ON SENSITIVITY ANALYSIS METHODS IN MEDICAL DECISION-SUPPORT SYSTEMS
DOI:
https://doi.org/10.32689/maup.it.2023.5.1Keywords:
sensitivity analysis, data mining, artificial neural network, integrated gradients, medical diagnosis, decisionmaking.Abstract
Abstract. This article delves into developing the vital part of computerized medical monitoring systems, namely part of the stratification of patient data – methods of identification parameters informativeness. The inherent stochastic nature of data generated by these systems necessitates advanced methods for discerning patient states, often requiring predefined logic or expert intervention. Leveraging Machine Learning methods for data analysis in medical monitoring systems can help uncover complex relationships between data and patient states, ultimately enhancing treatment quality. The study explores a combination of the artificial neural network model with methods for defining data parameter informativeness, providing insights into parameter impact on the model output. The study considered developed gradientbased methods for estimating overall parameters informativeness and modified integrated gradients method for estimating informativeness parameters of specific data. The research employs the UCI Heart Disease Data, a representative dataset mirroring typical patient data in computer medical monitoring systems. Challenges of this data: such as bias, missing values, and high dimensionality underscore the complexity of real-world medical data, posing a significant challenge for the proposed methods. The work showed the performance of the supposed methods and analyzed them by comparing them to PCA variance estimation. The supposed gradient-based method shows high awareness of the parameter importance and consideration of nonlinearities in the data. The integrated-gradients method shows a relation between overall informativeness values and informativeness for specific data. The results will impact the development of decision-supporting systems for computer medical monitoring systems.
References
Logeshwaran, J., Malik, J. A., Adhikari, N., Joshi, S. S., Bishnoi, P. IoT-TPMS: An innovation development of triangular patient monitoring system using medical internet of things. International Journal of Health Sciences. 2022. 6(S5), 9070–9084. DOI: 10.53730/ijhs.v6nS5.10765.
Yu, M., Li, G., Jiang, D., Jiang, G., Tao, B. Chen, D. Hand medical monitoring system based on machine learning and optimal EMG feature set. Personal and Ubiquitous Computing. 2019. 1 – 17.
Humayun, M., Jhanjhi, N.Z., Almotilag, A., Almufareh, M.F. Agent-Based Medical Health Monitoring System. Sensors. Basel, Switzerland. 2022. 22.
Miller, R., Acton, C., Fullerton, D.A., Maltby, J., Campling, J. Analysis of Variance (Anova). The SAGE Encyclopedia of Research Design. 2022.
Cozzi, M., Romano, S., Viccaro, M., Prete, C., Persiani, G. Wildlife Agriculture Interactions, Spatial Analysis and Trade-Off Between Environmental Sustainability and Risk of Economic Damage. 2015.
Chung, H., Ko, H., Kang, W.S., Kim, K.W., Lee, H., Park, C., Song, H., Choi, T., Seo, J.H., Lee, J. Prediction and Feature Importance Analysis for Severity of COVID-19 in South Korea Using Artificial Intelligence: Model Development and Validation. Journal of Medical Internet Research. 2021. 23.
Pereira, J.P., Stroes, E.S., Zwinderman, A.H., Levin, E. Covered Information Disentanglement: Model Transparency via Unbiased Permutation Importance. ArXiv, abs/2111.09744. 2021.
Ekanayake, I.U., Meddage, D.P., Rathnayake, U.S. A novel approach to explain the black-box nature of machine learning in compressive strength predictions of concrete using Shapley additive explanations (SHAP). Case Studies in Construction Materials. 2022.
Wu, Y., Zhou, Y. Hybrid machine learning model and Shapley additive explanations for compressive strength of sustainable concrete. Construction and Building Materials. 2022.
Lundstrom, D., Huang, T., Razaviyayn, M. A Rigorous Study of Integrated Gradients Method and Extensions to Internal Neuron Attributions. ArXiv, abs/2202.11912. 2022.
Enguehard, J. Sequential Integrated Gradients: a simple but effective method for explaining language models. ArXiv, abs/2305.15853. 2023.
Qi, Z., Khorram, S., Li, F. Visualizing Deep Networks by Optimizing with Integrated Gradients. CVPR Workshops. 2019.
Kovacs, I., Iosub, A., Topa, M.D., Buzo, A., Pelz, G. A Gradient-based Sensitivity Analysis Method for Complex Systems. 2019 IEEE 25th International Symposium for Design and Technology in Electronic Packaging (SIITME), 333-338. 2019.
Strilets, V., Bakumenko, N., Chernysh, S., Ugryumov, M., Donets, V. Application of Artificial Neural Networks in the Problems of the Patient’s Condition Diagnosis in Medical Monitoring Systems. Integrated Computer Technologies in Mechanical Engineering. 2020.
В. Є. Стрілець, С. І. Шматков, М. Л. Угрюмов. Методи машинного навчання у задачах системного аналізу і прийняття рішень : монографія / Харків : Харківський національний університет імені В. Н. Каразіна. 2020.
Janosi Andras, Steinbrunn William, Pfisterer Matthias, Detrano Robert. Heart Disease. UCI Machine Learning Repository. 1988.
Pereira, N. Using Machine Learning Classification Methods to Detect the Presence of Heart Disease. 2019.
