Selecting Optimal Long Short-Term Memory (LSTM) Architectures for Online Estimation of Mean Arterial Pressure (MAP) in Patients Undergoing General Anesthesia
Source abstract
Abstract
In the realm of anesthetic management during surgical procedures, the reliable estimation of mean arterial pressure (MAP) is critical for ensuring patient safety and optimizing drug administration. This paper investigates the determination of the optimal Long Short-Term Memory (LSTM) architectures aimed at enhancing the estimation of MAP. Using data from a trial involving 70 patients undergoing Total Intravenous Anesthesia (TIVA) provides the effect-site concentrations of Propofol and Remifentanil as key input variables for LSTM models. Our solution categorizes the selection strategies into three distinct methodologies: (i) a population-based method applying a single model across all patients, (ii) a patient-specific method tailoring models to individual physiological responses, and (iii) a novel category-specific method that groups patients based on the correlation between input variables, the effect-site concentrations of Propofol and Remifentanil, and MAP output. The novelty of this paper lies in the proposed method to identify the optimal architecture, evaluating 288 models to fine-tune the best model for each patient and category. Our findings suggest that the patient-specific model outperforms others, highlighting the benefits of personalized model architectures in medical artificial intelligence (AI) applications. The category-specific models provide a pragmatic solution, with reasonable accuracy and enhanced computational efficiency. By contrast, the population-based models, while efficient, have a lower estimation accuracy. This study confirms the significance of sophisticated LSTM architectures in medical AI, providing insights into their potential for advancing patient-specific anesthetic care by accurately online estimating MAP.
Source: crossref · Quality: baseline
Research context
OpenAlex topics: Hemodynamic Monitoring and Therapy · Optical Imaging and Spectroscopy Techniques · Non-Invasive Vital Sign Monitoring
SciVX discovery tags are terminology-based. OpenAlex topics are source-provided classifications.
Affiliations
Research Group on Dynamical Systems and Control, Department of Electromechanics, Systems and Metal Engineering, Ghent University, Tech Lane Science Park 125, 9052 Ghent, Belgium
Department of Anesthesia, Ghent University Hospital, Corneel Heymanslaan 10, 9000 Ghent, Belgium
Funding
Ghent University · Flanders Research Foundation
References (39)
Show source-provided references
1. Comparison between hemodynamic effects of propofol and thiopental during general anesthesia induction with remifentanil infusion: A double-blind, age-stratified, randomized study10.1007/s00540-019-02657-x ↗
2. Current practices and goals for mean arterial pressure and spinal cord perfusion pressure in acute traumatic spinal cord injury: Defining the gaps in knowledge10.1080/10790268.2019.1660840 ↗
3. Cava, C., Salvatore, C., and Castiglioni, I. (2023). Pan-cancer classification of gene expression data based on artificial neural network model. Appl. Sci., 13.10.3390/app13137355 ↗
4. Zhang, C., Zhang, L., Tian, Y., Bao, B., and Li, D. (2023). A machine-learning-algorithm-assisted intelligent system for real-time wireless respiratory monitoring. Appl. Sci., 13.10.3390/app13063885 ↗
5. Artificial intelligence for pain classification with the non-invasive pain monitor Anspec-Pro10.56126/73.S1.29 ↗
6. Shamshirband, S., Fathi, M., Dehzangi, A., Chronopoulos, A.T., and Alinejad-Rokny, H. (2021). A review on deep learning approaches in healthcare systems: Taxonomies, challenges, and open issues. J. Biomed. Inform., 113.10.1016/j.jbi.2020.103627 ↗
7. Real-time regional seismic damage assessment framework based on long short-term memory neural network10.1111/mice.12628 ↗
8. Shah, S.Y., Yuan, Z., Lu, S., and Zerfos, P. (2017, January 11–14). Dependency analysis of cloud applications for performance monitoring using recurrent neural networks. Proceedings of the 2017 IEEE International Conference on Big Data (Big Data), Boston, MA, USA.10.1109/BigData.2017.8258087 ↗
9. Recurrent neural networks for time series forecasting: Current status and future directions10.1016/j.ijforecast.2020.06.008 ↗
10. Ionescu, C.M., Keyser, R.D., Copot, D., Yumuk, E., Ynineb, A., Othman, G.B., and Neckebroek, M. (2024). Model Extraction From Clinical Data Subject to Large Uncertainties and Poor Identifiability. IEEE Control Syst. Lett.10.1109/LCSYS.2024.3402942 ↗
11. Pham, T.D. (2021). Time–frequency time–space LSTM for robust classification of physiological signals. Sci. Rep., 11.10.1038/s41598-021-86432-7 ↗
12. Heynen, J., Copot, D., Ghita, M., and Ionescu, C.M. (2021, January 20–23). Using convolutional neural network online estimators for predicting pain-level variability enables predictive control of anesthesia. Proceedings of the 2021 25th International Conference on System Theory, Control and Computing (ICSTCC), Iași, Romania.10.1109/ICSTCC52150.2021.9607270 ↗
13. PSO–LSTM for short term forecast of heterogeneous time series electricity price signals10.1007/s12652-020-02353-9 ↗
14. End-to-end CNN+ LSTM deep learning approach for bearing fault diagnosis10.1007/s10489-020-01859-1 ↗
15. The influence of method of administration and covariates on the pharmacokinetics of propofol in adult volunteers10.1097/00000542-199805000-00006 ↗
16. Influence of age and gender on the pharmacokinetics and pharmacodynamics of remifentanil: I. Model development10.1097/00000542-199701000-00004 ↗
17. Karnika, S. (2022). ISO 14155: Clinical Investigation of Medical Devices for Human Subjects. Medical Device Guidelines and Regulations Handbook, Springer.
18. Global, S.P., Windisch, F., Zimmermann, N., Knoll, V., Christodoulou, M., Habimana, K., Piccoli, G., and Del Prete, M. (2024). Study on the Implementation of Article 17 of Regulation
19. (EU) 2017/745 on Medical Devices on the EU Market, European Union.
20. Ionescu, C.M., Copot, D., Yumuk, E., De Keyser, R., Muresan, C., Birs, I.R., Ben Othman, G., Farbakhsh, H., Ynineb, A.R., and Neckebroek, M. (2024). Development, Validation, and Comparison of a Novel Nociception/Anti-Nociception Monitor against Two Commercial Monitors in General Anesthesia. Sensors, 24.10.3390/s24072031 ↗
21. A heterogeneous network embedded medicine recommendation system based on LSTM10.1016/j.future.2023.07.004 ↗
22. Kaushik, S., Choudhury, A., Sheron, P.K., Dasgupta, N., Natarajan, S., Pickett, L.A., and Dutt, V. (2020). AI in healthcare: Time-series forecasting using statistical, neural, and ensemble architectures. Front. Big Data, 3.10.3389/fdata.2020.00004 ↗
23. Goodfellow, I., Bengio, Y., and Courville, A. (2016). Deep Learning, MIT Press.
24. A survey on LSTM memristive neural network architectures and applications10.1140/epjst/e2019-900046-x ↗
25. Modified gate activation functions of Bi-LSTM-based SC-FDMA channel equalization
26. EEG signal classification using LSTM and improved neural network algorithms10.1007/s00500-019-04515-0 ↗
27. A review of recurrent neural networks: LSTM cells and network architectures10.1162/neco_a_01199 ↗
28. Beale, M.H., Hagan, M.T., and Demuth, H.B. (2018). Deep Learning Toolbox, The MathWorks, Inc.. User’s Guide.
29. How does weight correlation affect generalisation ability of deep neural networks?
30. Chen, J., Liu, Z., Yin, Z., Liu, X., Li, X., Yin, L., and Zheng, W. (2023). Predict the effect of meteorological factors on haze using BP neural network. Urban Clim., 51.10.1016/j.uclim.2023.101630 ↗
31. A novel long-term solar photovoltaic power forecasting approach using LSTM with Nadam optimizer: A case study of India10.1002/ese3.1178 ↗
32. Chang, Z., Zhang, Y., and Chen, W. (2018, January 23–25). Effective adam-optimized LSTM neural network for electricity price forecasting. Proceedings of the 2018 IEEE 9th International Conference on Software Engineering and Service Science (ICSESS), Beijing, China.10.1109/ICSESS.2018.8663710 ↗
33. Towards activation function search for long short-term model network: A differential evolution based approach
34. Single layer & multi-layer long short-term memory (LSTM) model with intermediate variables for weather forecasting10.1016/j.procs.2018.08.153 ↗
35. Abbasimehr, H., Shabani, M., and Yousefi, M. (2020). An optimized model using LSTM network for demand forecasting. Comput. Ind. Eng., 143.10.1016/j.cie.2020.106435 ↗
36. An Open source patient simulator for design and evaluation of computer based multiple drug dosing control for anesthetic and hemodynamic variables10.1109/ACCESS.2021.3049880 ↗
37. Performance of the hypotension prediction index with noninvasive arterial pressure waveforms in awake cesarean delivery patients under spinal anesthesia10.1213/ANE.0000000000005754 ↗
38. Kandil, H., Soliman, A., Alghamdi, N.S., Jennings, J.R., and El-Baz, A. (2023). Using mean arterial pressure in hypertension diagnosis versus using either systolic or diastolic blood pressure measurements. Biomedicines, 11.10.3390/biomedicines11030849 ↗
39. Khanna, A.K., Kinoshita, T., Natarajan, A., Schwager, E., Linn, D.D., Dong, J., Ghosh, E., Vicario, F., and Maheshwari, K. (2023). Association of systolic, diastolic, mean, and pulse pressure with morbidity and mortality in septic ICU patients: A nationwide observational study. Ann. Intensive Care, 13.10.1186/s13613-023-01101-4 ↗