PROJECT SUMMARY Perioperative neurocognitive disorders (PNCD) affect about 25% of patients in the period following surgery, and can persist for months or years in 10% of geriatric surgical patients. The presence of acute cognitive disturbances post-surgery increases the risk of patients eventually developing dementias such as Alzheimer's disease. Poor cognitive outcomes lead to longer hospital stays, decreased quality of life, and increased morbidity and mortality. Unfortunately, about half of elderly individuals require a surgical procedure at some time in their remaining years, and no interventions exist to prevent PNCD because the etiology is unclear. One of the main challenges in identifying the factors leading to chronic cognitive impairment is the lack of routine, comprehensive cognitive testing in the surgical care plan. This shortcoming is in part due to the lack of mobile, easy-to-use cognitive testing platforms. To establish the perioperative biomarkers contributing to cognitive decline, we will (1) integrate routine, comprehensive cognitive testing pre- and post-surgery, and (2) build a database and an analysis platform to mine this multidimensional dataset. Together, this aims to yield accurate models to pre-identify patients at risk and create targeted therapeutic interventions. We propose to build the foundational infrastructure for a precision medicine approach in geriatric surgical patients. In the R21 phase, we will build a novel comprehensive database of demographic and risk factor questionnaire responses, banked blood specimens, intraoperative electroencephalography (EEG), and inclusive cognitive testing. These data will be collected throughout the patient interaction, from the preoperative appointment through a year following the surgical procedure and available to other research teams. We will incorporate cognitive testing and collect large-scale data in the geriatric surgical setting, establishing a new precedent for subsequent multidisciplinary studies. This structure will afford us the opportunity to accurately track cognitive decline towards chronic conditions, such as dementia. In the R33 phase, we will develop an analysis platform capable of mining this multidimensional dataset. This phase will include EEG analysis and deep immune profiling using mass cytometry. Layered on these analyses we will build innovative machine learning tools to identify features and interactions contributing to both acute and chronic PNCD pathology. Our novel machine learning tools use prior knowledge to refine feature selection, thus addressing a common challenge faced by clinical research studies (having many measurements in a limited patient population), and will thus be of broad interest to other clinical research projects.