Khondakar Kamil Reza, Mazumdar Hirak, Das Suparna, Kaushik Ajeet
School of Technology, India and Woxsen University, Hyderabad 502345, Telangana, India.
School of Engineering and Technology, Department of Computer Science and Engineering, Adamas University, Kolkata 700126, India.
Adv Colloid Interface Sci. 2025 Jul 5;344:103594. doi: 10.1016/j.cis.2025.103594.
Surface-enhanced Raman spectroscopy (SERS) is a powerful and highly sensitive analytical tool that has found application in healthcare and environmental monitoring. Significant progress has been made in developing SERS-based sensing technology, enabling ultra-high sensitivity through its label-free and fingerprint-level detection capabilities. They are being utilized for molecular diagnostics, screening of clinical samples for food safety, and environmental toxic monitoring. In SERS techniques, vibrational spectra of complex chemical mixtures are acquired as large datasets are extracted from image analysis. Further, subtle variations of SERS signatures from thousands of clinical samples impose a major challenge in identifying analytes for accurate diagnosis. To address these issues, machine learning (ML) algorithms and multivariate statistical analysis have been combined with SERS for extracting and predicting the better outcome. Advancements in artificial intelligence (AI) and ML have shown promising potential to enhance the capabilities of SERS through rapid analysis and automated data processing. By leveraging AI/ML, SERS can transition from merely sensing to a more comprehensive sense, where the algorithms not only detect but also interpret complex patterns in the data. This review delves into the integration of ML with SERS, exploring how ML algorithms can improve these techniques by providing more accurate and insightful analyses. We discuss the overall process of merging ML with SERS, emphasize their applications in molecular diagnostics and screening, and offer insights into the future of ML-enhanced SERS sensor technologies, highlighting the transformative potential of AI/ML in moving from simple sensing to intelligent sensing.
表面增强拉曼光谱（SERS）是一种强大且高度灵敏的分析工具，已在医疗保健和环境监测中得到应用。基于SERS的传感技术取得了重大进展，通过其无标记和指纹级检测能力实现了超高灵敏度。它们正被用于分子诊断、食品安全临床样本筛查以及环境毒性监测。在SERS技术中，从图像分析中提取大量数据集时会获取复杂化学混合物的振动光谱。此外，数千个临床样本的SERS特征的细微变化给准确诊断中识别分析物带来了重大挑战。为了解决这些问题，机器学习（ML）算法和多元统计分析已与SERS相结合，以提取和预测更好的结果。人工智能（AI）和ML的进步显示出通过快速分析和自动化数据处理增强SERS能力的巨大潜力。通过利用AI/ML，SERS可以从单纯的传感转变为更全面的感知，即算法不仅能检测还能解释数据中的复杂模式。本综述深入探讨了ML与SERS的整合，探索ML算法如何通过提供更准确和有洞察力的分析来改进这些技术。我们讨论了将ML与SERS合并的整体过程，强调它们在分子诊断和筛查中的应用，并对ML增强的SERS传感器技术的未来提供见解，突出AI/ML从简单传感向智能传感转变的变革潜力。
