Passive smartphone-based assessment of cognitive changes in neurosurgery

Clinical observations suggest dynamic alterations in behavior after brain surgery. While some alterations reportedly occur within days others gradually develop over several months. These alterations can be attributed to the pre-surgical impact of the diseased tissue, neuronal damage caused by the surgery, and subsequent plasticity. A key step towards a systems-level understanding of the brain-behavior relationships is to capture the dynamics of the cognitive alterations. Here, we first established in 38 healthy individuals that the day-to-day smartphone interactions can be used to inform on cognitive processing speed. Next, we analyzed the smartphone interactions in 12 patients undergoing intracranial tumor surgery, with postsurgical follow-up of up to a year. In healthy individuals, the speed of the touchscreen interactions was highly correlated to choice reaction times (R2 = 0.71) but less so to simple reaction times (R2 = 0.15) on Deary-Liewald tests. Touchscreen interactions slowed immediately after surgery but the post-surgical changes varied between patients. Data-driven models revealed the time-constant of the short-term postsurgical changes and the time taken to stabilize after the surgery. Furthermore, by using conceptually distinct types of touchscreen interaction speeds - i.e. unlocking time and app locating speed - we established that the post-surgical changes are domain-specific. Interestingly, in this small sample, the pre-surgical smartphone speeds were highly related to the speeds post-stabilization (R2 = 0.75 to 0.95). The proxy measures of cognition seamlessly captured on the smartphone can reveal postsurgical dynamics inaccessible to conventional testing. We propose that the transient cognitive alterations indicate the time-constrained influence of distinct neuronal mechanisms triggered by the surgery.