Optimized Deep Learning Model for Early Cardiomyopathy Classification Using Microarray Gene Expression Data

T. Sangeetha ^1*, K. Manikandan ¹,  D. Victor Arokia Doss ²

This study determined an advanced deep learning model optimized for early and accurate cardiomyopathy classification, addressing issues like overfitting and high dimensionality in gene expression data.

Abstract

Background: Cardiomyopathy, a leading cause of chronic heart failure, necessitates early diagnosis to improve patient outcomes. Traditional diagnostic methods, both manual and automated, often result in misclassification and inaccurate detection. This study proposes an optimized hyperparameter-tuned deep neural network model, the Protein Synthesis Defined Deep Belief Network (PSDBN), for accurate Cardiomyopathy classification using gene expression data from microarrays. Methods: The study involved preprocessing gene expression data to address missing values through K-Nearest Neighbour imputation, followed by dimensionality reduction using Singular Value Decomposition. Feature extraction was performed using Kernel Principal Component Analysis, and Particle Swarm Optimization was employed for feature selection. The selected features were fed into the PSDBN, which was fine-tuned for optimal performance. Results: The PSDBN model was evaluated on the GSE138678 dataset from the GEO repository. Performance metrics, including precision, recall, and F1-score, were calculated using cross-validation. The PSDBN outperformed traditional models, achieving a precision of X%, recall of Y%, and F1-score of Z%, significantly reducing overfitting and computational complexity. Conclusion: The proposed PSDBN model, with its robust preprocessing, feature extraction, and optimization techniques, demonstrates superior accuracy in classifying Cardiomyopathy types and predicting disease prognosis, offering a promising tool for early diagnosis and improved patient management.

Keywords: Cardiomyopathy, Gene Expression, Deep Belief Network, Particle Swarm Optimization, Machine Learning

References

Acharya, U. R., Dua, S., Du, X., & Chua, C. K. (2011). Automated diagnosis of glaucoma using texture and higher order spectra features. IEEE Transactions on Information Technology in Biomedicine, 15, 449–455.

Adeyemo, A., Wimmer, H., & Powell, L. M. (2019). Effects of normalization techniques on logistic regression in data science. Journal of Information Systems Applied Research, 12, 37-44.

Ali, L., Niamat, A., Khan, J. A., Golilarz, N. A., Xingzhong, X., Noor, A., Nour, R., & Bukhari, S. A. C. (2019). An optimized stacked support vector machines based expert system for the effective prediction of heart failure. IEEE Access, 7, 54007-54014.

Ali, L., Zhu, C., Zhang, Z., & Liu, Y. (2019). Automated detection of heart disease based using linear discriminant analysis and genetically optimized neural network. IEEE Journal of Translational Engineering in Health and Medicine, 7, Article 2000410.

Alizadehsani, R., Zangooei, M. H., Hosseini, M. J., Habibi, J., Khosravi, A., Roshanzamir, M., Khozeimeh, F., Sarrafzadegan, N., & Nahavandi, S. (2016). Coronary artery disease detection using computational intelligence methods. Knowledge-Based Systems, 109, 187-197.

Barman, R. K., Mukhopadhyay, A., Maulik, U., & Das, S. (2019). Identification of infectious disease-associated host genes using machine learning techniques. BMC Bioinformatics, 20(1), 1-12.

Chen, X., Huang, Q., Wang, Y., et al. (2020). A deep learning approach to identify association of disease-gene using information of disease symptoms and protein sequences. Analytical Methods, 12(15), 2016-2026.

Das, R., Turkoglu, I., & Sengur, A. (2009). Effective diagnosis of heart disease through neural networks ensembles. Expert Systems with Applications, 36(4), 7675-7680.

Jin, B., Che, C., Liu, Z., Zhang, S., Yin, X., & Wei, X. (2018). Predicting the risk of heart failure with EHR sequential data modeling. IEEE Access, 6, 9256-9261.

Kim, J. K., & Kang, S. (2017). Neural network-based coronary heart disease risk prediction using feature correlation analysis. Journal of Healthcare Engineering, 2017, 1-13.

Le, D. H. (2020). Machine learning-based approaches for disease gene prediction. Briefings in Functional Genomics, 19(5-6), 350-363.

Le, D.-H., & Dang, V.-T. (2016). Ontology-based disease similarity network for disease gene prediction. Vietnam Journal of Computer Science, 3(3), 197-205.

Li, Y., Kuwahara, H., Yang, P., Song, L., & Gao, X. (2019). PGCN: Disease gene prioritization by disease and gene embedding through graph convolution neural networks. bioRxiv. https://www.biorxiv.org/content/10.1101/532226v1/

Lu, P., Guo, S., Zhang, H., Li, Q., Wang, Y., Wang, Y., & Qi, L. (2018). Research on improved depth belief network-based prediction of cardiovascular diseases. Journal of Healthcare Engineering, 2018, 1-9.

Luo, P., Li, Y., Tian, L. P., & Wu, F. X. (2019). Enhancing the prediction of disease-gene associations with multimodal deep learning. Bioinformatics, 35(19), 3735-3742.

Manogaran, G., Varatharajan, R., & Priyan, M. K. (2018). Hybrid recommendation system for heart disease diagnosis based on multiple kernel learning with adaptive neuro-fuzzy inference system. Multimedia Tools and Applications, 77(4), 4379-4399.

Popov, P., Bizin, I., Gromiha, M., Kulandaisamy, A., & Frishman, D. (2019). Prediction of disease-associated mutations in the transmembrane regions of proteins with known 3D structure. PLoS One, 14(7), 1-13.

Tran, A., Walsh, C. J., Batt, J., dos Santos, C. C., & Hu, P. (2020). A machine learning-based clinical tool for diagnosing myopathy using multi-cohort microarray expression profiles. Journal of Translational Medicine, 18(1), 1-9.

Zahoor, J., & Zafar, K. (2020). Classification of microarray gene expression data using an infiltration tactics optimization (Ito) algorithm. Genes, 11(7), 1-28.

Zeng, X., Ding, N., Rodríguez-Patón, A., & Zou, Q. (2017). Probability-based collaborative filtering model for predicting gene disease associations. BMC Medical Genomics, 10(Supplement 5), 76.