Defense Date


Graduation Date

Spring 5-23-2020


One-year Embargo

Submission Type


Degree Name





School of Pharmacy

Committee Chair

James K. Drennen

Committee Member

Carl Anderson

Committee Member

Wilson Meng

Committee Member

Frederick Fochtman

Committee Member

David J. Wargo


Counterfeit Medicine, Falsified Drug, Portable Spectroscopy, Near-Infrared, Raman, Chemometric, Pharmaceutical.


The spread of falsified drugs is increasing worldwide. Currently, 10%-30% of drugs in the world are falsified. Unfortunately, the supply chain system of developing countries (from manufacturers to customers) is not well monitored and suffers the highest rates of fraudulent activities. Also, the rise of product procurement from the internet increases the chance of American consumers' exposure to poor quality drugs. To combat this horrendous activity, surveillance of pharmaceutical materials is required in the supply chain system. Spectroscopic techniques (e.g., Near-Infrared and Raman spectroscopies) can be a potential solution to authenticate samples in different locations; they are non-destructive, safe, rapid, portable, and affordable. However, before deploying these techniques in the field, rigorous method development is required with the help of chemometrics. The chemometric tools to be used should declare the test sample as either target class (authentic samples) or non-target class (alternate class or falsified samples). One challenge of method development is the sensitivity of spectrometers toward unwanted variabilities including moisture, batch to batch variabilities, raw material variabilities, etc. Initial research for this dissertation observed that the traditional chemometric methods, which are based on distribution assumptions, provided a high number of false negatives due to violation of distribution assumptions in the presence of unwanted variations. It was demonstrated that adding samples from different seasons in the calibration set created binominal or multimodal distributions due to moisture variations - which violated the assumptions of the principal component analysis based soft independent modeling of class analogy (SIMCA) method. Hence, the support vector data description (SVDD) method—which has no distribution assumption—is proposed for use as a class modeling approach for authentication purposes. Implementing the SVDD algorithm improved model performance by reducing false negatives relative to the traditional multivariate class modeling approach (i.e., SIMCA). In addition, while developing the SVDD method, this dissertation suggested using different non-target class samples produced by competitor manufacturers or synthetic samples generated in the laboratory using design of experiment (DOE) as test or validation set to decrease false positives. This work also evaluated several commercially available products in two local pharmacies using portable spectrometers to validate the practical usefulness of the proposed methods. To summarize, the research performed for this dissertation has demonstrated the value of critical prior knowledge regarding pharmaceutical products, pharmaceutical manufacturing/processing, analytical methodology, and advanced chemometric techniques in a unique way to develop a successful spectroscopic authentication system.



Additional Citations

Hossain, M. N., Igne, B., Anderson, C. A., & Drennen III, J. K. (2019). Influence of moisture variation on the performance of Raman spectroscopy in quantitative pharmaceutical analyses. Journal of pharmaceutical and biomedical analysis, 164, 528-535.

Available for download on Saturday, May 08, 2021