B0833
Title: Machine learning methods for survival analysis to predict dementia risk in diabetic patients
Authors: Aamna AlShehhi - Khalifa University (United Arab Emirates) [presenting]
Abstract: Dementia is an insidious, progressive, and degenerative neurodegenerative disease that destroys normal brain functionality. It targets the elderly, although it is not part of the normal ageing process. Disease symptoms start with memory loss and language problems that progress over time to losing the ability to carry on normal daily activities. At the later stage, the patient becomes bed-bound and requires around-the-clock care. According to the World Health Organization (WHO), approximately 50 million people worldwide are diagnosed with dementia, with nearly 10 million new cases annually. Dementia has a physical, emotional, financial, and economic burden on the patient as well as on society, families, and caregivers. According to WHO, the estimated global community cost of dementia caring was US$\$$ 818 billion in 2015. On the bright side, dementia can be delayed or prevented by diagnosing it in its early stage. That is why we assessed the performance of different machine learning for survival analysis methods combined with various feature selection methods by the concordance index (C-Index) to predict patients at the risk of developing dementia. In the presented study, we developed a stable predictive model for early-stage dementia prediction using tree-based methods.