Amyotrophic lateral sclerosis (ALS) is a terminal neuromuscular disease in which motor neurons steadily disappear causing muscle atrophy and eventual loss of movement control. Statistical predictive validity of traditional diagnostic modalities in determining the prognosis remains low. This work further proposed a new deep learning model to examine clinical and genetic data. Correcting missing values and normalization were employed in data preprocessing section. The results were assessed with k-fold cross-validation; the chosen metrics were accuracy, precision, recall, and F1 score. Several strategies including data augmentation and the tuning of the hyperparameters were used to improve model durability. The developed algorithm received an accuracy of about 94.28%, with specific to true positive of 0.9256, and sensitivity of 0.8975 for predicting a high-risk population for rate of disease advancement. The variable importance analysis showed that key predictors of outcomes were Age of onset and Initial functional assessment scores. The study provided evidence of the application of deep learning for enhancing the prognostic information of ALS and called for cross-disciplinary strategy to enhance the diagnosis and management of rare diseases.