This paper proposes an approach to learning system dynamics when the time between observations is non-uniform and when the observations from different objects are not temporally aligned. There were concerns that a large body of related work was ignored.