Internatonal Journal of Computer Trends and Technology (IJCTT) volume 7 Number 4 Nov 204 Patent Health Care nalyss based on NFIS Sugeno Model Maylvaganan M #, Rajeswar K *2 # ssocate Professor & Dept.of Computer Scence &PSG College of rts and Scence, Combatore,Inda. # Research Scholar(Ph.D) Karpagam Unversty, Dept.of Computer Scence, Combatore,Inda. bstract The man research formulaton of the problem deals wth how blood pressure affects the dfferent parts of the human body wth the use of the proposed fuzzy logc controller. The proposed work focus about daptve Neuro Fuzzy Interface System (NFIS) depends on fuzzy logc controller to dagnose the varous level of health rsk factor value whch s aggregated wth Blood Pressure, Pulse Rate and Kdney functon based on varous Input Parameters. In ths paper, Fuzzy Logc crcut was developed wth 2 s Complement n full adder usng the nput such as Blood Pressure value taken from Systolc and Dastolc value, Pulse Rate and GFR value. Due to ncrease n blood pressure measurement values, such as systolc and dastolc values how the kdney and other parts of our body functon values are heavly affected are also dscussed n ths paper. The proposed NFIS system s valdated wth blood pressure data set values usng Mat Lab Fuzzy Tool Box, and smulated output analyse the rsk factor value of a human beng. Keywords Pulse rate; Glomerular Fltraton Rate; daptve Neuro Fuzzy Interface System; Systolc; Dastolc; membershp functon. I. INTRODUCTION Blood carres oxygen and food to all the body cells. Blood flow n a human body wll be regular f there s no dsturbance by the rsk factors. The rsk factors become a great rsk, due to nexactness of blood pressure and cholesterol. Hence, to capture and overcome the unexpected, unknown and uncertan level of rsk factors and kdney functons are dentfed usng the blood pressure levels. Usng the powerful mathematcal logc called fuzzy logc s appled to mtgate the rate of mortalty by kdney falure, n developng countres lke Inda. Blood pressure s one of the most pre-domnant dseases n the modern world and focuses not only a medcal but a socal problem. Fuzzy approaches can play an mportant role n data mnng []-[3]. Blood pressure s summarzed by two measurements, systolc and dastolc, whch depend on whether the heart muscle s contractng (systole) or relaxed between beats (dastole). Hence, n ths paper, t s planned to mplement an NFIS system to overcome the uncertantes of dfferent types of Blood pressure ranges. Ths proposed work helps to estmate the pulse rate and rsk factors value and Kdney functon value of a human beng. II. LITERTURE REVIEW- FUZZY LOGIC HELTH CRE SYSTEM The exstng work publshed upon Fuzzy Expert System n medcal cases encountered a broad spectrum ncludng feedbacks, applcatons, nnovatons, conceptual studes and development of medcal tools[8]-[9]. In the early tme reported studes, the need, specalzaton, potental, requrements of fuzzfcaton and strateges for mplementng medcal doman expert systems s dscussed. The expert system layers are dsplayed for automatc creaton of fuzzy expert systems wth applcable to certan dseases and decson support systems. The doman that are developed for mplementng a fuzzy expert systems wth reference to specfc dseases, common usage medcal dagnostc systems as well as for consultng of personal health[5][0]. The exstng methodology are proposed for the mplementaton and mprovng performance of a rule based dagnostc decson support systems related to medcal dagnoss. III. PROPOSED METHODOLOGY Problem Statement: Usng the NFIS fuzzy logc, controller can be desgned and expermented wth a set of avalable knowledge to estmate the rsk factor value of a human beng wth the use of sample data case related to varous range of blood pressure flow wth systolc and dastolc values and kdney functon GFR values are represented. Therefore, n ths research work, t s planned to propose a desgn methodology usng NFIS ISSN: 223-538 http://www.jcttjournal.org Page 200
Internatonal Journal of Computer Trends and Technology (IJCTT) volume 7 Number 4 Nov 204 controller to control the lfe rsk factor values and analyse the dfferent organs that are affected due to Layer 4 : The nodes n ths layer are adaptve blood pressure, whch n turn reduces the rate of and perform the consequent of the rules: Usng mortalty.nfis represent Sugeno Tsukamoto ND, OR operator to nput the value mf value to fuzzy models. the rule, the rule was analyse n to the engne and Mathematcal Model for NFIS - Layer : sngle truth value to be determne for the patent rsk Collect the Current status of the patent regardng stage. blood pressure by set of facts and represented usng O doman knowledge. Blood pressure are taken as a 4, w f w ( p x q y r ) parameter range s dvded nto four fuzzy sets, The parameters n ths layer ( p, q, r ) are to be namely, low, medum, hgh and very hgh. determned and are referred to as the consequent The output of each node for systolc and dastolc parameters. Blood pressure range value s represented by: Layer 5 : There s a sngle node here that computes the overall output: Defuzzfcaton s the process of O, ( x) for,2 convertng the fnal output of a fuzzy system to a crsp value. The health rsk are determnes the level O, B ( y) for 3,4 2 of severty of depresson rsk gven the nput Where O, ( x) s essentally the membershp grade varables. for x and y. µ represents the any membershp grade functons. For a zero-order Sugeno model, the w f O output level z s a constant (a=b=0).the output 5, w f w level z of each rule s weghted by the frng Pseudo code for NFIS strength w of the rule. Typcal membershp functon s followed by the formula, Step : start the process µ x = () 2b x-c + a where a, b, c are parameters to be learnt. These are the premse parameters. Layer 2 : The pulse rate was analyzed by gven Systolc and Dastolc values. Every node n ths layer s fxed. Ths s where the t-norm s used to sum OR the membershp grades s represented by O w ( x) ( y),,2 2, B Layer 3 : It contans fxed nodes whch calculate the rato of the frng strengths of the rules: When the IF (condton) part of the rule matches a fact, the rule s fred and ts THEN (acton) part s executed. The condton s check blood pressure s mf, pulse value represents mf2, and kdney functon s representng as mf3. If x s, O 3, w and y s B THEN f px q y r w w w 2 Step 2: State the varables Blood pressure, kdney functon GFR and pulse values. Step 3: Calculate the pulse rate and each w s scaled nto n the normalzaton layer. Step 4 : Each w weghs the result of ts lnear regresson f n the functon layer, generatng the rule output. Rule : IF X s low ND Y s low ND Z s low ND W s very hgh then P s Low Rule 2 : IF X s low ND Y s normal ND Z s low ND W s very hgh then P s very Low Rule 3 : IF X s low ND Y s low ND Z s low ND W s very hgh then P s medum Rule 4 : IF X s low ND Y s very hgh ND Z s low ND W s hgh then P s short Step 5: Estmate the NFIS membershp functons for dfferent set of blood pressure ranges. µagg(p) = max{mn(2/5, µl(p)), mn(/2, µvl(p)), mn(2/5, µm(p)), mn(/2, µs(p))} Membershp functon for calculatng blood pressure range s estmated as ISSN: 223-538 http://www.jcttjournal.org Page 20
Internatonal Journal of Computer Trends and Technology (IJCTT) volume 7 Number 4 Nov 204 Membershp functon for calculatng blood pressure range s estmated as. daptve neuro Knowledge rule base µ L(x) = 20-x/60 for60 <= x <= 20 80-x/40 for 40 <= x <= 80 µ N(x) =x-60/60 for 60<=x <= 20 µ BP(x) 40-x/20 for 40<=x <= 60 x-80/5 for 80<=x <= 85 95-x/0 for 85<=x <= 95 ffected organs BP BP symptoms 2. fuzzy Inference systems Step 6: Each rule output s added n the output layer w f by range value of w f = w Step 7 : Compute the patent rsk stage usng ND, OR operator and Fuzzy Inference System(FIS) defuzzfcaton whch r(t) = d(t)+s(t) Step 8: stop the process 3. NFIS :Blood Pressure Range Knowledge cquston Fuzzfcaton & new doman knowledge IV. PROPOSED NFIS RCHITECTURE The proposed adaptve neuro fuzzy generally comprses the followng steps: system ()Selecton of dfferent blood pressure range from database as nput ()mplementng the approprate fuzzy membershp functons and operators () dentfy the accurate fuzzy nference system (v)formulaton of knowledge Blood pressure rule base Data base, Blood pressure (systolc dastolc value), Kdney functon GFR value,pulse rate De Fuzzfcaton Control sgnal Error sgnal Output sgnal Fg. NFIS rchtecture Fgure represents the archtecture of a proposed system showng the data flow through the neuro fuzzy nterface system. It conssts of manly knowledge rule base, knowledge acquston, blood pressure range, kdney functon and pulse rate knowledge base and nference engne modules. In the adaptve neuro fuzzy system, the user can get ISSN: 223-538 http://www.jcttjournal.org Page 202
Internatonal Journal of Computer Trends and Technology (IJCTT) volume 7 Number 4 Nov 204 the detals such as blood pressure range, affected From fg.2 and fg.3 represents the member functon organs and symptoms from the knowledge base to of blood pressure are constructed for fndng the rsk nference. The knowledge acquston module factor. The nference engne compares each rule allows user to seek the nputs as well as to buld the stored n the knowledge base wth facts contaned n new doman knowledge. The nput varables are the database. When the IF (condton) part of the fuzzfed whereby the membershp functons rule matches a fact, the rule s fred and ts THEN defned on the nput varables are appled to ther (acton) part s executed. The condton s check actual values, to determne the degree of truth for blood pressure s mf, pulse value represent mf2, each rule antecedent. The fuzzy rule base s kdney functon s represent as mf3. The nference engne uses a system of rules to make decsons characterzed n the form of f-then rules n whch through the fuzzy logcal operator and generates a the antecedents and consequents nvolve lngustc sngle truth value that determnes the outcome of the varables.. The followng fgure represents the rules. Ths way, they emulate human cogntve archtecture of the proposed adaptve neuro fuzzy process and decson makng ablty and fnally they nterface system. represent knowledge n a structured homogenous and modular way. V. EXPERIMENTL RESULTS ND DISCUSSIONS Under the process fuzzfcaton was handle for frst step a proper choce of process state varables and control varables s essental to characterzaton of the operaton of a fuzzy logc control system. In decson makng logc, If..Then rule base follow for measurng the membershp values obtaned. Fnally the defuzzfcaton s processed for combnng the fuzzy outputs of all the rules to gve one crsp value Pressure B Kdney C D Pulse P X 0 X Y= 0 or Fg.2. Member functon of Blood Pressure Fg.4. Logc gate for fndng the rsk rate From the fgure 4 and 5 descrbe, and B are pressure value whch represent as X0, pulse values are derved from gven pressure value whch represent X, C and D are Kdney functon value whch represent X2. The pulse rate was analyzed by gven Systolc and Dastolc values. Fnally rsk factor was analyzed by pulse rate and analyss by GFR rate of kdney functons. Fg.3. Fnal plot of Member functon - Blood Pressure Fg.5 analyse the rule usng logcal operator ISSN: 223-538 http://www.jcttjournal.org Page 203
Internatonal Journal of Computer Trends and Technology (IJCTT) volume 7 Number 4 Nov 204 VI. Result and Dscusson [9]Mahfouf M, bbod MF & Lnkens D, survey of fuzzy montorng and control utlzaton n medcne,rtfcal Intellgence Usng Sugeno FIS method the patent health rsk was found from the gven nput of lngustc varable of Blood pressure, Pulse rate and kdney functons. Usng If... Then rule and nference strateges are chosen for processng the rule base to determne the rsk factor among the blood pressure, kdney functon and pulse rate by logcal decson makng analyss. Through the defuzzfcaton, fuzzy system provdes an objectve process of rsk factor, also to vew the surface vew of the rsk determnaton usng smulaton. Concluson The man focus of ths research work s the desgn of a fuzzy controller to analyse the rsk factors value of the human beng. The novel dea has been mplemented and tested usng Mat Lab successfully. The use of ths approach s contrbuted to medcal decson-makng and the development of computerasssted dagnoss n medcal doman and dentfes the major rsk of the patent n earler. The valdaton of the fuzzy logc controller s tested wth blood pressure data and t s used to determne the rsk factor value of an human beng whch s affected by blood pressure. The most nnovatve feature s the smulated of the rule vewer whch shows the fuzzfcaton and defuzzfcaton usng ND, OR gate. Rule base s also an predomnant for our system to add or delete the rules. In future, ths system s extended to analyse the rsk factor values of dfferent organs affected by blood pressure. REFERENCES [] braham. Rule-based Expert Systems. Handbook of Measurng System Desgn, John Wley & Sons, 909-99, 2005. [2]dlassng, K.P., Kolarz, G., Schethauer, W. 985. Present State of The Medcal Expert System CDIG-2. Methods of Informaton n Medcne, 24: 3-20. [3]gbonfo, Oluwatoyn C., jay, dedoyn O. Desgn of a Fuzzy xpert Based System for Dagnoss of Cattle Dseases, Internatonal Journal of Computer pplcatons & Informaton Technology. [5] dlassng, K.P. Fuzzy set theory n medcal dagnostcs, IEEE Trans. On Systems, Man, and Cybernetcs,Vol. 6 260-264. [6]Constantnos Koutsojanns and Ioanns Hatzlygerouds, FESMI: Fuzzy Expert System for Dagnoss and Treatment of Male Impotence, Knowledge-Based Intellgent Systems for Health Care, pp 06 3,2004. [7] Rahm F, Deshpande, Hossen, Fuzzy Expert System For Flud Management In General naesthesa, Journal of Clncal and Dagnostc Research, pp 256-267,2007. [8]Hamdreza Badel, Mehrdad Sadegh, Elas Khall Pour, btn Hedarzadeh, Glomerular Fltraton Rate Estmates Based on Serum Creatnne Level n Healthy People, Iranan Journal of Kdney Dseases, Volume 3,Number, 2009. logc n Medcne 2, pp 27-42, 200. [0] Sesng, Hstory of Medcal Dagnoss Usng Fuzzy Relatons,. Fuzzness n Fnland'04, -5, 2004. []Tomar, P.P., Saxena, P.K. 20. rchtecture For Medcal Dagnoss Usng Rule-Based Technque. Frst Int. Conf. on Interdscplnary Research & Development, Thaland, 25.-25.5. ISSN: 223-538 http://www.jcttjournal.org Page 204