I don't think there's a reliable way to detect such a state using just logic devices (I can think of two schmitt triggers, one with input pulled a bit up, the other - down, by appropriate voltage, so when they output two different values, something is wrong, but that's NOT reliable). So you need to attach ADC input to your "unreliable" data line and probe the voltage. But prevention is easier, cheaper and gives better results (the device just works, as opposed to "we know why it doesn't work") than detection, so why detect a fault when we can prevent it altogether? And of course, it still CAN happen, but then you need to be really paranoid to try to detect it in-application.