Fostering reproducibility and generalizability in machine learning for clinical prediction modeling in spine surgery

Current survival-prediction models (SPMs) were designed using historical data to predict future events. Over a dozen of SPMs exists in spine surgery, such as predicting survival in patients with spinal metastases or chordoma [1–5]. It cannot be overemphasized the importance of a robust methodology to construct a reliable SPM. However, random splitting the training and validation datasets, which was widely adopted as mentioned by Dr Azad et al., might not always be an ideal strategy in all clinical settings [6], especially in settings where medical treatment is evolving in a fast pace.