App-based assessment of memory functions in patients after transfemoral aortic valve replacement.

Background: Transfemoral aortic valve replacement (TAVR) is the standard treatment for elderly patients with aortic valve stenosis. Although safe and well-established, there is a risk of intraprocedural hemodynamic instability and silent cerebral embolism, which can lead to a decline in neurocognitive function and dementia. In clinical practice, comprehensive cognitive testing is difficult to perform. AI-assisted digital applications may help to optimize diagnosis and monitoring.

Methods: Neurocognitive function was assessed by validated psychometric tests using "∆elta -App", which uses artificial intelligence and computational linguistic methods for extraction and analysis. Memory function was assessed using the 'Consortium to Establish a Registry for Alzheimer's Disease' (CERAD) word list and digit span task (DST) before TAVR and before hospital discharge. The study is registered in the German Register of Clinical Trials (https://drks.de/search/de/trial/DRKS00020813).

Results: From October 2020 until March 2022, 141 patients were enrolled at University Hospital Heart Centre Brandenburg. Mean age was 81 ± 6 years, 42.6% were women. Time between the pre- and post-interventional test was on average 6 ± 3 days. Memory function before TAVR was found to be below average in relation to age and educational level. The pre-post TAVR comparison showed significant improvements in the wordlist repeat, < 0.001 and wordlist recall test of CERAD, < 0.001. There were no changes in the digital span test.

Conclusions: Despite impaired preoperative memory function before TAVR, no global negative effect on memory function after TVAR was detected. The improvements shown in the word list test should be interpreted as usual learning effects in this task.