A Regression Framework for Predicting Cognitive Decline in Frontotemporal Dementia using Recurrent Neural Networks

Frontotemporal dementia (FTD) is a progressive neurodegenerative disorder with a diverse range of symptoms, including personality changes, behavioral disturbances, language deficits, and impaired executive functions. FTD has three main subtypes: behavioral variant FTD, non-fluent variant primary progressive aphasia, and semantic variant primary progressive aphasia. While there has been extensive research on detecting FTD, there is a limited exploration of the progression of FTD severity in patients over time. FTD typically manifests at a younger age, occurring between 40 and 65 years, than any other dementia forms. Therefore, detecting specific FTD subtypes early on becomes more feasible when assessing disease severity in the initial stages. This study aims to forecast an individual's future cognitive status using their neuropsychiatric inventory. This inventory consists solely of neuropsychological test scores related to FTD markers collected from one or more time points. We proposed and applied a regression framework with Encoder-Decoder Long-Short-Term-Memory (ED-LSTM) model to the data from the Frontotemporal Lobar Degeneration Neuroimaging Initiative (FTLDNI/NIFD) comprising longitudinal data of 288 participant's 918 instances. This study represents the first comprehensive exploration of forecasting individuals' FTD markers (cognitive scores) over four years into the future. We compared the performance of the proposed model with two baseline recurrent neural network models (LSTM and Simple RNN). The results indicate that the suggested model outperforms other implemented models when considering mean absolute error and root mean square error performance metrics.Clinical relevance- This study aims to provide valuable insights into the early identification and prognosis of cognitive decline in individuals with FTD. This could contribute to more timely and targeted interventions, improving patient outcomes and enhancing the overall management of FTD.