Improving Emergency Department Efficiency by Patient Scheduling Using Deep Reinforcement Learning

. 2020 Mar 27;8(2):77. doi: 10.3390/healthcare8020077

Set:
$K_{g}$	A set of medical resource type $g$
$℘$	A set of patients
Indices:
$i$	individual patient
$j$	individual treatment
k	individual resource
$g$	medical resource type $g = 1, \dots, G$
Parameters:
$a r_{i}$	Arrival time of patient $i$
$a u_{i}$	Acuity level of patient $i$
$w a_{i}$	Weighted acuity level of patient $i$
$n_{i}$	Total number of treatments for patient $i$
$p t_{i j}$	Actual processing time of treatment $j$ for patient $i$
$H$	Large number
$O_{i j}$	Treatment $j$ for patient $i$
$c_{i}$	Weight for wait time of patient $i$
Decision variable:
$w t_{i j}$	Waiting time for $O_{i j}$ to be started
$s t_{i j}$	Commencing time for $O_{i j}$
$e t_{i j}$	Completion time for $O_{i j}$
$x_{i j k}$	1 if treatment $j$ for patient $i$ is assigned to resource $k$ , 0 otherwise