Predicting the Kidney Graft Survival Using Optimized African Buffalo-Based Artificial Neural Network

Area of Science:

• Nephrology
• Artificial Intelligence
• Medical Informatics

Background:

• Kidney graft survival is influenced by various donor and recipient factors.
• Accurate prediction of transplant effectiveness is vital for optimizing organ allocation and patient management.
• Previous studies on graft rejection predictors yielded contradictory results.

Purpose of the Study:

• To identify predictive risk variables for kidney graft survival using a novel hybrid feature selection technique.
• To develop an accurate prediction model for kidney transplant outcomes.
• To compare the proposed model's performance against existing classification methods.

Main Methods:

• Utilized an African buffalo-based artificial neural network (AB-ANN) approach combined with hybrid feature selection.
• Processed collected clinical data through training and testing methodologies.
• Evaluated prediction model performance using accuracy, precision, recall, and F-measure metrics.

Main Results:

• The hybrid feature selection identified key clinical factors influencing transplant survival.
• The AB-ANN model demonstrated superior accuracy in forecasting kidney graft survival compared to naive Bayesian, random forest, and J48 classifiers.
• The proposed method achieved higher precision, recall, and F-measure scores.

Conclusions:

• The AB-ANN approach offers a novel and accurate method for predicting kidney graft survival.
• This predictive model can aid physicians in making more personalized treatment decisions for kidney transplant recipients.
• Integrating these clinical tools into practice can enhance patient care and transplant outcomes.