Аннотации:
Genome-wide association studies are aimed at identifying associations between commonly occurring variations in a group of individuals and a phonotype, in which the Deoxyribonucleic acid is genotyped in the form of single nucleotide polymorphisms. Despite the exsistence of various research studies for the prediction of chronic diseases using human genome data, more investigations are still required. Machine learning algorithms are widely used for prediction and genome-wide association studies. In this research, Random Forest was utilised for selecting most significant single nucleotide polymorphisms associated to Alzheimer’s Disease. Deep learning model for the prediction of the disease was then developed. Our extesnive similation results indicated that this hybrid model is promising in predicting individuals that suffer from Alzheimer’s disease, achieving area under the curve of 0.9 and 0.93 using Convolutional Neural Network and Multilayer perceptron respectively.