Application of Bayesian classification for clinical early diagnosis of dementia

• Main Authors: LIAO, YU-MIN, 廖育民
• Other Authors: HORNG, TZYY-LENG
• Format: Others
• Language: zh-TW
• Published: 2017
• Online Access: http://ndltd.ncl.edu.tw/handle/fsh33j

碩士 === 逢甲大學 === 應用數學系 === 105 === Dementia is a progressive degeneration of cognitive function due to the damage or disease in the brain. Besides, the rate of degeneration is much higher than the one of normal aging. Dementia is not a single disease, but a combination of a group of symptoms (syndrome). Affected cognitive functions focus mainly on the areas of memory, orientation, judgment, abstract thinking, attention, and language. Additionally, it may be also accompanied with behavioral disturbances, personality changes, paranoia, or hallucination. If the symptoms are severe, then the function and performance of daily lives will be affected. Dementia can be classified as either reversible or irreversible, depending upon the etiology of disease. It is about less than 10％ of dementia that can be reversed by treatment. Although dementia is far more common in the elderly population (about 5％ of those over 65 years), but it is not the patent of the elderly. Many patients with familial or genetic history have early onset before ages 65, termed early-onset dementia. Such patients, mostly responsible for the family’s economy, have onset in the prime of life. Once the disease occurs, along with the cognition degeneration, this will affect the patients’ ability to work. The impact will be more enormous, compared to older patients. This study will introduce Bayesian classification. In order to construct a viable model of clinical early diagnosis for dementia, the risk factors, such as gender, age, education level, hypothyroidism, hypertension, heart disease, diabetes, and cerebral vascular accident, will be considered for dementia. Additionally, MMSE (Mini-mental state examination), commonly used in medicine to screen for dementia, is also included. Total 169 subjects (male: 87, female: 82) were recruited in this research. Using the above factors, we will develop an appropriate bayesian prediction model for dementia, and verify the reliability and validity of the model. In this study, CDR (clinical dementia rating scale) will serve as the target of prediction. With this study, a forecast in CDR will serve as a preliminary diagnosis of dementia. In the future, we will collect more sample data, which will also include dementia-related genetic type, etc., and do the training and validation of predictive model. To achieve the goal of early prevention and improve the quality of patients’ life, we will develop the software system of monitoring and preventing the occurrence of dementia.