Severe Complication Prediction Model Construction Combined with Statistical and Machine Learning Algorithms for Patient with Diabetes Related Fatigue, Sleep and Lifestyle

Fatigue, poor sleep quality and frailty are the common chief complaints of people with diabetes. Fatigue is a subjective symptom without specific characteristics. Though not an emergency or life threatening, it will affect the patient's daily activity and self-management ability. Lack of a solid definition and clinical management, it is still a trouble for people with diabetes. Previous studies have shown the correlation between glycemic control and fatigue, so the complexity among variables should be clarified.. However, there were neither enough large studies for the survey of fatigue and its risk factors in Taiwan, nor the relation among fatigue, sleep quality, lifestyle and severe comorbitidies. Therefore, the purpose of this study is to achieve accurate analysis and construct the best predictive model through machine learning algorithms based on data mining. A diabetes bioinformatics data set includes the information of subjective feeling of fatigue, sleep quality, lifestyle and treatment effects from 400 patients with type 2 diabetes from questionnaire survey and retrospective chart review. To explore the relationship among diabetes related fatigue, sleep quality and lifestyle under the data miming of feature selection and case selection, and regression analysis from above bioinformatics databases. The final output is to develop the best predictive model by the occurrence of severe complications including nephropathy, retinopathy and neuropathy as the effectiveness of diabetes control.

Project ID:PC10907-0997
External Project ID:MOST109-2314-B182-048