An Expert System Approach to Medical Region Selection for a New Hospital Using Data Envelopment Analysis

Transcription

1 usiess, 2010, 2, doi: /ib Published Olie Jue 2010 ( A Expert System Approach to Medical Regio Selectio for a New Hospital Usig Data Evelopmet Aalysis Chi-Tsai Li*, Chua Lee, Zhi-Ju Che Graduate School of Maagemet, Mig Chua Uiversity, Mig Chua, Taiwa, Chia. ctli@mail.ypu.edu.tw, clee@mcu.edu.tw, ivy.cute916@msa.hiet.et Received February 22 d, 2010; revised March 27 th, 2010; accepted May 1 st, ABSTRACT A appropriate medical regio must be selected before establishig a ew hospital. Oce established, a hospital may brig may medical doctors ad facilities to the area. Not oly ca the distributio of medical resources i that area be iflueced, but also competitio amog differet hospitals ca be ehaced. The govermet eeds to cosider the issue of medical resource distributio; therefore, medical regio selectio has importat policy implicatios. This study uses data evelopmet aalysis (DEA) to establish the effective idicators, ad also uses a expert system o the equality of medical resource distributio to idetify medical areas lackig medical resources ad where there is relatively less competitio. As a result, this study ca provide the ecessary iformatio to facilitate the choice of regio for a ew hospital. This method ot oly avoids assigig a hospital to a medical regio where there are surplus resources, but it ca also reduce the risk of excessive competitio. Keywords: Practice Locatio, Equality of Medical Resources Distributio, Data Evelopmet Aalysis, Effectiveess 1. Backgroud Prior to the establishmet of the Natioal Health Isurace system, there were 716 hospitals ad 61,105 acute beds i Taiwa. However, by 2006, the umber of hospitals was reduced by 193 to 523, whereas the umber of acute beds had icreased by 11,827 to 72,932 [1]. This developmet idicates the tedecy for hospitals to cotiue to expad to obtai higher medical paymets [2]. The govermet adopted a hads-off approach to physicia traiig ad hospital maagemet util the establishmet of the Departmet of Health, Executive Yua, i 1971 [3]. Taiwa is divided ito 17 differet healthcare districts, ad relevat criteria for assessig medical resources are also established by the Departmet of Health [4]. Accordig to the stadard developed by the Departmet of Health for medical districts without sufficiet medical resources (defied as districts with less tha 20 geeral sickbeds per 10,000 populatio), o medical districts lacked medical resources as of 2006, except the medical area of Yuli Couty, which probably had 25 geeral sickbeds per 10,000 populatio. Thus, most districts have over 30 geeral sickbeds per 10,000 populatio; half of these have a sufficiet amout of medical resources, with over 60 geeral sickbeds per 10,000 populatio. Hospitals established i districts with sufficiet medical resources immediately suffer ot oly excessive competitio i health-care markets, but also barriers to expasio or the establishmet of ew hospitals [3]. Moreover, the Departmet of Health has bee developig a ew proect called the Global Budget System sice 1998, i which aual total medical expediture is calculated based o medical-care paymets for each perso ad the actual umber of medical isurace obects [5]. Besides the paymet of total medical claims for geeral hospitals, paymet for detists, Chiese medicie doctors, ad family doctors exists o a differet basis. Therefore, if a iappropriate medical district is selected, oe might simultaeously face several problems; for example, greater competitio ad a lower average paymet value. Furthermore, sice medical services are ot mobile, ad patiets must atted hospitals i perso, careful medical district selectio has become a key issue for those establishig ew hospitals. Oce established, a hospital may brig may medical doctors ad facilities to the area, which strogly iflueces the distributio of medical resources, ad competitio amog differet hospitals ca be ehaced. Therefore, choice of the medical

2 A Expert System Approach to Medical Regio Selectio for a New Hospital Usig Data Evelopmet Aalysis 129 area for a ew hospital has importat policy implicatios. Whe the market is i balace, all doctors share the same effective demad. Similarly, it is wise for doctors to establish their practices i areas with a low desity of doctors per capita to equalize distributio [6]. Additioally, Newhouse et al. [7] foud that cities with smaller populatios could be cosidered oly after the umber of doctors i urba areas reaches a certai ratio. Ecoomic capability i the medical market may, therefore, profoudly affect doctors choices of where to locate their practices. Chiag [8] oted that the reaso remote areas have fewer doctors is ot because of the uequal distributio of medical resources, but rather because cosiderable room remais i cities for ew market etrats. Thus, more doctors are eeded to esure that remote areas have sufficiet medical resources ad services. The oly way to solve the problem of isufficiet umbers of doctors i remote areas is to icrease the supply of doctors. I Taiwa, for example, the Taiwaese govermet is tryig to provide more doctors to remote areas by icreasig the quota for medical studets ad also promotig the Medical Group Practice Ceter [3]. Because of this successful medical policy, the umber of doctors has icreased from 24,399, before the implemetatio of the health-care proect i 1994, to 34,864 i 2006; a icrease of 10,465 (about 43%) i 12 years [9]. Itesifyig competitio amog doctors i the medical market ca be a problem; thus, doctors must seriously cosider whether they should locate their practices i a medical district with a smaller umber of doctors to avoid excessive competitio. I ecoomics, the market cocetratio ratio is widely used to measure the degree of competitio [10], particularly the Herfidahl idex, which is used to calculate the square of the total occupacy rate of all hospitals i the market. Competitio icreases with decreasig Herfidahl idex [11]. I a medical district with a populatio of 10,000 people, whe the Herfidahl idex drops to 100 (from 1000), the umber of doctors icreases from 10 to 100, but the occupacy rate of each doctor remais uchaged; i.e., the degree of competitio icreases. Thus, capacity must be calculated before a ew hospital is established. Furthermore, the capacity itself, the umber of doctors, ad the amout of additioal medical resources i this market must also be cosidered. The quatity of medical resources i a medical district ca be measured by aalyzig whether it has a equal distributio of medical resources whe compared with other related medical districts. If a medical district suffers from a ufair distributio of medical resources, it might be a suitable locatio for establishig a ew hospital. Two methods are commoly adopted for measurig equality, amely the proportioal method ad the idex method [12]. Regardless of which method is used, umerous differet metrics or criteria must be cosidered i the data. Furthermore, uless a medical district exists havig its etire metrics superior to those of other districts, it will be extremely difficult to evaluate ad ustify districts which are superior to others, without usig ay weightigs or trasfer fuctios. Due to the limitatios of traditioal methods, the maor purpose of this study was to establish a Performace Idex usig Data Evelopmet Aalysis (DEA), i which the Performace Idex is adopted to evaluate the equality of distributio of medical resources i a medical district. The DEA model ca provide a obective ad fair solutio to the above problem without requirig ay presuppositios [13]. The fuctio of the expert system lies i offerig, udgig, explaiig, ad makig policy decisios to the specific field questio [14], ad compared with a huma expert, the expert system has several beeficial characteristics: o time restrictios, low cost of operatio, ease of distributio ad duplicatio, cosistecy, ability to deal with time-cosumig or complicated problems, ad ca be adapted to specific fields [15]. The questio of medical regio selectio for ew hospitals is complicated, ad thus a expert system ca perform better tha a huma expert. This study used data evelopmet aalysis (DEA) to establish the effective idicators ad to help choose the correct expert system. I particular, this study used the relevat medical care materials ad the peoples awareess of accessibility of medical treatmet i 2006, ad the applied the DEA method to idetify the equality of medical resource distributio i Taiwa, to uderstad which medical areas lack medical resources, have relatively less competitio, ad i which, therefore, it is appropriate to add medical resources. This study may provide the ecessary iformatio to facilitate the choice of regio for placemet of a ew hospital, avoidig placemet i a medical regio with a medical resource surplus, ad reducig the risks associated with excessive competitio. 2. Literature Review Five defiitios are most frequetly quoted for equity of health-care distributio [16-21], as follows: equality of choice sets, equality of access, equality of expediture, equality of eeds, ad equality of health. 2.1 Equality of Choice Sets This meas that all idividuals should be provided with idetical sets of choices [16,17]. This idea is similar to the theory of Pechasky ad Thomas [22] regardig availability of eough doctors, hospitals, ad equipmet. This study measures choice set equality based o the percetage of doctors ad acute beds; the umber of choice sets icreases with icreasig percetage. Whe medical care facilities are established i districts with fewer choice sets, Phelps spatial theory, which holds

3 130 A Expert System Approach to Medical Regio Selectio for a New Hospital Usig Data Evelopmet Aalysis that doctors would like to ru their practices i sparsely populated areas, is demostrated. Similarly, medical istitutes face reduced competitio as the umber of choice sets decreases. 2.2 Equality of Access This meas that all idividuals have equal access to medical care. Access ot oly idicates the coveiece of obtaiig medical care for all idividuals, but also demostrates the relatioship betwee people ad sites of medical care. For example, access ivolves distaces ad traffic problems [22]. Equality of access thus improves with decreasig distace. This study measures the equality of access based o the percetage of doctors, the legth of road, ad the percetage of the people able to obtai medical treatmet withi a acceptable time, with a higher value represetig higher accessibility. Actio [23], Coffey [24] ad Cauley [25] idicated that the time take to get medical treatmet iflueces cosumers medical demad. Whe the distace ad time to obtai medical treatmet are log, the the demad for medical treatmet is very low. Usig a questioaire survey, oe ca idetify the time delay that cosumers fid acceptable for obtaiig medical treatmet. Whe a medical istitute chooses a medical area with lower accessibility, it is i agreemet with D Aspremot et al. [26], who idicated that factory dealers try their best to be far away from each other, the so-called priciple of maximum differetiatio. I additio, by choosig a medical area with lower accessibility, cosumer demad will also icrease, ad there will ot be much competitio for medical facilities. 2.3 Equality of Expeditures This meas medical care is distributed fairly amog differet idividuals. Outpatiet services ad hospitalizatio charges are geerally icluded i medical expeditures. Two idicators, average expediture o outpatiet services ad average expediture o hospitalizatio, are used to measure expediture equality. A higher value idicates that each perso s expediture is higher. Oe does ot use the percetage of doctors (wester medicie) as a idicator of expediture o outpatiet services or the percetage of acute beds as a idicator of expediture o ipatiet services because these two idicators caot tell us the differeces that arise from regioal differeces i the umber of people seekig medical treatmet or differeces i each perso s hospitalizatio cost. O the other had, the average of each perso s outpatiet expeditures ad the average of each perso s ipatiet expeditures ca accout for the differece i the umber of people. I additio, after Taiwa implemeted the Global Budget System, patiet s expeditures after seekig medical advice were paid by the Bureau of Natioal Health Isurace [27], so the relatioship with the umber of patiets is dimiished [28]. Furthermore, i a medical regio with a large umber of doctors, more doctors icreasig demad, thus icreasig outpatiet umbers, which icreases each perso s average expediture o medical treatmet [29]. Thus, i medical regios with higher average idividual expeditures o medical treatmet, the umber of doctors will be higher, resultig i relatively high competitio. 2.4 Equality of Needs Equality of eeds meas that medical resources should be distributed based o public eed. Sometimes, ecessity ca be defied as obtaiig or maitaiig high-level medical services. Geerally, medical treatmet ca be cosidered as health-care provided based o the medical kowledge of doctors ad professioals [30,31]. I fact, whe the uit of medical service is used to assess the eed for medical treatmet, the commoly applied methods iclude times of outpatiet service, hospitalizatio, or surgery. Two idicators, percetage of doctors providig outpatiet services ad percetage of acute beds, are used to measure equality of eeds. A higher value represets a greater amout of medical resources that the medical demads obtai each time. Whe a hospital chooses a area with lower medical resources, ad the differece i the amout of medical treatmet is cosidered, the effective demad should be the same for each doctor [12]. Similarly, whe a medical istitute is established i a regio with lower medical resources, the competitio is relatively low for the medical demads of each perso. 2.5 Equality of Health This meas that the distributio of medical resources must result i all persos havig equal health status, as reflected by life expectacies, death rates, ifat death rates, ad rates of iury ad sickess. Owig to the elemets of care that ifluece health status, umerous other relevat physiological ad social factors ca be idetified [32]. Male ad female survival rates are two idicators used to measure health equality. A higher value represets better health status i a medical area. Sice oprofit hospitals do ot cosider maximizig profit their mai purpose, ad they obtai their fuds through charitable cotributios or govermet subsidies, oprofit hospitals ca always maage [12]. The oprofit hospital exists, as stated by Weisbrod [33], because there are some umet demads for medical treatmet; whe the service provided is less tha that required by society, people will orgaize a oprofit hospital i order to offer services to provide the eeded services. Thus, oprofit hospitals must cosider health equality as oe of the stadards used to determie where they should be established. A medical regio that has a worse health status eeds a oprofit hospital.

4 A Expert System Approach to Medical Regio Selectio for a New Hospital Usig Data Evelopmet Aalysis 131 The defiitio of the equality of distributio for medical resources idicates that o-profit hospitals should cosider all five dimesios. Sice for-profit hospitals do ot always have to assume a maor social resposibility, they oly eed to cosider the first four dimesios. 3. Methods The proportioal ad idex methods are the most commo methods of measurig equity. Whe the proportioal method is used to distribute health care, the medical isurace expeditures amog differet groups ca be adopted to compesate for the percetage differeces [34]. The idex method is the most frequetly quoted method to measure equality; for example Va Doorslaer et al. [35] ad Schalick et al. [36]. Wagstaff et al. [37] discuss three idices: the disease cetralizatio idex, the cetralizatio idex of medical treatmet expeditures, ad the idex of fairess. This permits us to measure the degree of iequity i medical resources i advace. However, regardless of the method used, both methods (proportioal ad idex) are oly capable of dealig with sigle targets. Whe several idices are measured simultaeously, as i the preset study, oe weighted set or alteratively a trasfer fuctio must be provided. Additioally, the weighted set ad the fuctio choice are cosidered subective. The DEA model provides a sigle obective ad fair method that does ot require settig default weights to solve problems ivolvig multiple obectives. Furthermore, the proposed method has good uit ivariace. The earliest DEA model uses efficiecy as the evaluatio basis [38]. decisio maagemet uits (DMUs) ( 1,, ) are accessed, ad m types of iput X (, x 1 T 2, y p) T x 2,, x m ) ad p types of output Y ( y 1, y, are produced. The model for assessig the o th DMU is as follows: Mi s. t. 1 1 θ, r, s o θ ε( o r x i r y r r 1 sio s ro p ro s, s ro y 0 i1 θ x io o ro m io s io ), i 1,, m, r 1,, p where θ o represets techical efficiecy, ad Equatio (1) is the iput-orieted CCR model. Cosequetly, the efficiecy of this equatio is also kow as iput based efficiecy. Furthermore, output ca also be take as the output-orieted CCR model, which ca be preseted as follows: p m p m 1 Max o ( sro sio) ( sro s io) r1 i1 o r1 i1 (1) s.. t r y s y, r 1,, p r ro o ro 1 rx s x, i 1,, m (2) i io io 1, r, s, s 0 o ro io where the value of o is termed the output based efficiecy. The relatio betwee iput ad output based 1 efficiecy ca be represeted aso. o Chag et al. (1995) explaied measuremet effectiveess by implemetig the CCR model. Whe the iput i Equatio (2) is set to X 1, the output-orieted CCR model with a costat iput should be: p m o sro sio r1 i1 Max ( ) s.. t r y s y, r 1,, p r ro o ro 1 1 r s 1, i 1,, m io, r, s, s 0 o ro io The effectiveess is the reciprocal of efficiecy, o, 1 by calculatig Equatio (3) amely o. That is, o whe all DMU iputs are costats, the output-orieted DEA model equals the relative efficiecy measured based o outputs. Accordig to its meaig i maagemet, it ca be explaied as value the target uscrupulously. Uscrupulousess here does ot mea igorig efficiecy but istead stresses the macro-perceptio for measurig evirometal chage ad developig a strategy for seekig log-term efficiecy [39]. 4. Research Process This study implemeted the DEA method by aalyzig the equality of distributio of medical resources usig four (for-profit hospitals) ad five (oprofit hospitals) dimesios, respectively. Furthermore, to avoid losig the meaig of performace value, i which too may DMUs, resultig from performace assessmet, cotribute to a performace value of 1, every effort should be made to esure that the relatioship betwee the umber of DMUs ad the umber of iput-output items meets the experiece priciple: amely, the umber of DMUs is three times that of iput-output items [40]. However, i this study, the DMU cotais 16 medical districts (with Keelug City belogig to the same medical district as Taipei City ad Taipei Couty), ad the evaluatio idex cotais eight items for for-profit (3)

5 132 A Expert System Approach to Medical Regio Selectio for a New Hospital Usig Data Evelopmet Aalysis hospitals ad te items for oprofit hospitals. Therefore, the ratio idex ca be cosidered the criterio for evaluatig the developmet of each medical district, where Table 1 illustrates the defiitio of each idex. Although the experiece priciple is ot always met, the maor goal of this study was still to determie which medical districts have isufficiet medical resources. It is extremely satisfyig to obtai a aalysis result i which few DMUs have a performace value of 1. This study used the DEA to build the structure for a expert system of assessmet, as i Figure 1. The relevat data from the Health ad Vital Statistics, Statistical Yearbook of the Highway Bureau, ad Natioal Health Isurace Statistics for 2006 were gathered [1,41]. The percetage of people who ca obtai medical treatmet withi a acceptable time, which is part of the equality of access idicator, was determied from a questioaire survey coducted by the Departmet of Health, which idetified the time delay that cosumers fid acceptable for obtaiig medical treatmet. Equality of health survival rate of females survival rate of males Peghu Hwalia DEA Equality of eeds pereetage of acute bedsi total hospitalizatio pereetage of doctors providig outpatiet services Taidog health eare distributio equity Equality of expeditures average expediture o hospitalizatio average expediture o outpatiet services Equality of access Equality of choice sets pereetage of the people is aware of acceptable time of take medical pereetage of doetorsi the legth of road pereetage of doetors with traiig i westem medicie pereetage of acute bods Taoyua Yila Taipei Figure 1. Structure of the equality of medical resources distributio

6 A Expert System Approach to Medical Regio Selectio for a New Hospital Usig Data Evelopmet Aalysis 133 Table 1. Idex for measurig health care distributio equity Dimesio Idex Formula Equality of choice sets Percetage of doctors with traiig i wester medicie (te thousads) The umber of doctors populatio (perso) 1000 Percetage of acute beds (te thousads) The umber of acute beds populatio (perso) 1000 Percetage of doctors alog a 10-km legth of road The umber of doctors the legth of the road (km) 10 Equality of access Percetage of people who ca obtai medical treatmet withi a acceptable time Percetage of people who ca obtai medical treatmet withi a acceptable time the umber of people studied (persos) 100 Equality of expeditures Average expediture o outpatiet services (thousad dollars) Average expediture o hospitalizatio (hudred dollars) Ma-time i outpatiet services the average expediture o outpatiet services (dollars) 1000 populatio (persos) The umber of hospitalizatios the average expediture of hospitalizatio 100 populatio (persos) Equality of eeds Equality of health Percetage of doctors providig outpatiet services The umber of doctors ma-time i health isurace 1000 Percetage of acute beds i total hospitalizatio Male survival rate (1 death rate of male) 100 Female survival rate (1 death rate of female) 100 The umber of acute beds the umber of hospitalizatios 1000 Source: Departmet of Health, Health ad Vital Statistics [1] A telephoe questioaire survey was coducted, usig stratified percetage systematic samplig; 3,269 people were icluded i the sample. The equality idex of the distributio of medical resources for each medical district was calculated based o the formula i Table 1, ad the observed values are listed i Table 2. The Departmet of Health curretly divides Taiwa ito 17 medical areas ad 63 medical sub-areas. This divisio is based o the priciple of regardig adacet couties ad cities as sigle areas [4]. For example, Taipei city ad Taipei Couty are icluded i the same area. However, give that Keelug City is geographically surrouded by Taipei Couty, it seems sesible to also iclude it i the Taipei city ad couty medical area. Thus, this study aalyzed 16 medical areas rather tha 17. The Departmet of Health directly sigs isurace cotracts with hospitals/cliics, but ot with idividual doctors. Doctors are allowed to simultaeously work for hospitals/cliics both withi ad outside the NHS system, so it is extremely reasoable to use the Departmet of Health figure for the umber of hospitals. However, idividuals who participate i the Natioal Health Isurace system i specific areas are ot ecessarily residets of those areas, so a sigificat differece exists betwee the umber of idividuals isured uder the NHS ad the umber of residets i a area. Thus, to avoid misuderstadig, data gathered by the Departmet of Health shows a rather high ratio of the isured. 5. Discussios Based o Equatio (3) i this study, Table 3 shows the aalysis for the 16 medical districts. The closer the performace value calculated usig Equatio (3) is to 1, the fairer is the distributio of medical resources i that district, resultig i more sufficiet medical resources. However, whe the performace value is less tha oe, icludig situatios where it is much less tha 1, that district has a ufair distributio of medical resources, leadig to greater isufficiecy. Therefore, if the efficiecy value of all medical districts is ear 1, the distributio of medical resources i each district is fair. Table 3 shows that, whether from the perspective of for-profit or oprofit hospitals, medical districts with fair distributio of (i.e., sufficiet) medical resources i 2006 icluded Taipei (Taipei City, Taipei Couty, ad Keelug City), Yila Couty, Taichug (Taichug City, Taichug Couty) ad Peg-hu Couty. Based o the results of this study, it is ot wise to establish a ew hospital i a district with sufficiet medical resources, regardless of whether it is a for-profit or oprofit hospital. Establishig a ew hospital i such a area meas that it will face excessive competitio.

7 134 A Expert System Approach to Medical Regio Selectio for a New Hospital Usig Data Evelopmet Aalysis

8 A Expert System Approach to Medical Regio Selectio for a New Hospital Usig Data Evelopmet Aalysis 135 Table 3. Efficiecy values ad their raks of distributio i medical treatmet of each medical area i Taiwa Medical regio 2006 Efficiecy value For-profit hospital 2006 Rak of Medical resources 2006 Efficiecy value Noprofit hospital 2006 Rak of Medical resources Taipei (Taipei City, Taipei Couty ad Keelug City) Yila Couty Tyoyua Couty Hsichu (Hsichu City ad Hsichu Couty) Miaoli Couty Taichug (Taichug City ad Taichug Couty) Chaghua Couty Natou Couty Yuli Couty Jiayi (Jiayi City ad Jiayi Couty) Taia (Taia City ad Taia Couty) Kaoshog (Kaoshog City ad Kaoshog Couty) Pigdog Couty Taidog Couty Hwalia Couty Peghu Couty Additioally, if the performace values of idividual medical districts are all close to 1, from the perspective of oprofit hospitals, the distributio of medical resources i each medical district will be abudat, which seems clearly differet from the perspective of for-profit hospitals. This situatio applies because, regardless of the differeces betwee medical districts i terms of medical resources, these differeces have almost o effect o their survival rate. The Departmet of Health defies districts with less tha 20 sickbeds per 10,000 people as havig isufficiet medical resources. The preset study foud that, of the 16 medical areas, the fewest acute beds were foud i YuLi Couty (20.76 beds per 10,000 people) i Thus, o medical district i Taiwa has isufficiet medical resources. Notably, although Jiayi ad Peghu Couties were assessed as havig isufficiet medical resources, the performace values of PegHu Couties i 2006 were 1.00, suggestig that the medical resources i Peghu Couties are sufficiet. The govermet has regarded Jiayi couty as seriously lackig medical resources, but if oe examies Jiayi (Jiayi City ad Jiayi Couty) medical area, the medical resources have already bee improved. Careful ivestigatio shows that the umber of doctors i Jiayi Couty icreased 54% from 1997 to 2006, while the umber of emergecy sickbeds icreased 66%. Therefore, upgradig the umber of doctors ad hospital facilities appears highly effective i maitaiig sufficiet medical resources. Particularly after God s Help Hospital was expaded ad the Buddhist Tzu-Chi Dali Geeral Hospital ad Chag Gug Memorial Hospital were established, Jiayi Couty appears to have already achieved sufficiet medical resources. Peghu Couty is a separate islad located off the coast of Taiwa, ad the local govermet has bee cotiuously ad progressively promotig ad implemetig the Itegratio Medical Service Outsourcig Proect for Off-shore Islads Area of Peghu Couty, which provides itegrated medical services, icludig outpatiet services, emergecy treatmet, health-care plaig, home healthcare iterviewig services, health educatio programs, ad a patiet referral system based o outsourcig ad multiple solutios, thus helpig maitai sufficiet medical resources. I Peghu Couty, from 1997 to 2006, icreases were achieved of 30% i the umber of doctors, 30% i the umber of outpatiets, ad 65% i the average outpatiet service fee. From the perspective of for-profit hospitals, the raks of medical districts with the greatest isufficiecy of medical resources i 2006 were Hwalia Couty, Taia (Taia City ad Taia Couty), Natou Couty, Hsichu (Hsichu City, Hsichu Couty), Kaoshog (Kaoshog City ad Kaoshog Couty), Chaghua Couty, Taidog Couty, Pigdog Couty, Jiayi (Jiayi City ad

9 136 A Expert System Approach to Medical Regio Selectio for a New Hospital Usig Data Evelopmet Aalysis Jiayi Couty), Miaoli Couty, Tyoyua Couty, ad Yila Couty. From Figure 2, accordig to the first quartile (Q1 = ), secod quartile (Q2 = ), ad third quartile (Q3 = ) of relative efficiecy value, the 16 regios ca be grouped ito four areas. I the first group are those areas with sufficiet medical resources (Taipei (Taipei City, Taipei Couty, Keelug City), Yila Couty, Taichug (Taichug City, Taichug Couty) ad Peghu Couty). The secod group is those areas of isufficiet medical resources (Jiayi (Jiayi City, Jiayi Couty), Miaoli Couty, Taoyua Couty ad Yuli Couty) with a relative efficiecy value betwee the secod quartile (Q2, 50%) ad the third quartile (Q3, 75%). The third group is those areas of scat medical resources (Kaoshog (Kaoshog City, Kaoshog Couty), Chaghua Couty, Taidog Couty ad Pigdog Couty) with a relative efficiecy value betwee the first quartile (Q1, 25%) ad the secod quartile (Q2, 50%). The fourth group is those areas with serious isufficiecy of medical resources (Hwalia Couty, Taia (Taia City, Taia Couty), Natou Couty, ad Hsichu (Hsichu City, Hsichu Couty)), with a relative efficiecy value less tha the first quartile (Q1, 25%). Usig o-parametric methods, the Ma-Whitey test ad the Kruskal-Wallis test, it was foud that the medical efficiecy values amog the four groups was sigificatly differet, (Ma-Whitey P = 0.001; Kruskal- Wallis P = 0.002), which implies a ufair medical resource distributio amog the four groups. Though the secod group ((Jiayi (Jiayi City, Jiayi Couty), Miaoli Couty, Taoyua Couty, ad Yuli Couty) has a acceptable level of medical resources, they are ot sufficiet. The third group (Kaoshog (Kaoshog City, Kao- shog Couty), Chaghua Couty, Taidog Couty ad Pigdog Couty) has scat medical resources less tha the average level ad eeds to improve as quickly as possible. The fourth group has a serious isufficiecy of medical resources (Hwalia Couty, Taia (Taia City, Taia Couty), Natou Couty ad Hsichu (Hsichu City, Hsichu Couty)), ad also eeds to improve as quickly as possible. Thus, for-profit hospitals should regard this area of very poor medical resources as a priority area to establish ew medical facilities where they will ot face excessive competitio. I fact, it is difficult to cosider Hsichu medical district as a place with isufficiet medical resources, because of its high-tech idustry image. But the biomedical garde of Hsichu Couty has already bee classified as 1 of 12 costructio proects i the Lovig Taiwa pla i 2008, ad buildig is to commece i 2009, which should improve the medical resources of the Hsichu area i the future. Other areas with isufficiet medical resource iclude Natou Couty, Hwalia Couty, Taidog Couty, ad Pigdog Couty. I particular, Natou Couty, Hwalia Couty, Taidog Couty, ad Pigdog Couty have poor medical resources i remote areas of all four couties; the hospital ceters are i the city, which results i large differeces i accessibility to medical treatmet. The govermet always classifies Jiayi Couty, Peghu Couty, ad Yuli Couty as areas that eed early improvemet, but oly the medical resources i Peghu Couty have bee effectively improved, suggestig that the govermet has ot bee successful i improvig the medical resources i every medical area. Whe selectig a appropriate Taipei, 1 Yila, 1 Taichug, 1 Peghu, 1 Q3= Yuli, Taoyua, Miaoli, Jiayi, Q2= Q1= Pigdog, Taidog, Chaghua, Kaoshog, Taia, Hsichu, Natou, Hwalia, Figure 2. The relative efficiecy of the distributio of medical treatmet i every medical area of Taiwa

10 A Expert System Approach to Medical Regio Selectio for a New Hospital usig Data Evelopmet Aalysis 137 resources ad the local traffic situatio, which are disadvatages faced by distat couties such Natou, Pig locatio for a ew hospital, the sufficiecy of medical dog, Hwalia ad Taidog Couty, should be the mai cosideratios. Thus, the govermet should ecourage the establishmet of ew private hospitals i Hsichu (Hischu City, Hsichu Couty), Miaoli Couty, Chaghua Couty, Yuli Couty, Jiayi (Jiayi City, Jiayi Couty), Taia (Taia City, Taia Couty), ad Kaoshog (Kaoshog City, Kaoshog Couty). I the cases of Natou Couty, Pigdog Couty, Hwalia Couty, ad Taidog Couty, the govermet still eeds to offer sufficiet icetives, for example, by providig govermet scholarships to medical studets, promotig expasio proects for local public hospitals, ad providig subsidies. 6. Coclusios The value of this paper lies i evaluatig the equality of the distributio of medical resources usig the DEA model, while also settig up a assessmet model usig a expert system that measures may idicators while avoidig subective comparisos. This study is extremely helpful i expadig the theory of DEA modelig ito a practical applicatio related to medical ecoomics ad maagemet policies. The medical regios i Taiwa with the greatest eed to improve medical resources were listed idividually based o the results of the DEA method. This study also illustrates how the DEA method may be implemeted to assist decisio-makers i avoidig risk ad avoidig the establishmet of ew hospitals i medical districts with a excess supply of medical resources. These results may the provide a referece poit for for-profit ad oprofit hospitals whe selectig ew hospital locatios; ad, for the govermet, these results are a guidelie for developig future medical policies. I particular, this study assessed the equality of access idicator (percetage of people who ca obtai medical treatmet withi a acceptable time) usig a questioaire survey. The purpose of the survey was to idetify the time it took to obtai medical treatmet that people foud acceptable, which ca more clearly reflect people s perceptio of the accessibility of medical care, ad serve as a covicig idicator of accessibility. Fially, this study proposes the followig: 1) Both the DEA method ad the process metioed i this study should be applicable whe private medical istitutes are seekig locatios for ew hospitals. These fidigs ot oly help to evaluate the actual differeces before establishig ew hospitals, but also to reduce uecessary risks resultig from excessive competitio. 2) The first priority for improvemet are those areas with a serious isufficiecy of medical resources, icludig Hwalia Couty, Taia (Taia City, Taial Couty), Natou Couty, ad Hsichu (Hsichu City, Hsichu Couty). Next, those areas with scat medical resources, icludig Kaoshog (Kaoshog City, Kaoshog Couty), Chaghua Couty, Taidog Couty ad Pigdog Couty), eed improvemet. Fially, areas with isufficiet medical resources, icludig Jiayi (Jiayi City, Jiayi Couty), Miaoli Couty, Taoyua Couty ad Yuli Couty, eed to be addressed. 3) Based o the study coducted here, the govermet should ecourage private medical istitutes to establish ew hospitals i medical regios such as Taoyua Couty, Hsichu (Hsichu City, Hsichu Couty), Miaoli Couty, Chaghua Couty, Yuli Couty, Jiayi (Jiayi City, Jiayi Couty), Taia (Taia City, Taia Couty), or Kaoshog (Kaoshog City, Kaoshog Couty), where they ca beefit from relatively low competitio ad a reasoable traffic situatio. 4) For regios i which traffic adversely affects accessibility to medical treatmet, such as Natou Couty, Pigdog Couty, Hwalia Couty, ad Taidog Couty, the govermet should cotiue providig icetives, such as scholarships to studets, expadig the local public hospital, ad offerig suitable subsidies. 5) For medical regios with sufficiet medical resources, icludig Taipei (Taipei City, Taipei Couty ad Keelug City), Yila Couty, Taichug (Taichug City, Taichug Couty), ad Peghu Couty, the govermet should cosider delayig the establishmet of ew public hospitals ad trasferrig public hospitals to private orgaizatios, which would also reduce the govermet s fiacial burde. REFERENCES [1] Departmet of Health, Health ad Vital Statistics, Departmet of Health, Taiwa, [2] T. C. Liu ad P. C. Wu, The Choice of Medical Istitutios uder NHI-A Example of Pediatric Patiets i Taipei, Joural of Healthcare Maagemet, Vol. 2, No. 2, 2002, pp [3] D. L. Jiag, Health Policy: Taiwa Experiece, ChuLiu, Taipei, [4] W. C. Hsiao, C. L. Yag ad J. R. Lu, Health Care Fiacig ad Delivery i the ROC: Curret Coditios ad Future Challeges, Idustry of Free Chia, 1990, pp [5] Y. C. Lee, M. S. Lai ad P. C. Sheg, The Developmet ad the Implicatio of Aual Health Care Expediture Target for Global Budget Paymet System of Natioal Health Isurace i Taiwa, Joural of Healthcare Maagemet, Vol. 2, No. 2, 2001, pp [6] C. E. Phelps, Health Ecoomics, Addiso Wesley Educatioal Publishers Ic, New York, [7] J. P. Newhouse, A. P. Williams, B. W. Beett ad W. B. Schwartz, Does the Geographical Distributio of Physicias Reflect Market Failure, Bell Joural of Ecoomics, Vol. 13, No. 2, 1982, pp

11 138 A Expert System Approach to Medical Regio Selectio for a New Hospital Usig Data Evelopmet Aalysis [8] T. L. Chiag, Deviatio from the Carryig Capacity for Physicias ad Growth Rate of Physicia Supply: The Taiwa Case, Social Sciece ad Medicie, Vol. 40, No. 3, 1995, pp [9] Departmet of Health, Health ad Vital Statistics, Departmet of Health, Taiwa, [10] F. X. Xiao, Idustrial Ecoomics, Root Iteratioal Iformatio Co., Ltd., [11] R. E. Saterre ad S. P. Neu, Health Ecoomics: Theories, Isights, ad Idustry Studies, Harcourt Brace ad Compay, Fort Worth, [12] R. F. Lu ad Q. R. Xie, Health Ecoomics, Taipei, [13] L. M. Seiford, Data Evelopmet Aalysis: The Evolutio of the State of the Art ( ), Joural of Productivity Aalysis, Vol. 7, No. 2-3, 1996, pp [14] L. G. Xue, Expert System i Library, Idex of NCL Taiwa Brach Bulleti, Vol. 5, No. 1, 1991, p. a10. [15] B. K. Duval ad L. Mai, Expert Systems: What is a Expert System? Library Software Review, Vol. 13, No. 1, 1994, pp [16] J. Le Grad, Equity ad Choice, Harper Collis, Lodo, [17] J. Le Grad, Health ad Health Care, Social Justice Research, Vol. 1, 1987, pp [18] G. Mooey, Equity i Health Care: Cofrotig the Cofusio, Effective Health Care, Vol. 1, No.4, 1983, pp [19] G. Mooey, J. Hall, C. Doaldso ad K. Gerard, Utilizatio as a Measure of Equity: Weighig Heat? Joural of Health Ecoomics, Vol. 10, No.4, 1991, pp [20] G. Mooey, J. Hall, C. Doaldso ad K. Gerard, Reweighig Heat: Respose to Culyer, Va Doorslaer ad Wagstaff, Joural of Health Ecoomics, Vol. 11, No.2, 1992, pp [21] A. J. Culyer ad A. Wagstaff, Equity ad Iequity i Health ad Health Care, Joural of Health Ecoomics, Vol. 12, No. 4, 1993, pp [22] R. Pechasky ad J. W. Thomas, The Cocept of Access: Defiitio ad Relatioship to Cosumer Satisfactio, Medical Care, Vol. 19, No. 2, 1981, pp [23] J. P. Actio, Nomoetary Factors i the Demad for Medical Services: Some Empirical Evidece, Joural of Political Ecoomy, Vol. 83, No. 3, 1975, pp [24] R. M. Coffey, The Effect of Time Price o the Demad for Medical Care Services, Joural of Huma Resources, Vol. 18, No. 3, 1983, pp [25] S. D. Cauley, The Time Price of Medical Care, Review of Ecoomics ad Statistics, Vol. 69, No. 1, 1987, pp [26] C. D Aspremot, J. J. Gabszewicz ad J. F. Thisse, O Hotellig s Stability i Competitio, Ecoometrical, Vol. 47, No. 5, 1979, pp [27] Bureau of Natioal Health Isurace, The Natioal Health Isurace Statistics, Taiwa, [28] T. E. Getze, Health Care is a Idividual Necessity ad a Natioal Luxury: Applyig Multilevel Decisio Model to the Aalysis of Health Care Expeditures, Joural of Health Ecoomics, Vol. 19, No. 2, 2000, pp [29] L. F. Rossiter ad G. R. Wilesky, Idetificatio of Physicia-Iduced Demad, Joural of Huma Resources, Vol. 19, No. 2, 1984, pp [30] R. M. Aderso ad J. F. Newma, Societal ad Idividual Determiates of Medical Care Utilizatio i the Uited States, Milbak Mem Fud Quarterly, Vol. 51, No. 1, 1973, pp [31] C. Cox, Physicia Utilizatio by Three Groups of Ethic Elderly, Medical Care, Vol. 24, No. 8, 1986, pp [32] C. Doaldso ad K. Gerard, Ecoomics of Health Care Fiacig: The Visible Had, St. Marti s Press, New York, [33] B. A. Weisbrod, Toward a Theory of the Volutary No- Profit Sector i a Three-Sector Ecoomy. I E. Phelps, Ed., Altruism, Mortality ad Ecoomic Theory, Rusell Sage Foudatio, New York, [34] S. W. H. Cheg ad J. R. Su, The Icidece of Expeditures ad Reveues i Taiwa s Natioal Health Isurace, Taipei Iteratioal Coferece o Health Ecoomics, Taipei, [35] E. Va Doorslaer, A. Wagstaff, H. Va Der Burg, T. Christiase, D. D. Graeve, I. Duchese, U. G. Gerdtham, M. Gerfi, J. Geurts, L. Gross, U. Hakie, J. Joh, J. Klavus, R. E Leu, B. Nola, O. O Doell, C. Propper, F. Puffer, M. Schellhor, G. Sudberg ad O. Wikelhake, Equity i the Delivery of Health Care i Europe ad the US, Joural of Health Ecoomics, Vol. 19, No.5, 2000, pp [36] L. M. Schalick, W. C. Hadde, E. Pamuk, V. Navarro ad G. Pappas, The Wideig Gap i Death Rates amog Icome Groups i the Uited States From 1967 to 1986, Iteratioal Joural of Health Services, Vol. 30, No. 11, 2000, pp [37] A. Wagstaff, E. Doorslaer ad P. P. Va, Equity i the Fiace ad Delivery of Health Care: Some Tetative Cross -Coutry Compariso, Oxford Review of Ecoomic Policy, Vol. 5, No. 1, 1989, pp [38] A. Chares, W. W. Cooper ad E. Rhodes, Measurig the Efficiecy of Decisio Makig Uits, Europea Joural of Operatioal Research, Vol. 2, No. 6, 1978, pp [39] P. L. Chag, S. N. Hwag ad W. Y. Cheg, Usig Data Evelopmet Aalysis to Measure the Achievemet ad Chage of Regioal Developmet i Taiwa, Joural of Evirometal Maagemet, Vol. 43, No.1, 1995, pp [40] F. Pedraa-Chaparro, J. Salias-Jimeez ad P. Smith, O the Quality of the Data Evelopmet Aalysis Model, Joural of the Operatioal Research Society, Vol. 50, No. 6, 1999, pp [41] Highway Bureau, Statistical Yearbook of Highway Bureau, M. O. T. C., Miistry of Trasportatio ad Commuicatios, Taiwa, 2006