Background: Perioperative Neurocognitive Disorders range from short term (postoperative delirium) through medium (delayed neurocognitive recovery) through to long term (mild and major postoperative neurocognitive disorders, sometimes called postoperative cognitive dysfunction) [1]. They are particularly prevalent in the older population, those with pre-existing cognitive impairments and those undergoing major or emergency surgery [2]. Multiple studies have demonstrated correlation between depth of anaesthesia measured using intra-operative processed EEG and postoperative neurocognitive outcomes [3]. Similarly, links between markers of systemic inflammation and neurocognitive outcomes have been explored using both targeted studies of candidate biomarkers and untargeted proteomics studies utilising high throughput multiplex assays [4]. As yet however, the link between systemic markers of inflammation, depth of anaesthesia and perioperative neurocognitive outcomes has not been the subject of prognostic modelling. Aims: The primary aim of this study will be to use supervised machine learning to build a prognostic model for postoperative delirium based on measures of systemic inflammation and depth of anaesthesia. The secondary aim of this study is to use supervised machine learning to build a prognostic model of postoperative cognitive dysfunction based on measured of systemic inflammation and depth of anaesthesia. Methods: This pragmatic, observational, pilot study will aim to recruit 50 participants aged over 65 years, undergoing non-emergency spinal surgery. Participants will undergo serum inflammatory biomarkers analysis immediately prior to, and the day following surgery and this data, together with metrics derived from intraoperative processed EEG, will be compared to delirium incidence over the first 5 postoperative days. Similarly, this data will also be used to predict cognitive decline between standardised cognitive tests completed prior to surgery, and those performed at 6 months postoperatively.