Table 1.

Samples of synthetic chief complaints using greedy sampling and beam search, along with key variables from the corresponding records

Age	Gender	CCS code	CCS description	Chief complaint	Greedy sample	Beam samples (k= 5)
70–74	M	660	Alcohol-related disorders	“alcoholic beverage consumption today”	“pt admits to drinking alcohol”	“etoh? “etoh consumption” “found on street” “found on the streef,’ “fond on street admits to drinking”
50–54	F	2607	E Codes: Motor vehicle traffic	“mv c pt denies complaints”	“pt was rear ended”	“mvc? “mva? “pt was rear ended? “pt was in a car? “pt was rear ended in mvc”
20–24	F	128	Asthma	“shortness of breath on and off x 3 week”	“shortness of breath”	“wheezing? “cough? “chest pain? “chest tightness? “shortness of breath”
25–29	F	661	Substance-related disorders	“od narcan given by ems”	“found on street”	“od? “as per ems? “as per ems patient was found unresponsive? “as per ems patient was found on the street? “as per ems patient was found in the street”
85–89	F	205	Spondylosis; intervertebral disc disorders; other back problems	“m5440 lumbago with sciatica unspecified side i10 needs dialysis”	“lower back pain m549”	“m542 cervicalgia i 10? “back pain m545 low back? “back pain m545 low back pain? “back pain m545 low back pain i 10? “back pain m545 low back injury i 10”
50–54	F	98	Essential hypertension	“generalized chest pain and blurred vision x few days no travel see above complaint”	“elevated bp”	“elevated bp? “elevated blood pressure? “high blood pressure? “elevated bp no travel? “elevated blood pressure no travel”

We include the authentic chief complaint for comparison, but we omit from the table hospital code, mode of arrival, disposition, month, and year to increase patient anonymity