The authors of this paper have been studying the system-scientific modeling of the inpatient flow in an acute-care hospitalusing the recorded order log data for the movement of patients over the observation period. We have thus far published theresults for obstetric \cite{Takagi-Kanai-Misue} and neonatal \cite{Kanai-Takagi} patients.The order log in the University of Tsukuba Hospital during the two fiscal years 2010--2011 is used as a data source. In the present study, we deal with pediatric patients who are transferred over several wards from their admission untiltheir discharge. We calculate the number of admissions, ward transfers, and discharges during the two years of observation.Special attention is paid to the numbers of patients who are hospitalized before and after the finite observation periods.We consider the total number of days on which patients stay in each ward over the observation period, which is called the\textit {patient-days} (PD) as a measure of clinical load brought by patients on the hospital. We count the numbers ofpatients who arrive to, depart from, and stay in each ward over the observation period and show that these numberssatisfy the \textit{conservation law} for the patient flow in a hospital. We also confirm an identity relation among the meannumber of patients staying in each ward per day, mean number of patients who arrive to that ward per day, and mean numberof days a patient stays in that ward, called the mean \textit{length-of-stay} (LoS), which may be thought of as adiscrete-time, finite-horizon version of \textit{Little's law} in queueing theory. We propose a \textit{patient residence chart}to visualize the movement of each patient during their hospitalization. We present a method to classify the paths of patients,i.e., a sequence of pairs (ward number, length-of-stay on that ward), from which we calculate a variety of measures ofpatient movement.