Abstract
Abstract Introduction Alzheimer’s Disease (AD) is one of the most common neurological diseases, affecting about 50 million people worldwide. Despite its prevalence and many promising recent developments, the factors that contribute to AD are still poorly understood. In this study, we examined the influence of sleep disturbances (sleep apnea, insomnia, and poor sleep quality) on the chances of developing AD and conversion from normal cognitive functioning (NL), to mild cognitive impairment (MCI) to AD, using the ADNI dataset and state-of-the-art machine learning models. Methods We conducted a series of Random Tree classifiers to determine whether sleep disturbances predicted changes in cognitive status (NL, MCI, or AD). Using three sleep target variables OSA, insomnia, and Neuropsychiatric Inventory’s sleep related questions (NPIK) markers, participants that were surveyed for sleep patterns (n=1848) self-declared if they have insomnia (n=49), obstructive sleep apnea (OSA) (n=65), or poor sleep quality via NPIK (n=114). NPIK contains 11 variables, including OSA and NPIK, we explored a total of 7089 observations. Following that, a logistic regression and k-nearest neighbor (KNN) model using only the NPIK questionnaire as NPIK provides a rich and coherent inventory of sleep parameters. Results Based on Random Tree classifiers, the best model predicted a change from MCI to AD with an accuracy of 70%. Logistic regression and KNN models predict that caregiver distress, frequency of sleep disturbances and excessive daytime sleepiness play important roles in worsening the cognitive functioning status from MCI to AD. For the MCI patients, a random forest model with 60% accuracy predicted re-conversion from MCI to NL. Caregiver distress, frequency of sleep disturbances, and OSA predicted re-conversion from MCI-NL, and was most pronounced among males. Conclusion Our research broadly suggests that sleep disturbances increases Alzheimer's disease risk. However, the data on sleep in the ADNI dataset has limitations. All markers for sleep were subjectively reported by the patients or their caregivers and were not verified with objective observations. Further research under clinical conditions would be of value to find conclusive evidence of sleep disturbance as a risk factor for Alzheimer's Disease. Support (if any)