Abstract:
Noise and missing data corrupt medical data. In medical settings, missing patterns occur in bursts due to sensors being off or data being collected unevenly. Sequential variational autoencoders (VAEs) are proposed to model heterogeneous medical data records with bursty missing data. The Shi-VAE extends VAEs to sequential streams of data with missing observations. We compare our model to state-of-the-art solutions in an intensive care unit (ICU) database and a passive human monitoring dataset. We include the cross-correlation between the ground truth and the imputed signal in our analysis because standard error metrics like RMSE cannot assess temporal models. Shi-VAE outperforms the state-of-the-art medical record method, GP-VAE, in both metrics and computational complexity.
Note: Please discuss with our team before submitting this abstract to the college. This Abstract or Synopsis varies based on student project requirements.
Did you like this final year project?
To download this project Code with thesis report and project training... Click Here
