In recent years, brain-related illnesses like Parkinson's disease have gained increasing attention due to the economic strain caused by healthcare expenses associated with these diseases in developed nations. Parkinson's disease (PD) is a neurodegenerative disorder that affects nerve cells responsible for producing dopamine, a neurotransmitter in the brain. A recent study utilized the recursive feature elimination (RFE) algorithm in four academic papers to select features and apply machine learning techniques. These techniques included direct and indirect separation methods, as well as decision tree and potential separator approaches. The study evaluated the performance of each category using five selected features identified through the RFE approach. Notably, the indirect category, specifically random forest and bagging cart, demonstrated excellent performance with an accuracy of 96.93% and 97.43%, respectively. This analysis aids physicians in effectively classifying neurodegenerative diseases by utilizing gait symptoms to differentiate them from healthy individuals.
TopIntroduction
Neurodegenerative diseases, particularly in countries experiencing rapid aging, pose significant challenges. These disorders have a profound impact on neurons in the brain, leading to their degeneration and eventual death, resulting in movement or cognitive impairments. Diseases such as Alzheimer's and Parkinson's disease (PD) are among the most common neurodegenerative disorders.
The economic burden of these diseases is substantial, primarily due to their prolonged duration and increasing levels of disability. Dementia, including Alzheimer's disease, has a particularly heavy global economic impact, with estimated costs reaching 800 million in 2015.
Alzheimer's disease (AD) accounts for 60-70% of dementia cases and is a progressive and irreversible disorder that primarily affects memory and cognitive abilities, eventually rendering individuals unable to perform even simple tasks. It ranks as the 6th leading cause of death in the United States, but recent reports suggest it may climb to the 3rd position. While there is currently no cure for AD, early detection plays a crucial role in improving patients' lives. Additionally, misdiagnosis can lead to the prescription of medications that can be highly debilitating or have psychological side effects.
Parkinson's disease (PD) is another neurodegenerative disorder that affects the nerve cells responsible for producing dopamine, a neurotransmitter in the brain. PD gradually damages the nervous system, causing symptoms such as tremors, bradykinesia, muscle stiffness, and postural instability. It is estimated that 3 to 4 out of every 1,000 people are affected by PD. Similar to AD, there is no known cure for PD, but medications can help improve patients' quality of life. Early diagnosis is important in reducing the long-term treatment costs associated with PD. Complex neuroimaging techniques can aid in diagnosis. However, studies have shown that about 25% of PD patients remain undiagnosed during the early stages. These patients do not receive the benefits of anti-parkinsonian drugs and must endure the side effects and financial burden of unnecessary prescriptions.
Machine learning is a sub-field of artificial intelligence (AI) focused on understanding the structure of data and developing models that can be used and interpreted by humans. While it falls within the realm of computer science, machine learning differs from traditional algorithms. Instead of being programmed with explicit commands, machine learning algorithms allow computers to train themselves on input data and use mathematical analysis to extract meaningful patterns within a given range. As a result, machine learning enables computers to build models from sample data and make automated decisions based on new input data.
Machine learning is a progressive and continuously evolving field. Consequently, it is important to consider certain factors when working with machine learning methods or assessing the impact of machine learning processes. Figure 1 illustrates the architecture of machine learning.
Figure 1. Machine learning architecture
TopLiterature Survey
During the literature survey, multiple research papers were reviewed to gather information on disease detection methods, specifically focusing on the Yolo Classification approach. Data mining, a widely used technique for extracting patterns and information from large datasets, was found to be essential in this research field.
Numerous existing projects and surveys were explored to gather references and insights. These projects emphasized addressing complex queries, accuracy results, compactness, connectedness, and high-speed time consumption. Each project had its strengths, such as ease of model interpretation, access to detailed information, and accuracy. Once the project idea was developed, the search for existing projects commenced, and several surveys provided valuable references. The following are the processes and steps involved in the power prediction process of existing research: