As there is a rapid growth in healthcare systems and biomedical data. Machine learning algorithms are utilized in many researches for predicting the risk of the diseases. The major intuition of the present paper is to plan for a novel methodology for multi-disease prediction using deep learning. The overall prediction methodology involves several steps such as “(a) Data Acquisition, (b) Optimal Feature selection, (c) Statistical feature Extraction, and (d) prediction”. In the initial step, the medical datasets of diverse diseases is gathered from multiple benchmark sources. Further, the optimal feature selection is applied to the available set of attributes. This is accomplished by hybridizing two meta-heuristic algorithms such as Lion Algorithm (LA), and Butterfly Optimization Algorithm (BOA). In these prediction algorithms, the hidden neuron count of NN and DBN is finely tuned or optimized by the same hybrid Lion-based BOA (L-BOA). The experimental evaluation of various medical datasets validates that the prediction rate of the developed model outperforms several traditional methods.

随着医疗保健系统和生物医学数据的快速增长。机器学习算法被用于许多研究中，以预测疾病的风险。本文的主要直觉是计划使用深度学习进行多病害预测的新方法。总体预测方法包括几个步骤，例如“（a）数据采集，（b）最佳特征选择，（c）统计特征提取和（d）预测”。在第一步中，从多个基准源收集各种疾病的医学数据集。此外，将最佳特征选择应用于可用的属性集。这是通过混合两种元启发式算法（例如Lion算法（LA）和Butterfly Optimization算法（BOA））来实现的。在这些预测算法中 NN和DBN的隐藏神经元计数可以通过相同的基于Lion的混合BOA（L-BOA）进行微调或优化。对各种医学数据集的实验评估证实，所开发模型的预测率优于几种传统方法。