Diagnosis of opioid use disorder with high sensitivity and specificity by advanced computational analysis of Fourier transform infrared spectroscopy

Abstract

Opioid use disorder (OUD), which is a chronic disease and widespread public health problem results in death of the patient in the early period. The consumption of opioid depending on the use disorder time, affects blood biochemical parameters. OUD diagnosis can be clinically determined from biochemical values and with the screening tools in substance abuse. However, long-term OUD causes multiple pathological and physiological changes in many organs. Opioid initiated diseases, such as changes in tissue level and molecular level can be detected with advanced diagnostic methods after a long period of usage. It is important to diagnose these changes without causing organ dysfunction. The tests for the diagnose of chronic effects of OUD are expensive and time-consuming. In this study, we used Fourier transform infrared (FTIR) spectroscopy to discriminate the whole blood samples of opioid-addicted patients (n = 20) from healthy persons (n = 14) with an average addiction period of for 8 +/- 5.8 years. We purposed to compare whole blood vibrational results of OUD and HC. In order to identify absorption bands on structures of proteins, lipids and nucleic acids, their corresponding absorbance of band spectra were measured. We had chosen following bands 2959, 2931, 1646, 1550, 1453, 1400, 1314, 1243, 1080, 1079, 1542, 1045, 1467 to identify peak heights and to compare peak height ratios such as H1646 to H1550, H1079 to H1542, H2959 to H2931, H1453 to H1400, H1314 to H1243, H1045 to H1467, H1080 to H1550. Statistically, the p value of group heights were significantly different (p < 0.001). As a chemometric method, features extracted by principal component analysis (PCA) and then classified by linear discriminate analysis (LDA) and support vector machine (SVM) to determine the spectral data of disorder zones. The confidence of specificity and sensitivity and accuracy were obtained as 93.33%, 85%, 80.57% in raw data, and 100%, 100% and 100% in the second derivative respectively. Our research illustrates that whole blood analyses by FTIR, on the selected peak heights may discriminate pathological and healthy structural changes induced long term opioid use disorder.Thus, we have demonstrated that infrared spectroscopy can provide a simple and available diagnostic test for OUD patients.