A deep learning model for text regression. What it does is, it takes the speech sample of people of two groups (Dementia patient and Healthy control) from Pitt corpus (Cookie Theft Description Task) of DementiaBank dataset. It then models the data longituidinally to predict the clinical score that is Mini Mental State Exam (MMSE) score which reflects their cognitive status so that the model can predict the score their MMSE score of next visit based on based on the currebt speech sample and previous visit score.
How to Run:
Run the script model_longituidinal.py under Dementia-Detection\Dementia-Detection\karnekal_niu_bansal_2018\karnekal_niu_bansal_2018 folder.