As a beneficial component with critical impact, computer-aided decision making systems have infiltrated many fields, such as economics, medicine, architecture and agriculture. The latent capabilities for facilitating human work propel high-speed development of such systems. Effective decisions provided by such systems greatly reduce the expense of labor, energy, budget, etc. The computer-aided decision making system for traumatic injuries is one type of such systems that supplies suggestive opinions when dealing with the injuries resulted from accidents, battle, or illness. The functions may involve judging the type of illness, allocating the wounded according to battle injuries, deciding the severity of symptoms for illness or injuries, managing the resources in the context of traumatic events, etc. The proposed computer-aided decision making system aims at estimating the severity of blood volume loss. Specifically speaking, accompanying many traumatic injuries, severe hemorrhage, a potentially life-threatening condition that requires immediate treatment, is a significant loss of blood volume in process resulting in decreased blood and oxygen perfusion of vital organs. Hemorrhage and blood loss can occur in different levels such as mild, moderate, or severe. Our proposed system will assist physicians by estimating information such as the severity of blood volume loss and hemorrhage , so that timely measures can be taken to not only save lives but also reduce the long-term complications as well as the cost caused by unmatched operations and treatments. The general framework of the proposed research contains three tasks and many novel and transformative concepts are integrated into the system. First is the preprocessing of the raw signals. In this stage, adaptive filtering is adopted and customized to filter noise, and two detection algorithms (QRS complex detection and Systolic/Diastolic wave detection) are designed. The second process is to extract features. The proposed system combines features from time domain, frequency domain, nonlinear analysis, and multi-model analysis to better represent the patterns when hemorrhage happens. Third, a machine learning algorithm is designed for classification of patterns. A novel machine learning algorithm, as a new version of error correcting output code (ECOC), is designed and investigated for high accuracy and real-time decision making. The features and characteristics of this machine learning method are essential for the proposed computer-aided trauma decision making system. The proposed system is tested agasint Lower Body Negative Pressure (LBNP) dataset, and the results indicate the accuracy and reliability of the proposed system.