Figure 4. DT model for predicting postoperative motor function and prediction performance using the internal test.

Regarding the DT, the elliptical node represents the condition, and the rectangular node represents the predicted outcome, i.e., stable or decline in postoperative motor function. Starting from the root node (i.e., “Amplitude of last measured MEP ≤98 μV”), the corresponding value of the patient is compared with the condition in the root node. If the patient’s last measured MEP amplitude is less than 98 μV, the next node is “Preoperative motor weakness”; alternatively, the algorithm jumps to the “Changes in intraoperative MEP response” node. The patient’s attribute values continue to be compared with other internal nodes of the tree until a rectangular node is reached, at which point the outcome prediction is obtained. The amplitude of the last measured MEP is the root node of the DT. It rates the function of the corticospinal tract and is crucial for predicting postoperative motor function. In patients with the last measured MEP amplitude >98 μV, changes in intraoperative MEP responses are of significance, and the corresponding postoperative motor function will be stable if “Changes in intraoperative MEP response” decline to less than 50%. In patients with the last measured MEP amplitude below 98 μV, preoperative motor weakness plays a significant role, and postoperative motor function would decline if the patient shows preoperative motor weakness.

DT: decision tree; MEP: motor evoked potential