An intelligent framework of heuristic approach-aided optimal gene selection and residual LSTM with MLP for disease prediction in rice crop using gene expression data

• Published in: Signal, image and video processing Vol. 19; no. 4; p. 307
• Main Authors: Adluri, Vijaya Lakshmi, Raju Bhukya
• Format: Journal Article
• Language: English
• Published: London Springer London 01.04.2025; Springer Nature B.V
• Subjects: Agriculture; Algorithms; Bioinformatics; Computer Imaging; Computer Science; Crop diseases; Crops; Deep learning; Farming; Gene expression; Heuristic methods; Illnesses; Image Processing and Computer Vision; Literature reviews; Machine learning; Multilayers; Multimedia Information Systems; Neural networks; Optimization; Original Paper; Pattern Recognition and Graphics; Perception; Rice; Signal,Image and Speech Processing; Support vector machines; Trends; Vision; Multilayer perception; Gene expression data; Optimal features selection; Rice crop disease prediction; Residual long short-term memory
• ISSN: 1863-1703; 1863-1711
• DOI: 10.1007/s11760-025-03859-5

The use of gene expression information for disease prediction in the cultivation of rice offers a novel approach to improve agricultural output and health. In order to predict future epidemics, it involves examining how genes are expressed based on different illnesses. Early identification of illness by gene expression might be achieved even before signs appear, allowing for timely treatments to reduce crop loss and stop the propagation of the illness. However, because gene expression information is so complicated and requires advanced quantitative and bioinformatic knowledge, understanding it may be quite difficult. This paper leverages cutting-edge technological solutions by developing a system based on deep learning and machine learning for disease prediction in rice crops using gene expression data. The initial step involves gathering gene expression data from reputable sources to ensure a diverse and comprehensive dataset. Once collected, variable-length gene data are filled by padding to standardize data input formats. Following this, repetitions in gene data are identified using multi-similarity methods, ensuring the uniqueness and relevance of the data for analysis. The obtained data are further used for optimal gene selection using the enhanced red-tailed hawk algorithm (ERTH), which ensures superior performance. Finally, the prediction is performed using an innovative approach named adaptively optimized residual long short-term memory with multilayer perception (AO-RLSTM-MLP) that combines residual long short-term memory (RLSTM) and multilayer perception (MLP). During this phase, the parameters of the AO-RLSTM-MLP are optimally tuned using the same ERTH approach. Experimental analysis is conducted on this deep learning-based model to validate its effectiveness in disease prediction in rice crops using gene expression data. This analysis demonstrates the potential of this robust solution in the field, highlighting its capabilities to predict disease outbreaks effectively and contribute to sustainable agriculture practices.