.
Dementia is the failure of brain function, understanding, recognizing, thinking, and behavioral skills to such a level that an individual faces problems in everyday life and be- haviors [28]. Few people with Dementia are unable to deal with their emotions, and their personalities can be changed [29]. From the mildest stage, Dementia varies in severity [30]. It mainly affects older people. No cure is available other than treatment [31].
There is little data available on Alzheimer’s patients in Bangladesh. According to the WHO data published in 2017, Alzheimer’s or Dementia deaths in Bangladesh reached 9,917 or 1.26% of the total deaths, which was the last data found in this aspect that ranks Bangladesh number 152 globally [32]. In Bangladesh, the awareness about Alz- heimer’s is now in the primary stage. Therefore, impacted patients and families are regularly experiencing various is- sues [33]. The fund for researching Alzheimer’s is limited. A lower-middle-income country like Bangladesh is not yet prepared for the management of Alzheimer’s [34]. Besides that, almost one-fifth of the Bangladeshi adult population is overweight, according to a global study [35], which is the leading risk factor for Alzheimer’s. Therefore, there are more chances of occurrences of Alzheimer’s [36]. To give a treatment for this disease, physicians tend to test individuals for Alzheimer’s disease by obtaining a medical and family history and psychiatric history from the point of view of specialists such as neurologists, neuropsychologists, geria- tricians, and geriatric psychiatrists [37].
Studies show that the situation may improve if people can detect Alzheimer’s disease early by taking therapy at the initial stage [38]. For this, they have to predict the progress of the disease accurately from mild condition to Dementia. Machine learning technology can help to predict accurately early Alzheimer’s disease. There are many machine learning systems, but they give inconsistent and inaccurate predic- tions. They also have overfitting and underfitting issues. Therefore, a model has been developed by us which can indicate Alzheimer’s disease early, using machine learning to support medical technicians. It will verify and show if anyone has Alzheimer’s disease or not.
The remainder of the paper is organized as follows: Section 2 discusses methods and methodology, and Section 3 provides the results and analysis. Finally, in Section 4, the conclusion of the presented work is provided.