In a dataset with time-series, that is dependant on a given input, which the time-series are given only on an irregular cycle of 10-12 time steps that makes lots of missing observations what is the best way to impute the missing observations?
p.s: Data is the measurement of a bio index(continues) every 10 -12 days and input is daily taken drug dosage(discrete and without missing values) and the desired predictive model is IOHMM(HMMs with inputs).