Reconstruction of missing cardiovascular signals using adaptive filtering

Here we introduce a robust method for filling in short missing segments in multiparameter ICU cardiovascular data inspired by the “PhysioNet/Computing in Cardiology Challenge 2010: Mind the Gap”. Using the signals' history we identified the interconnections between the signals in the form of composite IIR transfer functions. Assuming that the connections do not vary in time, we managed to reconstruct the missing signals using the yet available parallel measured signals and the transfer functions. Since this assumption holds only for short timeperiod, we restricted the identification of the transfer function to segments prior the missing signal shorter than 30 seconds. Our results are promising on the challenge dataset. We concluded that this approach can be efficient in reconstructing and even detecting missing or corrupted cardiovascular signals or other type of datasets with several modalities and strong interconnections between them.