In this study we explored the possibility of detecting blood loss in patients with hemorrhage symptoms (N=14) from photoplethysmographic (PPG) signals collected with pulse oximeters (PO) at forehead, ear and finger sites. We used variable frequency complex demodulation (VFCDM) technique to estimate amplitude modulations in heart rate frequency range (AMHR) of PPG signals. We determined the trend of these AMHR values over time, and used it to classify each patient's recording. The obtained results were compared to the clinical classifications made by physicians at the UMass Medical Center, which were considered as references. The accuracy of our algorithm was about 79%. These are the preliminary results of an ongoing study, and we foresee that this device and technique can be applied in battlefield and combat casualty care.