Automatic generation of new rules for processing synthetic events using ...

EIEEMU

106/ MM ‘ |]|]|]|] I I SERVER 206 200 \ PROCESSING \‘ UNIT 210 202 it 208 216 236 \ \ / / / GRAPHICS MAIN AUDIO PROCESSOR <:> NB/MCH <'L'> MEMORY ADAPTER

so 204 1; U H 240 \ 238 BUS BUS / SB/ICH 1 H H H U_> ROM

USBAND KEYBOARD DEK ODROM flggggg OTHER gggggg ANDMOUSE MODEM PORTS ADAPTER / / / / \ \ \ \ 226 230 212 232 234 220 222 224

FIG. 2

COHORT 306 SYSTEM / 300 COHORT T‘ APPLICATION FEATURE _ DATA MINING ‘ ' APPLICATION DATABASE | / K ‘ 308 310 V V / CLINICAL CLINICAL TEST CONTROL 302/ INFORMATION COI-IORT SELECTION SYSTEM PROGRAM FIG. 3 400 \ A 418 C “ 414>\_/ /’ \ 408 / \l 416 >’T\\ 406‘’1 CLUSTER\\< / CLUSTER \ \ 1 / B I 2 \\_// \ \ / AGE /<1/¢_‘>\ Q 410 412 2 ~ 422 Y S. a W’ \\ / \ I CLUSTER I CLUSTER 43/ \ 4 \ 3 I \\ N \ V 420 _ / Z \ \ __ / / x SEVERITY OF SEIZURE FIG. 4A @

502 \

504 /

‘
7

400
\ A
418
414 f_ _
/
/ \
/ \
>’ T T \ \ 406 “IOLUSTER
/ CLUSTER \ \ 1
\

\ / AGE /(IF/\\ 416 Q 410 422 Y / // I CLUSTER 41,- \\ 4 I 424 \ 3 I \ \< 420%//N2 \\___// X SEVERITY OF SEIZURE T FIG. 4B % 500 / O PATENT 506 508 510 POPULATION RECORDS / CLUST/ERED / CLUSTERING A PATIENT A FEATURE ® ALGORITHM ' RECORDS ' SELECTION TREATMENT COHORT RECORDS

FIG. 5

600 602 604 606 CLUSTER \ / ASSIGNMENT TREATMENT / CRITERIA g CLUSTERING COHORT HLTER ' ALGORITHM RECORDS \ CLUSTERED RECORDS FROM TREATMENT FIG 6 COHORT ' \ 610 CLUSTER Z90 7O4\ ASSIGNMENT CRITERIA II 702 POTENTIAL CLUSTERWG RE(C_0JRSDrg|I=ER|:0M CONTROL COHORT SCORING pO‘|'EN‘|'|A|_ RECORDS A‘-GOREHM CONTROL COHORT / \ 706 708 FIG. 7 (3 80° / 802 \ TREATMENT COHORT CLUSTER 806 808 ASSIGNMENTS \ / / \ / INTEGER OPTIMAL PROGRAMMING = CONTROL <3 SYSTEM COHORT POTENTIAL / CONTROL COHORT CLUSTER ASSIGNMENTS FIG‘ 8