Saliva analysis has been gaining interest as a potential non-invasive source of disease indicative biomarkers due to being a complex biofluid correlating with blood-based constituents on a molecular level. For saliva to cement its usage for analytical applications, it is paramount to gain underpinning molecular knowledge and establish a ‘baseline’ of the salivary composition in healthy individuals as well as characterize how these factors are impacting its performance as potential analytical biofluid. Here, we have systematically studied the molecular spectral fingerprint of saliva, including the changes associated with gender, age, and time. Via hybrid artificial neural network algorithms and Raman spectroscopy, we have developed a non-destructive molecular profiling approach enabling the assessment of salivary spectral changes yielding the determination of gender and age of the biofluid source. Our classification algorithm successfully identified the gender and age from saliva with high classification accuracy. Discernible spectral molecular ‘barcodes’ were subsequently constructed for each class and found to primarily stem from amino acid, protein, and lipid changes in saliva. This unique combination of Raman spectroscopy and advanced machine learning techniques lays the platform for a variety of applications in forensics and biosensing.

由于唾液分析是一种复杂的生物流体，在分子水平上与血液成分相关联，因此它作为一种潜在的非侵入性疾病指示性生物标志物来源越来越受到关注。对于唾液来巩固其在分析应用中的用途，最重要的是获得基础分子知识并建立健康个体唾液成分的“基线”，以及表征这些因素如何影响其作为潜在分析生物流体的性能。在这里，我们系统地研究了唾液的分子光谱指纹，包括与性别、年龄和时间相关的变化。通过混合人工神经网络算法和拉曼光谱，我们开发了一种无损分子分析方法，能够评估唾液光谱变化，从而确定生物流体源的性别和年龄。我们的分类算法成功地从唾液中识别出性别和年龄，具有很高的分类准确率。随后为每个类别构建了可识别的光谱分子“条形码”，发现主要源于唾液中的氨基酸、蛋白质和脂质变化。拉曼光谱和先进机器学习技术的这种独特组合为法医和生物传感领域的各种应用奠定了平台。