Introduction: Effective sepsis treatment relies on early diagnosis. As yet there is a lack of reliable and fast detection methods for sepsis aiding the treating physician to differentiate non-infectious and infectious stimuli as origin for inflammation and shock. Host specific immune response could be helpful for identifying infection as cause for organ dysfunction. Raman spectroscopy is an emerging method in the field of biomedical research with diagnostic potential. This technique provides information on the biochemical features of the cells in a non-destructive and label-free manner, which in turn enables discrimination of different cell types as well as their functional states. In Raman spectroscopy laser light gets scattered upon interacting with the biomolecules and the inelastically scattered light contains information about the molecular vibrations that are very specific to the molecule. Hence, complex information about the cell and its activation status is captured by the Raman spectra. Previously, Raman spectroscopy has been applied for leukocyte phenotyping [2, 3] and it has been demonstrated in an ex vivo study of splenocytes isolated from endotoxemic mice possibility to follow dynamic changes in the T-lymphocytes [4]. In the current work we have applied Raman spectroscopy as a new diagnostic platform for leukocytes screening to obtain infection-specific information and for sepsis detection. Objectives: The study aims at validating Raman spectroscopy signature of isolated blood leukocytes for fast, reliable and precise identification of sepsis patients and to stratify patients having infection or non-infectious shock. Methods: In this observational study (HemoSpec, DRKS-ID: DRKS00006265) 54 patients, with pre-defined criteria of (a) inflammation: patients with no proven infection, undergone elective surgery, (b) infection: patients with clinically or microbiologically proven infection, (c) sepsis: patients with organ dysfunction in accordance to sepsis-3 definition were recruited. Peripheral EDTA blood (4.9 ml) was collected for routine clinical diagnostic (Blood hemogram), biomarkers (IL6, procalcitonin [PCT], C-reactive protein [CRP], suPAR) and for Raman spectroscopy. Blood leukocytes were isolated from 500 ll blood by removing erythrocytes by lysis method followed by chemical fixation with 4% formaldehyde. The cells were coated on CaF2 and measured immediately. Raman spectral data analysis was performed with aid of chemometric analysis methods: canonical powered partial least square (CPPLS) analysis and logit regression analysis to build Raman spectral based model for patient stratification. Results: Raman spectroscopy allowed extracting phenotype information of leukocyte subtypes. The logit regressions of the Raman spectroscopic data shows good separation between patients having infection (AUC = 0.83) and patients with sepsis (AUC = 0.82). The major spectral differences were observed in the vibrational peak of proteins and nucleic acid. To calculate the added value of Raman spectroscopy, advanced statistical analysis was performed by employing CPPLS, which allows including demographic data of the patients as additional responses. All the models were validated by leave-one-patient-out cross validation, that allows robust evaluation of models, similar to testing on an independent data set. The CPPLS model for Raman spectroscopic data allowed discrimination of sepsis patients from the two other groups with a mean sensitivity of 72% (AUC = 0.82) being equivalent to the mean sensitivity obtained by using the biomarker panel consisting of IL6, PCT, CRP and suPAR. The mean sensitivity for sepsis detection was increased to 91% (AUC = 0.99) by combining Raman responses and the biomarkers. There was no change in the sensitivity for sepsis detection by including the qSOFA values. Similarly, we also show presence of infection can be identified with mean sensitivity of 80% (AUC =0.86) by using only Raman spectroscopic data and when combined with biomarkers, the sensitivity was increased to 99% (AUC = 0.97). In this study we demonstrate diagnostic capability of the Raman spectroscopy and show the added value of Raman spectroscopy for fast, accurate and reliable detection of presence of infection and sepsis. Conclusions: With the aid of chemometric methods, the Raman spectroscopy model allows identifying patients with reactive leukocytes leading to adverse immune responses manifesting as inappropriate responses, i.e. sepsis. Further, we show Raman spectroscopy brings in added value for qSOFA and biomarkers by increasing sensitivity for sepsis patient detection. Raman spectroscopy has high potential especially in a decentralized health care system for patients screening with suspected infection.