Machine Learning-Enabled NIR Spectroscopy. Part 3: Hyperparameter by Design (HyD) Based ANN-MLP Optimization, Model Generalizability, and Model Transferability

Machine Learning-Enabled NIR Spectroscopy. Part 3: Hyperparameter by Design (HyD) Based ANN-MLP Optimization, Model Generalizability, and Model Transferability Hussain Ali Prakash Muthudoss http://orcid.org/0000-0003-2136-3903 Chirag Chauhan Ilango Kaliappan Dinesh Kumar http://orcid.org/0000-0001-9785-0500 Amrit Paudel http://orcid.org/0000-0002-3325-7828 Gobi Ramasamy http://orcid.org/0000-0002-0176-5414 Abstract

Data variations, library changes, and poorly tuned hyperparameters can cause failures in data-driven modelling. In such scenarios, model drift, a gradual shift in model performance, can lead to inaccurate predictions. Monitoring and mitigating drift are vital to maintain model effectiveness. USFDA and ICH regulate pharmaceutical variation with scientific risk-based approaches. In this study, the hyperparameter optimization for the Artificial Neural Network Multilayer Perceptron (ANN-MLP) was investigated using open-source data. The design of experiments (DoE) approach in combination with target drift prediction and statistical process control (SPC) was employed to achieve this objective. First, pre-screening and optimization DoEs were conducted on lab-scale data, serving as internal validation data, to identify the design space and control space. The regression performance metrics were carefully monitored to ensure the right set of hyperparameters was selected, optimizing the modelling time and storage requirements. Before extending the analysis to external validation data, a drift analysis on the target variable was performed. This aimed to determine if the external data fell within the studied range or required retraining of the model. Although a drift was observed, the external data remained well within the range of the internal validation data. Subsequently, trend analysis and process monitoring for the mean absolute error of the active content were conducted. The combined use of DoE, drift analysis, and SPC enabled trend analysis, ensuring that both current and external validation data met acceptance criteria. Out-of-specification and process control limits were determined, providing valuable insights into the model’s performance and overall reliability. This comprehensive approach allowed for robust hyperparameter optimization and effective management of model lifecycle, crucial in achieving accurate and dependable predictions in various real-world applications.

Graphical Abstract
12 07 2023 254 2697 1 10.1007/springer_crossmark_policy link.springer.com false 8 August 2023 1 November 2023 7 December 2023 The authors declare no competing interests. Graz University of Technology http://dx.doi.org/10.13039/100008332 https://creativecommons.org/licenses/by/4.0 https://creativecommons.org/licenses/by/4.0 10.1208/s12249-023-02697-3 20231230080141402 https://link.springer.com/10.1208/s12249-023-02697-3 https://link.springer.com/content/pdf/10.1208/s12249-023-02697-3.pdf https://link.springer.com/content/pdf/10.1208/s12249-023-02697-3.pdf https://link.springer.com/article/10.1208/s12249-023-02697-3/fulltext.html J Pharm Sci Elsevier D Saravanan 110 833 2021 10.1016/j.xphs.2020.09.022 Saravanan D, Muthudoss P, Khullar P, Rose VA. Quantitative microscopy: particle size/shape characterization, addressing common errors using ‘analytics continuum’ approach. J Pharm Sci Elsevier. 2021;110:833–49. AAPS PharmSciTech P Muthudoss 23 277 2022 10.1208/s12249-022-02403-9 Muthudoss P, Tewari I, Chi RLR, Young KJ, Ann EYC, Hui DNS, et al. Machine learning-enabled NIR spectroscopy in assessing powder blend uniformity: clear-up disparities and biases induced by physical artefacts. AAPS PharmSciTech. 2022;23:277 (Springer). AAPS PharmSciTech. H Ali 24 34 2023 10.1208/s12249-022-02493-5 Ali H, Muthudoss P, Ramalingam M, Kanakaraj L, Paudel A, Ramasamy G. Machine learning–enabled NIR spectroscopy. Part 2: workflow for selecting a subset of samples from publicly accessible data. AAPS PharmSciTech. 2023;24:34 (Springer). Innov Food Sci Emerg Technol. JA Cayuela-Sánchez 51 139 2019 10.1016/j.ifset.2018.05.018 Cayuela-Sánchez JA, Palarea-Albaladejo J, García-Martín JF, del Carmen Pérez-Camino M. Olive oil nutritional labeling by using Vis/NIR spectroscopy and compositional statistical methods. Innov Food Sci Emerg Technol. 2019;51:139–47 (Elsevier). Food Anal Methods Springer AC de Oliveira Moreira 14 865 2021 10.1007/s12161-020-01933-x de Oliveira Moreira AC, Braga JWB. Authenticity identification of copaiba oil using a handheld NIR spectrometer and DD-SIMCA. Food Anal Methods Springer. 2021;14:865–72. 10.1146/annurev.food.080708.100915 Mauer LJ, Taylor LS. Water-solids interactions: deliquescence. Annual review of food science and technology. 2010;10;1:41–63. Food Addit Contam Part A S Kar 36 863 2019 10.1080/19440049.2019.1600746 Kar S, Tudu B, Jana A, Bandyopadhyay R. FT-NIR spectroscopy coupled with multivariate analysis for detection of starch adulteration in turmeric powder. Food Addit Contam Part A. 2019;36:863–75 (Taylor & Francis). J Appl Pharm Sci D Saravanan 9 1 2019 10.7324/JAPS.2019.90901 Saravanan D, Muthudoss P, Khullar P, Rosevenis A. Vendor qualification: utilization of solid state characterization “Toolbox” to assess material variability for active pharmaceutical ingredient. J Appl Pharm Sci. 2019;9:1–9. Rajesh PKS, Kumaravelu C, Gopal A, Suganthi S. Studies on identification of medicinal plant variety based on NIR spectroscopy using plant leaves. 2013 15th Int Conf Adv Comput Technol. 2013. p. 1–4. Talanta. P Mishra 225 121971 2021 10.1016/j.talanta.2020.121971 Mishra P, Herrmann I, Angileri M. Improved prediction of potassium and nitrogen in dried bell pepper leaves with visible and near-infrared spectroscopy utilising wavelength selection techniques. Talanta. 2021;225:121971 (Elsevier). Chemom Intell Lab Syst. P Mishra 212 104190 2021 10.1016/j.chemolab.2020.104190 Mishra P, Roger JM, Marini F, Biancolillo A, Rutledge DN. Parallel pre-processing through orthogonalization (PORTO) and its application to near-infrared spectroscopy. Chemom Intell Lab Syst. 2021;212:104190 (Elsevier). Postharvest Biol Technol. P Mishra 168 111271 2020 10.1016/j.postharvbio.2020.111271 Mishra P, Roger JM, Rutledge DN, Woltering E. SPORT pre-processing can improve near-infrared quality prediction models for fresh fruits and agro-materials. Postharvest Biol Technol. 2020;168:111271 (Elsevier). 10.5772/intechopen.99017 Sampaio PS, Brites CM. Near-Infrared spectroscopy and machine learning: analysis and classification methods of rice. Integrative Advances in Rice Research. 2022;26:257. https://doi.org/10.5772/intechopen.99017. Int Dairy J F Pi 19 624 2009 10.1016/j.idairyj.2009.05.002 Pi F, Shinzawa H, Ozaki Y, Han D. Non-destructive determination of components in processed cheese slice wrapped with a polyethylene film using near-infrared spectroscopy and chemometrics. Int Dairy J. 2009;19:624–9 (Elsevier). J Food Sci Technol Y Wang 52 5305 2015 10.1007/s13197-014-1574-5 Wang Y, Ding W, Kou L, Li L, Wang C, Jurick WM. A non-destructive method to assess freshness of raw bovine milk using FT-NIR spectroscopy. J Food Sci Technol. 2015;52:5305–10 (Springer). Food Chem. N Núñez-Sánchez 190 244 2016 10.1016/j.foodchem.2015.05.083 Núñez-Sánchez N, Martínez-Marín AL, Polvillo O, Fernández-Cabanás VM, Carrizosa J, Urrutia B, et al. Near infrared spectroscopy (NIRS) for the determination of the milk fat fatty acid profile of goats. Food Chem. 2016;190:244–52 (Elsevier). 10.1208/s12248-022-00725-x Rish AJ, Henson SR, Alam A, Liu Y, Drennen JK, Anderson CA. Comparison between pure component modeling approaches for monitoring pharmaceutical powder blends with near ‑ infrared spectroscopy in continuous manufacturing schemes. AAPS J [Internet]. Springer International Publishing; 2022;24:1–10. https://doi.org/10.1208/s12248-022-00725-x J Pharm Biomed Anal Y Sulub 49 48 2009 10.1016/j.jpba.2008.10.001 Sulub Y, Wabuyele B, Gargiulo P, Pazdan J, Cheney J, Berry J, et al. Real-time on-line blend uniformity monitoring using near-infrared reflectance spectrometry: a noninvasive off-line calibration approach. J Pharm Biomed Anal. 2009;49:48–54. 10.1016/j.aca.2013.12.002 Ni W, Nørgaard L, Mørup M. Non-linear calibration models for near infrared spectroscopy. Anal Chim Acta [Internet]. Elsevier B.V.; 2014;813:1–14. https://doi.org/10.1016/j.aca.2013.12.002 J Pharm Biomed Anal P Mishra 192 113684 2021 10.1016/j.jpba.2020.113684 Mishra P, Nordon A, Roger J-M. Improved prediction of tablet properties with near-infrared spectroscopy by a fusion of scatter correction techniques. J Pharm Biomed Anal. 2021;192:113684 (Elsevier). Anal Chim Acta W Kessler 642 127 2009 10.1016/j.aca.2009.01.057 Kessler W, Oelkrug D, Kessler R. Using scattering and absorption spectra as MCR-hard model constraints for diffuse reflectance measurements of tablets. Anal Chim Acta. 2009;642:127–34 (Elsevier). Drug Test Anal H Rebiere 8 370 2016 10.1002/dta.1843 Rebiere H, Ghyselinck C, Lempereur L, Brenier C. Investigation of the composition of anabolic tablets using near infrared spectroscopy and Raman chemical imaging. Drug Test Anal. 2016;8:370–7 (Wiley Online Library). Drug Dev Ind Pharm AS Hussain 20 1739 1994 10.3109/03639049409038390 Hussain AS, Shivanand P, Johnson RD. Application of neural computing in pharmaceutical product development: computer aided formulation design. Drug Dev Ind Pharm. 1994;20:1739–52. J Pharm Sci JA Dowell 88 154 1999 10.1021/js970148p Dowell JA, Hussain A, Devane J, Young D. Artificial neural networks applied to the in vitro-in vivo correlation of an extended-release formulation: initial trials and experience. J Pharm Sci. 1999;88:154–60. Pharm Res AS Hussain 8 1248 1991 10.1023/A:1015843527138 Hussain AS, Yu X, Johnson RD. Application of neural computing in pharmaceutical product development. Pharm Res. 1991;8:1248–52 (Springer). SN Comput Sci IH Sarker 2 160 2021 10.1007/s42979-021-00592-x Sarker IH. Machine learning: algorithms, real-world applications and research directions. SN Comput Sci. 2021;2:160 (Springer). 10.1038/s41573-019-0024-5 Vamathevan J, Clark D, Czodrowski P, Dunham I, Ferran E, Lee G, Li B, Madabhushi A, Shah P, Spitzer M, Zhao S. Applications of machine learning in drug discovery and development. Nat Rev Drug Discov 2019;18(6):463–77. Qual Reliab Eng Int R Arboretti 38 1131 2022 10.1002/qre.3025 Arboretti R, Ceccato R, Pegoraro L, Salmaso L. Design of experiments and machine learning for product innovation : a systematic literature review. Qual Reliab Eng Int. 2022;38:1131–56. Bioprocess Biosyst Eng D Rodriguez-Granrose 44 1301 2021 10.1007/s00449-021-02529-3 Rodriguez-Granrose D, Jones A, Loftus H, Tandeski T, Heaton W, Foley KT, et al. Design of experiment (DOE) applied to artificial neural network architecture enables rapid bioprocess improvement. Bioprocess Biosyst Eng. 2021;44:1301–8 (Springer). 10.1109/ICCoSITE57641.2023.10127717 Gaurav D, Tiwari S. Interpretability vs explainability: the black box of machine learning. 2023 Int Conf Comput Sci Inf Technol Eng. 2023. p. 523–8. Entropy. P Linardatos 23 18 2020 10.3390/e23010018 Linardatos P, Papastefanopoulos V, Kotsiantis S. Explainable ai: a review of machine learning interpretability methods. Entropy. 2020;23:18 (MDPI). 10.1053/j.semdp.2023.02.002 Albahra S, Gorbett T, Robertson S, D'Aleo G, Kumar SV, Ockunzzi S, Lallo D, Hu B, Rashidi HH. Artificial intelligence and machine learning overview in pathology & laboratory medicine: A general review of data preprocessing and basic supervised concepts. Semin Diagn Pathol 2023;40(2):71–87. 10.1016/B978-0-323-89925-3.00015-0 Chaudhary S, Muthudoss P, Madheswaran T, Paudel A, Gaikwad V. Artificial intelligence (AI) in drug product designing, development, and manufacturing. In: A handbook of artificial intelligence in drug delivery. Academic Press; 2023. p. 395–442. Appl Spectrosc M Dyrby 56 579 2002 10.1366/0003702021955358 Dyrby M, Engelsen SB, Nørgaard L, Bruhn M, Lundsberg-Nielsen L. Chemometric quantitation of the active substance (containing C≡N) in a pharmaceutical tablet using near-infrared (NIR) transmittance and NIR FT-Raman spectra. Appl Spectrosc. 2002;56:579–85. https://www.evidentlyai.com/. Open-source machine learning monitoring. Accessed 14th Nov 2023. Madkour AH, Elsayed A, Abdel-Kader H. Historical Isolated Forest for detecting and adaptation concept drifts in nonstationary data streaming. IJCI. Int J Comput Inf. 2023;10(2):16–27. J Chemom CM Andersen 24 728 2010 10.1002/cem.1360 Andersen CM, Bro R. Variable selection in regression—a tutorial. J Chemom. 2010;24:728–37 (Wiley Online Library). Int J Pharm T Rajalahti 417 280 2011 10.1016/j.ijpharm.2011.02.019 Rajalahti T, Kvalheim OM. Multivariate data analysis in pharmaceutics: a tutorial review. Int J Pharm. 2011;417:280–90 (Elsevier). J Open Source Softw S Raschka 3 638 2018 10.21105/joss.00638 Raschka S. MLxtend: Providing machine learning and data science utilities and extensions to Python’s scientific computing stack. J Open Source Softw. 2018;3:638 (The Open Journal). J Mach Learn Res F Pedregosa 12 2825 2011 Pedregosa F, Varoquaux G, Gramfort A, Michel V, Thirion B, Grisel O, et al. Scikit-learn: machine learning in Python. J Mach Learn Res. 2011;12:2825–30 (JMLR.org). Comput Sci Eng JD Hunter 9 90 2007 10.1109/MCSE.2007.55 Hunter JD. Matplotlib: a 2D graphics environment. Comput Sci Eng. 2007;9:90–5 (IEEE Computer Society). Food Control. C Lörchner 133 108601 2022 10.1016/j.foodcont.2021.108601 Lörchner C, Horn M, Berger F, Fauhl-Hassek C, Glomb MA, Esslinger S. Quality control of spectroscopic data in non-targeted analysis–development of a multivariate control chart. Food Control. 2022;133:108601 (Elsevier). Processes. MDPI M Malindzakova 11 1246 2023 10.3390/pr11041246 Malindzakova M, Čulková K, Trpčevská J. Shewhart control charts implementation for quality and production management. Processes MDPI. 2023;11:1246. 10.1109/ACCESS.2023.3282569 Pérez-Benítez BE, Tercero-Gómez JG, Khakifirooz M. A review on statistical process control in healthcare: data-driven monitoring schemes. IEEE Access. 2023;11:56248–272. https://doi.org/10.1109/ACCESS.2023.3282569 (https://ieeexplore.ieee.org/abstract/document/10144935). Pharm Res LX Yu 25 781 2008 10.1007/s11095-007-9511-1 Yu LX. Pharmaceutical quality by design: product and process development, understanding, and control. Pharm Res. 2008;25:781–91. J Valid Technol PL Pluta 17 51 2011 Pluta PL. FDA lifecycle approach to process validation-what, why, and how? J Valid Technol. 2011;17:51 (MultiMedia Healthcare Inc). U.S. Food and Drug Administration (USFDA), Process validation: general principles and practices, ID: FDA-2008-D-0559, USFDA, Silver Spring, MD 2011. 10.1007/978-3-031-20298-8_32 Lange R, Schnor T. Product quality, quality control and validation. In: Practical Pharmaceutics: an international guideline for the preparation, care and use of medicinal products. Cham: Springer International Publishing; 2023. p. 767–83. Int J Pharm Investig K Pramod 6 129 2016 10.4103/2230-973X.187350 Pramod K, Tahir MA, Charoo NA, Ansari SH, Ali J. Pharmaceutical product development: a quality by design approach. Int J Pharm Investig. 2016;6:129 (Wolters Kluwer--Medknow Publications). Acta Pharm B Kovács 71 497 2021 10.2478/acph-2021-0039 Kovács B, Kovács-Deák B, Székely-Szentmiklósi I, Fülöp I, Bába L-I, Boda F, et al. Quality-by-design in pharmaceutical development: from current perspectives to practical applications. Acta Pharm. 2021;71:497–526 (Hrvatsko farmaceutsko društvo). J Pharm Investig S-H Lee 52 649 2022 10.1007/s40005-022-00575-x Lee S-H, Kim J-K, Jee J-P, Jang D-J, Park Y-J, Kim J-E. Quality by design (QbD) application for the pharmaceutical development process. J Pharm Investig. 2022;52:649–82 (Springer).