第十七卷第二期 2015 年 6 月 (pp.263~290) CKD a, * a a b c a b c Chronic Kidney Disease, CKD CKD OWL Web Ontology Language Web-based 36 Developing A Dietary Con

Size: px

Start display at page:

Download "第十七卷第二期 2015 年 6 月 (pp.263~290) CKD a, * a a b c a b c Chronic Kidney Disease, CKD CKD OWL Web Ontology Language Web-based 36 Developing A Dietary Con"

建于
5 years ago
Views:

1 第十七卷第二期 205 年 6 月 (pp.263~290) a, * a a b c a b c Chroic Kidey Disease, CKD CKD OWL Web Otology Laguage Web-based 36 Developig A Dietary Cosolatio System for Chroic Kidey Disease Patiet Care Yu-Liag Chi a Pei-Rog Chuag a Tsag-Yao Che a Tsair-Yua Chag b Mei-Lig Liao c a Departmet of Iformatio Maagemet, Chug Yua Christia Uiversity b Departmet of Iformatio Maagemet, Mig Chua Uiversity c Nutritio Departmet, Ladseed Hospital Abstract The care for patiets with chroic kidey disease (CKD) is closely related to the patiet s daily diet maagemet. The type, quatity, ad utriet cotet of the patiet s * maxchi@cycu.edu.tw ( H MY2) DOI: 0.688/JEB.205.7(2).04 Jue

2 food itake eed to be strictly cotrolled. CKD dietary cosultatio ivolves multiple kowledge sources ad is thus a kowledge-itesive task. We approach the task from the perspective of kowledge system developmet by costructig related problem solvig kowledge ito a kowledge model. I additio, cosiderig kowledge sharability ad reuse, we adopt a otological methodology ad use OWL (Web Otology Laguage) as the modelig laguage for the developmet of a dietary cosultatio kowledge system. I the early stage of the study, we have excluded the iterferece factors of complicatios & comorbidities from the kowledge model for scope defiitio. The costructed kowledge base is the developed ito a Web-based kowledge system for better user experiece. I terms of experimetal evaluatio, data from 36 case patiets are used for testig. The system s iferece capability was tested, respectively, i cliical coditio estimatio, theoretical food group recommedatio, ad key utriet restrictio. The dietitia-advised diet lists are the added to examie the differeces betwee the theoretical ad actual itake values. The evaluatio results show that, excludig the iterferece factors ad certai o-medical reasos, the system has achieved the research goal of CKD dietary cosultatio. For future studies, the problem solvig scope ca be expaded to implemet a more comprehesive dietary cosultatio system. Keywords: Chroic Kidey Disease (CKD), Kowledge-based System, Problem Solvig, Otology, Sematic Rules. Chroic Kidey Disease, CKD CKD () (2) Servigs (3) (4) Natioal Kidey Foudatio, 2002; Jue 205

3 McCullough et al., 2002 Otology Kowledge-based System, KBS CKD Problem Solvig KBS CKD Co-morbidity Complicatio W3C 2 Sematic Web OWL 3 OWL-based Otology Gaies, 203; Shadbolt, 203 KBS Ausseac-Gilles ad Gado, 203 Kasse, 203 Breuker, 203 KBS Welty ad Guario, 200 OWL-based Otology Class Property Istace KBS Gómez-Pérez ad Bejamis,999; Welty ad Guario, 200 OWL-based KBS Cairó ad Guardati, 202; Ferádez-López et al., 203; Muse, 203 OWL-based KBS OWL Beimel ad Peleg, 20; 2 3 Otology Otology W3C OWL Web Otology Laguage Jue

4 García-Castro ad Gómez-Pérez, 200 O Coor et al., 2008 ; Wieliga, 203 CKD Taxoomy OWL Sematic Rules OWL-based KBS () Domai Otology is-a (2) Task otology has-a Compositio Subsumptio Aggregatio (3) Web-based 慢性腎臟病的相關背景 Richards, Levey et al., 2003 CKD Levey ad Coresh, 202 CKD CKD 4 Glomerular Filtratio Rate, GFR 4 The Natioal Kidey Foudatio Jue 205

5 CKD estimated GFR egfr Myers et al., 2006; Steves et al., 2006 egfr 5 MDRD Modificatio of Diet i Real Disease Serum Creatiie, Scr egfr egfr 90 60~89 3A 3B 45~59 30~44 5~29 5 WTO 6 Calorie Protei Phosphorous Potassium Sodium 5 CKD Richards, 20 CKD NKF-KDOQI 203 CKD KBS 7 Basal Metabolic Rate, BMR Jue

6 2.2 食物與營養攝取 McCullough et al., a 9 20 MyPlate 0 My Pyramid 6 Grais, Protei Food, Daily, Fruits, Vegetables, Oils b CKD CKD CKD CKD ~3A 3B~5 ~3A 3B~5 2.3 本體技術與開發知識系統的挑戰 Otology Plessers et al., 2007 Welty ad Guario, 200 Cotrolled Vocabulary García-Castro ad Gómez-Pérez, 200 XML OWL Web Otology Laguage OWL Jue 205

7 Horrocks et al., 2003 OWL Descriptio Logic Motik et al., 2009 OWL Protégé OWL-based Otology Geari et al., 2003 SWRL Sematic Web Rule Laguage 2 Horrocks et al., 2005 SWRL OWL-based Corsar ad Sleema, 2006 SWRL protégé SWRLTab 4 SWRL JESS Java Expert System Shell Lezcao et al., 20 Declarative Kowledge Procedural Kowledge Lai, 2007 Wieliga, 203 Gómez-Pérez ad Bejamis, 999 OWL-based Otology Class OWL-based KBS Cairó ad Guardati, 202 ; Ferádez-López, et. al, 203; Muse, KBS Domai Otology Task Otology Guario, 997; Bejamis ad Perez, 2000 OWL-based Otology Ausseac-Gilles ad Gado, 203 KBS Motta, 203 Muse 203 Expert s Head Protégé 2 SWRL Jue

8 axiom OWL class, property, idividual is-a, has-a, restrictios Beimel ad Peleg, 20; García-Castro ad Gómez-Pérez, 200 Busiess Logic O Coor et al., 2008 ; Wieliga, 知識系統在醫學資訊的應用 Abu-Haa et al Protégé Roveta ad Rosu 2009 Prolog Meemasuk ad Chatraporchai 203 Che et al. 202 NCPM / - PEM 3. OWL-based KBS Axiom Logical Axiom No-Logical Axiom OWL class, property, idividual SWRL 270 Jue 205

9 OWL-based KBS 3. 問題分析 (A)~(C) ()~(4) () egfr (A) (B) (a) egfr MDRD egfr (b) BMR (c) egfr CKD (d) Nurtiet Restritio, NR () p NR p N stage calorie protei phosphorous potassium sodium w p NRp = ( N' stage, Calorie, wp N' stage, Pr otei, N' stage, Phosphorus, N' stage, Potassium, N' stage, Sodium ) () =( NCalorie, NPr otei, NPhosphorus, NPotassium, N ) Sodium (e) Food Servigs, FS 2 ~3A, 3B~5 7 6 (2) FS (p, i, s) p i Stage S FS(,, ) ( S, S, S, S, S p i stage grais meat milk vegetable fruit Soil uts, & ) (2) Jue

10 (B) () (A) CKD (3) (C) (2) (4) (2) (A) (C) (a) (3) f i,calorie *s i i calorie (s i ) ( f * s ), ( f * s ), ( f i, Calorie i i, Protei i i, Phosphorus i i= i= i= ( f * s ), ( f * s ) i, Potassium i i, Sodium i= i= i * s ), (3) (f) (4) f i,grais,servigs i Grais Servigs ( f ), ( f ), f i, grais, servigs i, meat, servigs iv, egetable, servigs i, Potassium i= i= i= i= ( f ), ( f ) i, fruit, servigs i, oil& uts, servigs i= i= ), ( f * si ), (4) (3) p Case (p, utriets) 272 Jue 205

11 (5) (3) () Case ( p, utriets) ( ( ficalorie, * si ) NCalorie), ( ( fi, Pr otei * si) NPr oei ), i= i= = ( fi, Phosphorus * si ) NPhosphorus ), ( fi, Potassium * si) NPotassium ), i= i= ( ( fi, Sodium * si ) NSodium ) i= (5) (4) p Case (p, food_group) (6) (4) (2) Case ( p, food _ group) ( ( fi, grais, servigs ) Sgrais ), ( ( fi, meat, servigs ) S ), meat i= i= = ( ( fi, vetetable, servigs ) Svegetable),( fi, milk, servigs ) S milk ), i= i= ( ( fi, fruit, servigs ) Sfruit), ( ( fi, oil& uts, servigs ) Soil& uts ) i= i= (6) 3.2 建置知識模型之領域本體 is-a 3. 2 CKD Protégé OWL-based Otology 2 Protégé 3 CLASS BROWSER INSTANCE BROWSER INDIVIDUAL EDITOR ()~(3) CKD -3a 200 Jue

2 Protégé OWL editor () 4 (a) 200 2700 7 Kcal_200 (b) 2. 5.

12 2 Protégé OWL editor () 4 (a) Kcal_200 (b) (c) 5 (d) (e) 6 (2) CKD 3 (a) CKD 6 has_ stage, has_egfr_mi, has_egfr_max 3 CKD egfr egfr (b) CKD 5 has_ (c) CKD 7 has_ 274 Jue 205

6 has_ (3) (a) (b) (c) 3 Excel Protégé DataMaster xlsql-y8 3 3.

13 6 has_ (3) (a) (b) (c) 3 Excel Protégé DataMaster xlsql-y 建置知識模型之任務本體 OWL-based protégé Asserted Property Iferred property Domai Rage Data Property Object Property CKD Jue

14 5 CSVData () CKD 5 (A)~(A0) ( ) (2) 7 has_ perso 6 CKD (B)~(B6) (3) 4 2 has_perso, has_ 2 _ _ (C), (C2), (D)~(D5), (E)~(E6) (4) 3 (5) CSVData Nutritio, Servig Atom (atom atom )à Cosequece has_stage) egfr CKD atom ~atom 7 { atom (x) atom 2 (x) has_egfr (y) 276 Jue 205

15 atom 3 CKD (s) atom 4 (s) egfr has_egfr_mi (b) atom 5 (s) egfr has_egfr_max (c) atom 6 SWRL (y) (b) atom 7 SWRL (y) (c) Cosequece (s) (x) has_stage } x, y, s, b, c Protégé SWRL-Tab SWRL (A) (?x) has_egfr (?x,?y) CKD (?s) has_egfr_mi (?s,?b) has_egfr_max (?s,?c) swrlb:greatertha(?y,?b) swrlb:lessthaorequal (?y,?c) has_stage(?x,?s) (A) 29 3 () (a) has_egfr egfr (A2) (A3) MDRD egfr (A2) (?x) has_cr (?x,?y) has_ (?x,?z) has_ (?x, Male ) swrlb:pow (?a,?y, -.54) swrlb:pow (?b,?z, ) swrlb:multiply (?c,?a,?b) swrlb:multiply (?M_as,?c, 86) has_egfr (?x,?m_as) (A2) (b) (has_ ) (A4) (A5) BMR (A4) (?x) has_ (?x,?w) has_ (?x,?t) has_ (?x,?a) has_ (?x, Male ) has_ (?x,?z) swrlb:multiply (?a,?w, 3.7) swrlb:multiply (?a2,?t, 5) swrlb:multiply (?a3,?a, 6.8) swrlb:add(?b,?a, 66) swrlb:add (?c,?b,?a2) swrlb:subtract (?C_Male,?c,?a3) swrlb:multiply (?as,?c_male,?z) has_ (?x,?as) (A4) Jue

16 (c) () 4 (A7)~A(0) (A7) (?x) has_ (?x,?w) has_stage (?x,?s) CKD (?c) has_stage (?c,?s) has_ (?c,?p) swrlb:multiply (?as,?p,?w) has_ (?x,?as) (A7) (2) (2) (B)~( B6) (B) (?x) has_perso (?x,?p) (?p) has_ stage (?p,?s) has_ (?p,?k) CKD (?y) has_stage (?y,?s) has_ (?y,?k) has_ (?y,?as) has_ (?x,?as) (B) (3) CKD SQWRL 3 CSV 4 (a) (3) (C) 5 CSV 4 3 SQWRL(Sematic Query-Ehaced Web Rule Laguage) 是一個 SWRL 為基礎的語義查詢語言, 提供類似於 SQL 的操作, 主要用於補充 SWRL 的運算能力, 例如分類集合運算 4 CSV(Comma Separated Value) SQWRL OWL CSV OWL 278 Jue 205

17 (?x) has_perso (?x,?x) has_case_ame(?x,?x2) has_ (?x,?y) (?y) has_food (?y,?f) has_servigs (?y,?) p:has_food_ame (?z,?f) p:has_ Calorie (?z,?k) swrlb:multiply (?a,?k,?) sqwrl:makeset (?s,?a) sqwrl:groupby (?s,?x) sqwrl:sum (?as,?s) sqwrl:select (?x2,?as) sqwrl:columnames ( CSV_NAME, CSV_CALORIE ) (C) (b) (4) (C2) (?x) has_perso (?x,?x) has_case_ame (?x,?x2) has_ (?x,?y) has_food_group (?y,?g) has_food_group_ame (?g,?g) has_servigs (?y,?) sqwrl:makebag (?s,?) sqwrl:groupby (?s,?x,?g) sqwrl:sum (?as,?s) sqwrl:select (?x2,?g,?as) sqwrl:columnames ( CSV_NAME, CSV_FOOD_GROUP, CSV_SERVINGS ) (C2) (c) (5) (D)~(D5) _ (D) (?x) has_perso (?x,?x) has_case_ame (?x,?x2) (?x) has_ (?x,?y) Nutriet. Nutriet (?z) CSV_NAME (?z,?x2) CSV_CALORIE (?z,?z) swrlb:subtract (?as,?z,?y) swrlb:strigcocat(?as, :,?y, :,?z, :,?as) _ (?x,?as) (D) (d) (6) (E)~(E6) _ (E) Jue

18 (?x) has_perso (?x,?x) has_case_ame (?x,?x2) (?y) has_perso (?y,?x) has_ (?y,?y) Servigs.Servigs (?z) CSV_NAME (?z,?x2) CSV_FOOD_GROUP (?z, ) CSV_SERVINGS (?z,?z) swrlb:subtract (?as,?z,?y) swrlb:strigcocat (?as, :,?y, :,?z, :,?as, ) _ (?x,?as) (E) 開發 Web-based 知識系統 Web-based 4 Java Protégé, SWRL rule egie bridge, JESS API JSP Apache Tomcat Web-based OWL-based iput () ivoke Java-based (Protégé API, JESS API, JENA, SWRL rule egie bridge, etc.) respose JESS (2) OWL-bascd Otology 4 Web-based () 66 77cm 66Kg Cr Jue 205

(2) 5 CKD 4 egfr 27.6 4 (a) - (A6)~(A0) 279.92, 39.

19 (2) 5 CKD 4 egfr (a) - (A6)~(A0) , 39.6 g, 528 mg, 3000 mg, 3000 mg (b) - (B)~(B6) 3.5, 4.0,.0, 4.0, 3.0, 6.0 (c) - (C) (D)~(D5) 0.8% -7.%, -5.9%, -43.8%, -68.5% 5 CKD Jue

20 (d) - (C2) (E)~(E6) 實驗設計與評估 20~202 CKD () CKD (2) CKD (3) (4) 36 2 CKD % % 282 Jue 205

21 2 00% 00% 00% 8% 89% 6. CKD 36 8% 89% OWL-based () 2.3 (2) OWL-based KBS OWL Jue

22 Diabetes CKD CKD a b Abu-Haa, A., Coret, R., de Keizer, N., Crubézy, M., & Tu, S. (2005). Protégé as a vehicle for developig medical termiological systems. Iteratioal Joural of Huma-Computer Studies, 62(5), Ausseac-Gilles, N., & Gado, F. (203). From the kowledge acquisitio bottleeck to the kowledge acquisitio overflow: A brief Frech history of kowledge acquisitio. Iteratioal Joural of Huma-Computer Studies, 7(2), Beimel, D., & Peleg, M. (20). Usig OWL ad SWRL to represet ad reaso with situatio-based access cotrol policies. Data & Kowledge Egieerig, 70(6), Bejamis, V. R., & Perez, A. G. (2000). Kowledge System Techology: Otologies ad problem-solvig methods. Retrieved February 2, 204, from l/usr/richard/pdf/kais.pdf Breuker, J. (203). A cogitive sciece perspective o kowledge acquisitio, Iteratioal Joural of Huma-Computer Studies, 7(2), Cairó, O., & Guardati, S. (202). The KAMET II methodology: Kowledge acquisitio, kowledge modelig ad kowledge geeratio. Expert Systems with Applicatios, 39(9), Che, Y., Hsu, C.-Y., Liu, L., & Yag, S. (202). Costructig a utritio diagosis expert 284 Jue 205

23 system. Expert Systems with Applicatios, 39(2), Corsar, D., & Sleema, D. (2006). Reusig JessTab rules i Protégé. Kowledge Based Systems, 9(5), Ferádez-López, M., Gómez-Pérez, A., & Suárez-Figueroa, M. C. (203). Methodological guidelies for reusig geeral otologies. Data & Kowledge Egieerig, 86, Gaies, B. R. (203). Kowledge acquisitio: Past, preset ad future. Iteratioal Joural of Huma-Computer Studies, 7(2), García-Castro, R., & Gómez-Pérez, A. (200). Iteroperability results for Sematic Web techologies usig OWL as the iterchage laguage. Web Sematics: Sciece, Services ad Agets o the World Wide Web, 8(4), Geari, J. H., Muse, M. A., Fergerso, R. W., Grosso, W. E., Crubézy, M., Eriksso, H., Noy, N. F., & Tu, S. W. (2003). The evolutio of Protégé: a eviromet for kowledge-based systems developmet. Iteratioal Joural of Huma-Computer Studies, 58(), Gómez-Pérez, A., & Bejamis, V. R. (999). Applicatios of Otologies ad Problem- Solvig Methods. AI-Magazie, 20(), Guario, N. (997). Uderstadig ad buildig, usig otologies. Iteratioal Joural of Huma-Computer Studies, 46(2-3), Horrocks, I., Patel-Scheider, P. F., & Va Harmele, F. (2003). From SHIQ ad RDF to OWL: The makig of a web otology laguage. Web Sematics: Sciece, Services ad Agets o the World Wide Web, (), Horrocks, I., Patelscheider, P., Bechhofer, S., & Tsarkov, D. (2005). OWL rules: A proposal ad prototype implemetatio. Web Sematics: Sciece, Services ad Agets o the World Wide Web, 3(), Kasse, M. (203). A promisig pheomeo of ope data: A case study of the Chicago ope data project. Govermet Iformatio Quarterly, 30(4), Lai, L. F. (2007). A kowledge egieerig approach to kowledge maagemet. Iformatio Scieces, 77(9), Lezcao, L., Sicilia, M. -A., & Rodríguez-Solao, C. (20). Itegratig reasoig ad cliical archetypes usig OWL otologies ad SWRL rules. Joural of Biomedical Iformatics, 44(2), Levey, A. S., & Coresh, J. (202). Chroic kidey disease. The lacet, 379(98), Levey, A. S., Coresh, J., Balk, E., Kausz, A. T., Levi, A., Steffes, M. W., Hogg, R.J. Jue

24 Perroe, R. D., Lau, J., & Ekoya, G. (2003). Natioal kidey foudatio practice guidelies for chroic kidey disease: evaluatio, classificatio, ad stratificatio. Aals of iteral medicie, 39(2), McCullough, M. L., Feskaich, D., Stampfer, M. J., Giovaucci, E. L., Rimm, E. B., Hu, F. B., Spiegelma, D., Huter, D. J., Colditz, G. A., & Willett, W. C. (2002). Diet quality ad major chroic disease risk i me ad wome: movig toward improved dietary guidace. The America Joural of Cliical Nutritio, 76(6), Meemasuk, P., & Chatraporchai, C. (203). O the Developmet of Nutritio Iformatio Systems for Kidey Disease Patiets. Iteratioal Joural of Database Theory & Applicatio, 6(3), Motik, B., Cueca Grau, B., Horrocks, I., & Sattler, U. (2009). Represetig otologies usig descriptio logics, descriptio graphs, ad rules. Artificial Itelligece, 73(4), Motta, E. (203). 25 Years of Kowledge Acquisitio. Iteratioal Joural of Huma- Computer Studies, 7(2), Muse, M. A. (203). The kowledge acquisitio workshops: A remarkable covergece of ideas. Iteratioal Joural of Huma-Computer Studies, 7(2), Myers, G. L., Miller, W. G., Coresh, J., Flemig, J., Greeberg, N., Greee, T., & Eckfeldt, J. H. (2006). Recommedatios for Improvig Serum Creatiie Measuremet: A Report from the Laboratory Workig Group of the Natioal Kidey Disease Educatio Program. Cliical Chemistry, 52(), 5-8. Natioal Kidey Foudatio (NKF) (2002). Cliical practice guidelies for chroic kidey disease: evaluatio, classificatio, ad stratificatio. Retrieved February 2, 204, from classificatio_ stratificatio.pdf O Coor, M. J., Shakar, R. D., Nyulas, C., Tu, S. W., & Das, A. K. (2008). Developig a Web-Based Applicatio usig OWL ad SWRL. AAAI Sprig Symposium: AI Meets Busiess Rules ad Process Maagemet, Plessers, P., De Troyer, O., & Casteley, S. (2007). Uderstadig otology evolutio: A chage detectio approach. Web Sematics: Sciece, Services ad Agets o the World Wide Web, 5(), Richards, M. (20). The kidey medical utritio therapy-yesterday ad today. Nutritio i cliical practice, 26(2), Roveta, E., & Rosu, G. (2009). The diagosis of some kidey diseases i a small prolog 286 Jue 205

25 Expert System. I 3rd Iteratioal Workshop o Soft Computig Applicatios, Shadbolt, N. (203). Kowledge acquisitio ad the rise of social machies. Iteratioal Joural of Huma-Computer Studies, 7(2), Steves, L. A., Coresh, J., Greee, T., & Levey, A. S. (2006). Assessig Kidey Fuctio- Measured ad Estimated Glomerular Filtratio Rate. New Eglad Joural of Medicie, 354(23), Welty, C., & Guario, N. (200). Supportig otological aalysis of taxoomic relatioships. Data & Kowledge Egieerig, 39, Wieliga, B. J. (203). Reflectios o 25 years of kowledge acquisitio. Iteratioal Joural of Huma-Computer Studies, 7(2), Jue

26 has_ Data Strig has_ Data it has_ Data float has_ Data float has _ Data float has _Cr Data float has _stage Object CKD (A) has_ egfr Data float (A2)(A3) has_ Data float (A4)(A5) has_ Object (A6) has_ Data float (A7) has_ Data float (A8) has_ Data float (A9) has_ Data float (A0) has _perso Object has_ Data float (B) has_ Data float (B2) has_ Data float (B3) has_ Data float (B4) has_ Data float (B5) has_ Data float (B6) has _perso Object has_ Object (C)/(C2) _ Data Strig (D)~(D5) _ Data Strig (E)~(E6) 288 Jue 205

27 2 CSVdata Nutritio Servig has_food Object has_food_group Object has_servigs Data float CSV_FOOD_GROUP Object Strig CSV_NAME Object Strig CSV_CALORIE Data double CSV_PHOSPHOROUS Data double CSV_POTASSIUM Data double CSV_PROTEIN Data double CSV_SERVINGS Data double CSV_SODIUM Data double CSV_FOOD_GROUP Data Strig CSV_NAME Data Strig CSV_SERVINGS Data double Jue

28 290 Jue 205

基于改进的TF-IDF算法的聚焦主题

基于改进的TF-IDF算法的聚焦主题 Joural of Computer Applicatios ISS 1001-9081 015- -10 计算机应用,015, ( ): - CODE JYIIDU http://www.joca.c 基于 TF-IDF 改进算法的聚焦主题网络爬虫王景中 1, 邱铜相 ( 北方工业大学计算机学院, 北京市,100144) (* 通