Details

Autor(en) / Beteiligte

• Chaplot, Neelam
• Pandey, Dhiraj
• Kumar, Yogesh
• Sisodia, Pushpendra Singh

Titel

A Comprehensive Analysis of Artificial Intelligence Techniques for the Prediction and Prognosis of Genetic Disorders Using Various Gene Disorders

Ist Teil von

• Archives of computational methods in engineering, 2023-06, Vol.30 (5), p.3301-3323

Ort / Verlag

Dordrecht: Springer Netherlands

Erscheinungsjahr

2023

Quelle

Alma/SFX Local Collection

Beschreibungen/Notizen

• A medical analysis of diagnosing rare genetic diseases has rapidly become the most expensive and time-consuming component for doctors. By combining predictive methods with growing knowledge of genetic disease, artificial intelligence (AI) has the potential to simplify and accelerate genome interpretation greatly. In this paper, multiple machine-learning models like support vector machine, Gaussian Naïve Bayes, KNN, Decision Tree, Gradient Boosting, logistic regression, light gradient boosting classifier, Random Forest, extreme gradient boosting classifier, and cat-boost are applied to the genetic disorder as well as genetic disorder sub-classes datasets. The dataset has been initially pre-processed to check for NAN values, which are graphically represented in various categories like genetic disorder, genetic disorder subclasses, five samples of symptoms, genes inherited from mother’s and father’s side, birth defects etc. to study their pattern. Later, the features have been selected using standardization technique on which the machine learning models are applied and later evaluated using accuracy, loss, recall, precision, root mean square error, and F1 score. Furthermore, the confusion matrix is also generated to compute false negative, true positive, false positive and true negative values for the classes drawn from both datasets. It has been found that the highest accuracy has been calculated by decision tree, random forest, gradient boosting, LGBM classifier, XGB classifier, and CatBoost by 99.9% for genetic disorder while as only the random forest, decision tree, LGBM classifier, and CatBoost, on the other hand, achieved 99.9% accuracy for genetic disorder sub-classes.