Anything related to a variety of fields from statistics, data mining, machine learning or related domains. Critical issues in medical diagnosis that often go ignored or cannot be handled due to some lack in the current technology.
Sounds quite interesting. This problem might involve several domains. I think it might be a good start to have the vital signs of all the wounded people which might be a key parameter in this kind of a decision making scenario.
Still, I would be interested to know what strategy to look for resource management considering the fact that no two individuals have the same level of injury.