Abstract
The problem of estimating breath alcohol concentration based on transdermal alcohol biosensor data is considered. Transdermal alcohol concentration provides a promising alternative to classical methods such as breathalyzers or drinking diaries. A physics-informed long short-term memory (LSTM) network with covariates for the solution of the estimation problem is developed. The data-driven nature of an LSTM is augmented with a first-principles physics-based population model for the diffusion of ethanol through the epidermal layer of the skin. The population model in an abstract parabolic framework appears as part of a regularization term in the loss function of the LSTM. While learning, the model is encouraged to both fit the data and to produce physically meaningful outputs. To deal with the high variation observed in the data, a mechanism for the uncertainty quantification of the estimates based on a recently discovered relation between Monte-Carlo dropout and Bayesian learning is used. The physics-based population model and the LSTM are trained and tested using controlled laboratory collected breath and transdermal alcohol data collected in four sessions from 40 orally dosed participants (50% female, ages 21–33 years, 35% BMI above 25.0) resulting in 256 usable drinking episodes partitioned into training and testing sets. Body measurement (e.g., BMI, hip to waist ratio, etc.), personal (e.g., sex, age, race, etc.), drinking behavior (e.g., frequent, rarely, etc.), and environmental (e.g., temperature, humidity, etc.) covariates were also collected from participants. The importance of various covariates in the estimation is investigated using Shapley values. It is shown that the physics-informed LSTM network can be successfully applied to drinking episodes from both the training and test sets and that the physics-based information leads to better generalization ability on new drinking episodes with the uncertainty quantification yielding credible bands that effectively capture the true signal. Compared to two machine learning models from previous studies, the proposed model reduces relative \(L_2\) error in estimated breath alcohol concentration by 58 and 72% and relative peak error by 33 and 76%.
Similar content being viewed by others
Data availability
The datasets generated during and/or analyzed during the current study are not publicly available due to restrictions on human subject data, but are available from the corresponding author on reasonable request.
References
Caddy G, Sobell M, Sobell L (1978) Alcohol breath tests: criterion times for avoiding contamination by “mouth alcohol’’. Behav Res Methods Instrum 10:814–818
Lister RG, Gorenstein C, Risher-Flowers D, Weingartner HJ, Eckardt MJ (1991) Dissociation of the acute effects of alcohol on implicit and explicit memory processes. Neuropsychologia 29(12):1205–1212. https://doi.org/10.1016/0028-3932(91)90034-6
Sakai JT, Mikulich-Gilbertson S, Long RJ, Crowley TJ (2006) Validity of transdermal alcohol monitoring: fixed and self-regulated dosing. Alcohol Clin Exp Res 30:26–33. https://doi.org/10.1111/j.1530.0277.2006.00004.x
Gamella M, Campuzano S, Manso J, de Rivera GG, López-Colino F, Reviejo AJ, Pingarrón JM (2014) A novel non-invasive electrochemical biosensing device for in situ determination of the alcohol content in blood by monitoring ethanol in sweat. Anal Chimica Acta 806:1–7. https://doi.org/10.1016/j.aca.2013.09.020
Nyman E, Palmlöv A (1936) The elimination of ethyl alcohol in sweat. Acta Physiol 74(2):155–159. https://doi.org/10.1111/j.1748-1716.1936.tb01150.x
Pawan GLS (1968) Physical exercise and alcohol metabolism in man. Nature 218:966–967
Swift RM (2000) Transdermal alcohol measurement for estimation of blood alcohol concentration. Alcohol Clin Exp Res 24(4):422–423. https://doi.org/10.1111/j.1530-0277.2000.tb02006.x
Swift RM (1993) Transdermal measurement of alcohol consumption. Addiction 88(8):1037–1039. https://doi.org/10.1111/j.1360-0443.1993.tb02122.x
Swift RM, Martin CS, Swette L, LaConti A, Kackley N (1992) Studies on a wearable, electronic, transdermal alcohol sensor. Alcohol Clin Exp Res 16(4):721–725. https://doi.org/10.1111/j.1530-0277.1992.tb00668.x
Labianca DA (1990) The chemical basis of the breathalyzer: a critical analysis. J Chem Educ 67(3):259–261. https://doi.org/10.1021/ed067p259
Dai Z, Rosen IG, Wang C, Barnett N, Luczak SE (2016) Using drinking data and pharmacokinetic modeling to calibrate transport model and blind deconvolution based data analysis software for transdermal alcohol biosensors. Math Biosci Eng 13(5):911–934. https://doi.org/10.3934/mbe.2016023
Dougherty DM, Charles NE, Acheson A, John S, Furr RM, Hill-Kapturczak N (2012) Comparing the detection of transdermal and breath alcohol concentrations during periods of alcohol consumption ranging from moderate drinking to binge drinking. Exp Clin Psychopharmacol 20:373–381
Dougherty DM, Hill-Kapturczak N, Liang Y, Karns TE, Cates SE, Lake SL, Roache JD (2014) Use of continuous transdermal alcohol monitoring during a contingency management procedure to reduce excessive alcohol use. Drug Alcohol Depend 142:301–306
Dougherty DM, Karns TE, Mullen J, Liang Y, Lake SL, Roache JD, Hill-Kapturczak N (2015) Transdermal alcohol concentration data collected during a contingency management program to reduce at-risk drinking. Drug Alcohol Depend 148:77–84
Dumett MA, Rosen IG, Sabat J, Shaman A, Tempelman L, Wang C, Swift RM (2008) Deconvolving an estimate of breath measured blood alcohol concentration from biosensor collected transdermal ethanol data. Appl Math Comput 196(2):724–743. https://doi.org/10.1016/j.amc.2007.07.026
Rosen IG, Luczak SE, Hu WW, Hankin M (2013) Discrete-Time Blind Deconvolution for Distributed Parameter Systems with Dirichlet Boundary Input and Unbounded Output with Application to a Transdermal Alcohol Biosensor. In: 2013 Proceedings of the Conference on Control and its Applications (CT) pp. 160–167. https://doi.org/10.1137/1.9781611973273.22
Rosen IG, Luczak SE, Weiss J (2014) Blind deconvolution for distributed parameter systems with unbounded input and output and determining blood alcohol concentration from transdermal biosensor data. Appl Math Comput 231:357–376. https://doi.org/10.1016/j.amc.2013.12.099
Fairbairn CE, Kang D, Bosch N (2020) Using machine learning for real-time bac estimation from a new-generation transdermal biosensor in the laboratory. Drug Alcohol Depend. https://doi.org/10.1016/j.drugalcdep.2020.108205
Oszkinat C, Shao T, Wang C, Rosen IG, Rosen AD, Saldich EB, Luczak SE (2022) Obtaining blood and breath alcohol concentration from biosensor measured transdermal alcohol level: Estimation and uncertainty quantification based on a covariate-dependent hidden markov model. Inverse Probl 38(5):005002
Oszkinat C, Luczak SE, Rosen IG (2021) Uncertainty quantification in the estimation of blood alcohol concentration using physics-informed neural networks. IEEE Trans Neural Netw Learn Syst
Leffingwell TR, Cooney NJ, Murphy JG, Luczak S, Rosen G, Dougherty DM, Barnett NP (2013) Continuous objective monitoring of alcohol use: twenty-first century measurement using transdermal sensors. Alcohol Clin Exp Res 37(1):16–22
Luczak SE, Ramchandani VA (2019) Special issue on alcohol biosensors: development, use, and state of the field: summary, conclusions, and future directions. Alcohol 81:161–165
Fairbairn CE, Kang D (2019) Temporal dynamics of transdermal alcohol concentration measured via new-generation wrist-worn biosensor. Alcohol Clin Exp Res 43(10):2060–2069
Hawekotte K, Luczak SE, Rosen IG (2021) Deconvolving breath alcohol concentration from biosensor measured transdermal alcohol level under uncertainty: a bayesian approach. Math Biosci Eng 18(5):6739–6770
Yao M, Luczak SE, Rosen IG (2021) Tracking and Deconvolving Blood Alcohol Concentration from Transdermal Alcohol Biosensor Measurements: A Population Model-Based LQG Approach in Hilbert Space
Banks HT, Ito K (1988) A unified framework for approximation in inverse problems for distributed parameter systems. Control Theor Adv Technol 4(1):73–90
Tanabe H (1979) Equations of Evolution. In: Monographs and Studies in Mathematics, Pitman https://books.google.com/books?id=Dn6zAAAAIAAJ
Pazy A (1983) Semigroups of Linear Operators and Applications to Partial Differential Equations. In: Applied Mathematical Sciences, Springer, https://books.google.co.in/books?id=80XYPwAACAAJ
Banks HT, Kunisch K (2012) Estimation Techniques for Distributed Parameter Systems. Springer, https://www.springer.com/us/book/9780817634339
Yao M, Luczak SE, Rosen IG (2021) A population model-based lqg compensator for the control of intravenously-infused alcohol studies with transdermal sensing
Hochreiter S, Schmidhuber J (1997) Long short-term memory. Neural comput 9:1735–80. https://doi.org/10.1162/neco.1997.9.8.1735
Hochreiter S (1991) Untersuchungen zu dynamischen neuronalen netzen
Raissi M, Perdikaris P, Karniadakis GE (2017) Physics informed deep learning (Part I): Data-driven Solutions of Nonlinear Partial Differential Equations
Raissi M, Perdikaris P, Karniadakis GE (2017) Physics informed deep learning (Part II): Data-driven Discovery of Nonlinear Partial Differential Equations
Raissi M, Perdikaris P, Karniadakis GE (2018) Multistep Neural Networks for Data-driven Discovery of Nonlinear Dynamical Systems
Raissi M, Perdikaris P, Karniadakis G (2018) Physics-informed neural networks: a deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations. J Comput Phys. https://doi.org/10.1016/j.jcp.2018.10.045
Sirlanci M, Rosen IG, Luczak SE, Fairbairn CE, Bresin K, Kang D (2018) Deconvolving the input to random abstract parabolic systems; a population model-based approach to estimating blood/breath alcohol concentration from transdermal alcohol biosensor data. Inverse problems 34(12) . arXiv:1807.05088v1
Sirlanci M, Luczak SE, Fairbairn CE, Kang D, Pan R, Yu X, Rosen IG (2018) Estimating the distribution of random parameters in a diffusion equation forward model for a transdermal alcohol biosensor. Automatica. to appear, arXiv:1808.04058
MacKay DJC (2003) Information Theory Inference and Learning Algorithms. Cambridge University Press, Cambridge
Gal Y, Ghahramani Z (2015) Dropout as a Bayesian approximation: Representing model uncertainty in deep learning. arXiv:1506.02142
Gal Y (2016) Uncertainty in deep learning. PhD thesis, University of Cambridge, Cambridge
Caflisch R (1998) Monte carlo and quasi-monte carlo methods. Acta Numerica 7:1–49. https://doi.org/10.1017/S0962492900002804
Hinton GE, van Camp D (1993) Keeping the neural networks simple by minimizing the description length of the weights. In: Proceedings of the Sixth Annual Conference on Computational Learning Theory. COLT ’93, pp. 5–13. Association for Computing Machinery, New York, NY, USA. https://doi.org/10.1145/168304.168306. https://doi.org/10.1145/168304.168306
Srivastava N, Hinton G, Krizhevsky A, Sutskever I, Salakhutdinov R (2014) Dropout: a simple way to prevent neural networks from overfitting. J Mach Learn Res 15(56):1929–1958
Saldich EB, Wang C, Rosen IG, Goldstein L, Bartroff J, Swift RM, Luczak SE (2020) Obtaining high-resolution multi-biosensor data for modeling transdermal alcohol concentration data. Alcoholism: Clinical and Experimental Research 44
Banks HT, Burns JA, Cliff EM (1981) Parameter estimation and identification for systems with delays. SIAM J Control Optim 19(6):791–828. https://doi.org/10.1137/0319051
Casella G, Berger RL (2002) Statistical Inference vol. 2. Duxbury, Pacific Grove, CA https://books.google.com/books/about/Statistical_Inference.html?id=0x_vAAAAMAAJ
Kingma D, Ba J (2014) Adam: A method for stochastic optimization. In: International Conference on Learning Representations
The Shapley Value (1988) Essays in Honor of Lloyd S. Shapley Cambridge University Press. https://doi.org/10.1017/CBO9780511528446
Funding
This research was supported in part by the National Institute on Alcohol Abuse and Alcoholism (NIAAA) grant no. R01AA026368.
Author information
Authors and Affiliations
Corresponding author
Ethics declarations
Conflict of interest
The authors have no relevant financial or non-financial interests to disclose.
Ethics approval
This work involved human subjects in its research. Approval of all ethical and experimental procedures and protocols was granted by USC IRB.
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Appendix A List of Covariates
Appendix A List of Covariates
Name | Description |
---|---|
Session Type | Single, dual, or steady |
Gender | Male or female |
Ethnicity | African/African-American, Asian, caucasian, hispanics, or mixed |
BMI | Body mass index |
RM | Resting metabolism |
SM | Skeletal muscle |
Wais_Hip | waist-to-Hip ratio |
AGE | Age of participants in years |
ATE | Number of hours from eating to start of drinking session |
WATER | Number of hours from time last drank water to time start drinking |
CAL_LUN | Amount of calories for lunch |
BPBase_S | Systolic blood pressure at rest |
BPBase_D | Diastolic blood pressure at rest |
BPPeak_S | Systolic blood pressure at peak |
BPPeak_D | Diastolic blood pressure at peak |
BPLow_S_S | Lowest systolic blood pressure reading before lunch |
BPLow_S_D | Diastolic blood pressure reading at lowest systolic reading before lunch |
BPLow_D_S | Systolic blood pressure reading at lowest diastolic reading before lunch |
BPLow_D_D | Lowest diastolic blood pressure reading before lunch |
28d_M | Past 28 days: maximum number of drinks on single day |
28d_B | Number of binge days (5 drinks for women, 4 for men) |
28d_T | Total number of drinks |
90d_M | Past 90 days: maximum number of drinks on single day |
90d_B | Number of binge days (5 drinks for women, 4 for men) |
90d_T | Total number of drinks |
Rights and permissions
About this article
Cite this article
Oszkinat, C., Luczak, S.E. & Rosen, I.G. An abstract parabolic system-based physics-informed long short-term memory network for estimating breath alcohol concentration from transdermal alcohol biosensor data. Neural Comput & Applic 34, 18933–18951 (2022). https://doi.org/10.1007/s00521-022-07505-w
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s00521-022-07505-w