Snowdon employs its expertise in statistics, chemometrics, and pattern recognition to assist partner organizations to solve problems in product quality and safety assurance. This is accomplished by transforming complex analytical data obtained from chromatographic and spectroscopic experiments on food, pharmaceutical, and other consumer products into simple computational models for data visualization, classification, and prediction. These models have become indispensable tools for issues in product quality assurance.
Chemometrics tools and techniques have become an essential component in modern chemical and biomedical industries. Chemometrics software has been widely used by product development scientists, process engineers, Process Analytical Technology (PAT) specialists, and Quality Assurance/Quality Control (QA/QC) scientists to build reliable models, ensure product quality, classify raw materials, and to monitor process end points in real-time. Snowdon provides a suite of customized tools to satisfy the needs for our clients and partners for data exploration and analysis including the short list cited below.
- Principal Component Analysis (PCA)
- Hierarchical Cluster Analysis
- Data Preprocessing (mean centering, autoscale, variance scale)
- Regression (PLS, PCR, MLR, 3-way PLS) and Prediction
- SIMCA and PLS-DA Classification
- Design of Experiments (DOE)
- ANOVA and Response Surface Methodology
- Clustering (K-Means)
- Genetic Algorithms (GAs)
- Machine learning, artificial neural networks (ANNs)
- Artificial Intelligence (AI)
In one example, Snowdon conducted a study for the FDA to develop chemometric tools and predictive models to distinguish authentic from impure and contaminated heparin samples. The impetus behind the study arose in March 2008, when oversulfated chondroitin sulfate (OSCS) was identified as a contaminant in heparin that caused nearly 100 deaths and widespread illness in the US. In response to this public health crisis, the FDA developed NMR methods so that manufacturers could screen their active pharmaceutical ingredients (APIs) for the presence of OSCS contaminant. This task was complicated in that heparin is a heterogeneous mixture of straight-chain anionic polysaccharides called glycosaminoglycans which can mask potential contaminants; therefore, the FDA contracted Snowdon to develop chemometric tools that consider heparin’s inherent variability. Snowdon pursued a strategy based on the concept of “pharmaceutical fingerprinting” that proved successful in earlier studies with the FDA. Processing of NMR and other analytical data provided by the FDA enabled Snowdon to generate a suite of chemometric models to identify and quantify the content of natural impurities (e.g., dermatan sulfate, or DS) and contaminants (e.g., oversulfated chondroitin sulfate, or OSCS) in heparin obtained from various commercial manufacturers. The predictive performance of the resulting models achieved an impressive >95% success rate in distinguishing acceptable heparin from unacceptable impure and contaminated heparin.
In a second example, Snowdon collaborated with partners in an NSF-supported project to implement principles of Quality by Design (QbD) and Process Analytical Technology (PAT) espoused by the FDA to design, monitor, and control pharmaceutical manufacturing processes. This is achieved by developing adaptive in-line feedback control systems for real-time regulation of the pharmaceutical manufacturing process. The central goal of QbD is to produce pharmaceuticals that are free of contamination and that reproducibly deliver the intended therapeutic benefit. Using near infrared spectroscopy (NIR-S) and near infrared chemical imaging (NIR-CI) data provided by our partner organization to explore the surface of drug tablets, Snowdon employed customized multivariate chemometric methods to explore both qualitative and quantitative relationships between the NIR spectral output and the pre-defined Critical Quality Attributes (CQAs) of the drug tablets. The ultimate goal is to deploy these methods and models for rapid and early detection of impurities and contaminants in these pharmaceutical products.
- Pharmaceutical fingerprinting: evaluation of neural networks and chemometric techniques for distinguishing among same-product manufacturers (Welsh, et al.)
- Preprocessing of HPLC trace impurity patterns by wavelet packets for pharmaceutical fingerprinting using artificial neural networks (Collantes, et al.)
- Application of a pruning algorithm to optimize artificial neural networks for pharmaceutical fingerprinting (Tetko, et al.)
- A strategy for developing consistent HPLC data for assessing sameness and difference in consistency of pharmaceutical products (Zielinski, et al.)
- Pharmaceutical fingerprinting in phase space. 1. Construction of phase fingerprints (Aksenova, et al.)
- Pharmaceutical fingerprinting in phase space. 2. Pattern recognition (Tetko, et al.)
- Fractal fingerprinting of chromatographic profiles based on wavelet analysis and its application to characterize the quality grade of medicinal herbs (Yiyu, et al.)
- Guidance for Industry: PAT – A Framework for Innovative Pharmaceutical Development, Manufacturing, and Quality Assurance (USFDA)
- Determination of galactosamine impurities in heparin samples by multivariate regression analysis of their (1)H NMR spectra (Zang, et al.)
- Identification of heparin samples that contain impurities or contaminants by chemometric pattern recognition analysis of proton NMR spectral data (Zang, et al.)
- Combining (1)H NMR spectroscopy and chemometrics to identify heparin samples that may possess dermatan sulfate (DS) impurities or oversulfated chondroitin sulfate (OSCS) contaminants (Zang, et al.)
- Class modeling analysis of heparin 1H NMR spectral data using the soft independent modeling of class analogy and unequal class modeling techniques (Zang, et al.)
- Experimental Design of Formulations Utilizing High Dimensional Model Representation (Li, et al.)