How Will Spectroscopy Benefit From Data-Driven Approaches?

The U. S. Pharmacopeia (USP) is a nonprofit organization that is primarily focused on improving the global supply chain, particularly when it comes to medicine. A scientific organization that strives to build greater trust in the supply of safe medicines (1). The USP, as part of their core mission, develops standards that are designed to improve the pharmaceutical drug development process, the goal of which being to inspire confidence and allow companies to move swiftly with certainty (2). Following these standards, according to USP, helps reduce the risk of incorrect results, which could lead to some negative externalities, including product delays and market withdrawals (2).

Some of these standards ultimately trickle down to spectroscopy, which is getting increasingly involved in pharmaceutical analysis. Primarily, spectroscopic techniques enable the precise classification and quantification of drugs and processes, which helps accelerate bringing drugs to market (3). For example, inductively coupled plasma–mass spectrometry (ICP-MS) and ICP–optical emission spectrometry (ICP-OES) can provide trace elemental analysis, which can support drug safety and stability assessments (3).

These tools are increasingly being used in conjunction with data-driven methodologies. By integrating artificial intelligence (AI) and machine learning (ML) into their work, scientists are reducing the calibration burdens while rapidly conducting quality measurements on pharmaceutical drugs during the manufacturing process (3).

So what does the integration of AI and ML means for the role of spectroscopy in pharmaceutical analysis going forward? To find out, we asked Yang Li, Staff Scientist II at USP who oversees the spectroscopy chapters at USP, about these trends.

In this Spectroscopy blog post, Liu offers his brief thoughts about how spectroscopy is set to benefit from artificial intelligence (AI), machine learning (ML), and data-driven approaches heading into 2026 and beyond. Liu’s commentary below showcases how ingrained AI and ML are in spectroscopic analyses, and the role these approaches will play in propelling spectroscopy forward in the digital age we now live in.

From the Expert

Spectroscopy is particularly well positioned to benefit from advances in data-driven methodologies, especially traceable and interpretable machine learning (ML) approaches. Spectroscopic techniques such as ultraviolet (UV), infrared (IR), near-infrared (NIR), Raman, and nuclear magnetic resonance (NMR) inherently generate multivariate data sets that align naturally with ML models designed to support transparency, traceability, and lifecycle management.

At present, the adoption of interpretable ML models in spectroscopic applications is expanding across research and development, quality control, and process monitoring, supporting use cases such as qualitative and quantitative analysis, real-time release testing, and continuous manufacturing. When appropriately developed and validated, these approaches can enhance method robustness, improve decision-making, and deepen understanding of spectral–property relationships, while remaining compatible with regulatory expectations for scientific justification and traceability.

At the same time, the application of black-box AI models in regulated spectroscopic environments is expected to remain limited until challenges related to explainability, model governance, and accountability are adequately addressed. These challenges include the ability to clearly explain model prediction pathways, define model boundaries and applicability domains, interpret and manage model drift over time, and ensure appropriate documentation, change control, and human oversight, including revalidation or verification approaches, throughout the model lifecycle. Future progress in spectroscopy will depend not only on algorithmic performance, but also on establishing regulatory pathways and fit-for-purpose frameworks or guidance that integrate model interpretability, validation, and lifecycle control into spectroscopic workflows.

References

1. USP, About USP. USP.org. Available at: https://www.usp.org/about#:~:text=About%20the%20U.S.%20Pharmacopeia%20(USP,Our%20code%20of%20ethics: (accessed 2025-12-18).
2. USP, USP Reference Standards. USP.org. Available at: https://www.usp.org/reference-standards?gad_source=1&gad_campaignid=209344496&gbraid=0AAAAADn1zwC049JTKegSLMLD83wS_c-OT&gclid=Cj0KCQiA6Y7KBhCkARIsAOxhqtOX_vqlp2-xQ9E60Ai37Noy7AxQxV4hskyDTLMLJCvmVNaWikOBhIMaAp2uEALw_wcB (accessed 2025-12-18).
3. Workman, Jr., J. A Review of the Latest Spectroscopic Research in Pharmaceutical and Biopharmaceutical Applications. Spectroscopy 2024, 39 (6), 25–29. DOI: 10.56530/spectroscopy.at8171q5 (accessed 2025-12-18).

About the Authors

Will Wetzel is a Senior Editor of Spectroscopy magazine, a MJH Life Sciences brand.

Yang Liu is a Scientific Liaison for General Chapters at the United States Pharmacopeia (USP). Dr. Liu graduated from the College of Pharmacy at the University of Illinois Chicago. He previously served as the Manager of Product Quality and Analytical Methods in the Digital and Innovation division at USP. During his tenure in the Digital and Innovation division, Liu was dedicated to evaluating emerging technologies and incorporating them into USP operations. He also led the development of digital products, including software and database creation. His focus was on advancing digital fluency and the application of digital technologies and solutions throughout the organization and for USP stakeholders. Currently, Liu is responsible for USP General Chapters development, including the spectroscopy technologies (such as UV, NIR, Raman, and NMR), process analytical technologies, and digital reference standard.