Ondieki Annah M, Birech Zephania, Kaduki Kenneth A, Mwangi Peter W, Juma Moses, Chege Boniface M
Laser Physics and Spectroscopy Research Group, Department of Physics, University of Nairobi, Nairobi, Kenya.
Department of Medical Physiology, University of Nairobi, Nairobi, Kenya.
PLoS One. 2025 May 23;20(5):e0323697. doi: 10.1371/journal.pone.0323697. eCollection 2025.
Growth hormone (GH) and testosterone (TE) levels in blood are crucial indicators of human health and performance in clean sports. Deviations from normal levels can signal serious health issues, such as fertility problems, cancer, or pituitary tumors. Existing detection methods for these hormones are often costly, time-consuming, and lack portability. In this study, we explored the potential of Surface-Enhanced Raman Spectroscopy (SERS) in distinguishing blood samples from Sprague Dawley (SD) rats injected with exogenous GH, TE and both hormones from those not injected. Then, used artificial neural network (ANN) models trained, and validated in predicting levels of these hormones in blood. Blood samples from SD rats injected with GH, TE, both hormones, and non-injected rats were analyzed using the SERS method upon 785 nm laser excitation. The recorded Raman spectra from blood of GH and TE injected and non-injected rats displayed hormone-specific band intensity variations. Additionally, Principal Component Analysis (PCA) showed temporal changes in band intensities post-injection, suggesting hormone-induced biochemical alterations. In particular, Raman bands centered around 1378 cm⁻¹ for all groups, 658 cm⁻¹ for GH, and 798 cm⁻¹ for GH and TE displayed significant intensity variations. The ANN models, trained using PCA scores from blood samples with varied hormone concentrations, achieved high predictive accuracy with coefficients of determination (R² > 87.71%) and low root mean square error (RMSE < 0.6436). Elevated hormone levels were initially observed in injected rats, gradually declining over time, with results aligning closely to those obtained via ELISA kits. This work showed that the SERS method can provide rapid (~2 minutes), hormone-independent detection with minimal sample preparation. This approach demonstrated the SERS method's potential for rapid, reliable hormone detection and with customized calibration may be applied in sports doping control, clinical diagnostics, and broader biomedical research.
血液中的生长激素（GH）和睾酮（TE）水平是纯净体育中人体健康和表现的关键指标。偏离正常水平可能预示着严重的健康问题，如生育问题、癌症或垂体肿瘤。现有的这些激素检测方法通常成本高昂、耗时且缺乏便携性。在本研究中，我们探索了表面增强拉曼光谱（SERS）在区分注射了外源性GH、TE以及两种激素的斯普拉格-道利（SD）大鼠的血样与未注射大鼠血样方面的潜力。然后，使用经过训练和验证的人工神经网络（ANN）模型来预测血液中这些激素的水平。在785 nm激光激发下，采用SERS方法对注射了GH、TE、两种激素的SD大鼠以及未注射大鼠的血样进行分析。从注射了GH和TE以及未注射大鼠的血液中记录的拉曼光谱显示出激素特异性的谱带强度变化。此外，主成分分析（PCA）显示注射后谱带强度随时间的变化，表明激素诱导的生化改变。特别是，所有组在1378 cm⁻¹附近、GH组在658 cm⁻¹以及GH和TE组在798 cm⁻¹处的拉曼谱带显示出显著的强度变化。使用来自不同激素浓度血样的PCA分数训练的ANN模型，具有较高的预测准确性，决定系数（R²＞87.71%）且均方根误差较低（RMSE＜0.6436）。最初在注射大鼠中观察到激素水平升高，随后随时间逐渐下降，结果与通过酶联免疫吸附测定（ELISA）试剂盒获得的结果非常吻合。这项工作表明，SERS方法能够在样品制备最少的情况下提供快速（约2分钟）、与激素无关的检测。这种方法证明了SERS方法在快速、可靠的激素检测方面的潜力，并且经过定制校准后可应用于运动兴奋剂检测、临床诊断以及更广泛的生物医学研究。
