Date of Publication

6-14-2021

Document Type

Master's Thesis

Degree Name

Master of Science in Manufacturing Engineering

Subject Categories

Manufacturing

College

Gokongwei College of Engineering

Department/Unit

Manufacturing Engineering and Management

Thesis Advisor

Renann G. Baldovino

Defense Panel Chair

Robert Kerwin C. Billones

Defense Panel Member

Rhen Anjerome R. Bedruz
Francisco Emmanuel T. Munsayac, III

Abstract/Summary

Currently, wet chemistry techniques such as HPLC dominate the pharmaceutical industry when complying with API content testing. UV-Vis spectroscopy proved to be an efficient and nondestructive method that could potentially obtain API assays simultaneously. However, its performance becomes limited due to severely overlapped spectra, such as the case of Acetaminophen (APAP), Dextromethorphan HBr (DEX), Guaifenesin (GUA), and Phenylephrine HCl (PHE) combination drug. Therefore, four machine learning models, namely, Partial Least Squares Regression (PLSR), Support Vector Regression (SVR), Random Forest Regression (RFR), and Artificial Neural Network (ANN), were developed, optimized, and tested that would breach this limitation. Subsequently, the four developed models were compared to each other, with ANN having R2 correlations of 99.96% for APAP, 95.65% for DEX, 99.51% for GUA, and 90.08% for PHE, which outperforms the three models. Finally, the optimum ANN model was validated, resulting in correlations of 99.25% for APAP, 95.61% for DEX, 99.34% for GUA, and 89.38% for PHE, proving its capability to generalize its prediction of API concentrations.

Abstract Format

html

Language

English

Format

Electronic

Physical Description

136 leaves, color illustrations

Keywords

Acetaminophen; Dextromethorphan; Guaifenesin; Spectrophotometry

Upload Full Text

wf_yes

Embargo Period

6-15-2021

Download

Share

COinS