Computational intelligence model of orally disintegrating tablets – an attempt to explain disintegration process

Traditional tablets are not an ideal dosage form. Many groups of patients, e.g. pediatric or geriatric, experience problems with swallowing or simply are not willing to take tablets, consequently reducing patient’s compliance. Therefore, to overcome these inconveniences, orally disintegrating tablets (OTDs) were introduced into the drug market. One of the methods of preparing ODTs is a direct compression, which is cost-efficient and simple. However, many factors are affecting the disintegration time of ODTs, which are usually optimized in a laboratory during the try-and-error assays.

We curated and greatly enhanced the database presented by Han et al. [1]. Moreover, we introduced chemical descriptors as active pharmaceutical ingredient (API) characteristics. We used H2O AutoML platform [2, 3] in order to develop a model and SHAP method to explain its predictions [4].

Based on the obtained DeepLearning model with NRMSE of 8.1% and R2 of 0.84, we have identified critical parameters affecting the process of disintegration of directly compressed ODTs.

1. R. Han, Y. Yang, X. Li, D. Ouyang. Predicting oral disintegrating tablet formulations by neural network techniques. Asian J. Pharm. Sci., 134 (2018), pp. 336-342

2. Szlęk J. 2021. h2o_AutoML_Python, Python script for AutoML in h2o. Available online: https://github.com/jszlek/h2o_AutoML_ Python (accessed on 10 April 2021).

3. LeDell, E.; Poirier, S. H2O AutoML: Scalable Automatic Machine Learning. 7th ICML Workshop on Automated Machine Learning. 2020.

4. Lundberg, S. M.; Erion, G.G; Lee S.I. Consistent Individualized Feature Attribution for Tree Ensembles. arXiv preprint arXiv:1802.03888. 2018.