You are on page 1of 7

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/224329945

Fuzzy Obesity Index for Obesity Treatment and Surgical Indication

Conference Paper  in  IEEE International Conference on Fuzzy Systems · July 2008


DOI: 10.1109/FUZZY.2008.4630703 · Source: IEEE Xplore

CITATIONS READS
15 233

2 authors:

Susana Abe Miyahira Ernesto Araujo


Hospital Municipal Dr. José de Carvalho Florence, São José dos Campos, Brazil IATECH - Inteligência Artificial em Tecnologia - Consultoria, Projeto e Inovação Ltda
42 PUBLICATIONS   189 CITATIONS    69 PUBLICATIONS   321 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Curso de Metodologia da Pesquisa em Ciências da Vida View project

Análise Econômica do Acesso da População ao Sistema Único de Saúde (SUS) pela Redução Difusa de Dimensão View project

All content following this page was uploaded by Ernesto Araujo on 27 August 2014.

The user has requested enhancement of the downloaded file.


Fuzzy Obesity Index for Obesity Treatment and Surgical Indication
Susana Abe Miyahira, Ernesto Araujo, Member, IEEE

Abstract— A Fuzzy Obesity Index for being used as an analysis. The search for a more accurate method to evaluate
alternative in obesity treatment and bariatric surgery indication obesity and, consequently, the better treatment, motivate the
(BSI) is presented in this paper. Obesity is nowadays understood proposition of a new index for surgical treatment and for a
as universal epidemy and became an important source of death
and co-morbidities. The search for a more accurate method to novel classification for obesity and the classes correlated to
evaluate obesity and to indicate a better treatment is important it. A novel Fuzzy Obesity Index is proposed for being used
in the world health context. In this paper the Body Mass as an alternative in obesity treatment and BSI, as well.
Index (BMI) is first modified and treated as fuzzy sets. BMI is The fuzzy set theory and fuzzy logic related to the
characterized by its capacity of weight excess and considered the approximate reasoning is the theory of classes with fuzzy
main criteria for obesity treatment and BSI. Nevertheless, the
fat excess related to the Body Fat (BF) is actually the principal borders and they present great appeal in the real world,
harmful factor in obesity disease, that is usually neglected. Due being of special interest to biosciences. Update publications
to that this paper also presents a new fuzzy mechanism for demonstrate the progressive growing in the use of fuzzy
evaluating obesity by associating BMI with Body Fat (BF) that logic in diverse medical areas such as internal medicine,
yields a fuzzy obesity index for obesity evaluation and treatment cardiology, vascular surgery, orthopedy, anesthesiology, der-
and allows to build up a Fuzzy Decision Support System (FDSS)
for BSI. Different values of BMI and BF (in terms of %BF) matology, ophthalmology, otorrinolaringology, gynecology,
used for validating the proposed method classify individuals in urology, neurology, pathology, biochemistry, pharmacology
distinct categories with degrees of compatibility more realistic and genetics [5]. These characteristics associated to the
than those accomplished by Boolean classification, as usually capacity to deal with linguistic variables and linguistic terms
occur. The proposed method may assume an important whole making easy the understanding of operators and users, the
in medicine as an index for obesity evaluation and surgery
treatment by using the advantages of BMI and BF in synergy. low computational cost and the ability to incorporate the
expertise of specialists justify the growing number of tech-
nical papers applied to the fuzzy sets theory and fuzzy
logic in biomedical questions. Due to that this mathematical
approach has become an interesting option to elaborate
I. I NTRODUCTION medical models, be it for diagnose systems, medical image
treatment, public health or epidemiology, only to mention
T HE clinical condition known as obesity currently as-
sumes the characteristic of universal epidemy. This
syndrome is important cause of death and related diseases
few [6], [7], [8], [9], [10], [11], [12], [13], [14].
The first main contribution of this proposal is to modify
(also denominate co-morbidities). Moreover, it directly and the Body Mass Index (BMI) by treating the crisp classes
proportionally corresponds to the magnitude of the obesity usually adopted by the World Health Organization (WHO) to
clinical condition [1], [2], [3], [4]. Nevertheless, the obesity fuzzy sets. Although the BMI is the most known mechanism
definition, measurement, classification and treatment are not employed for classify obesity, there is a consensus concerned
easily determined. Finding out a mechanism for obesity eval- the gravity that pathological condition of the chronic obesity
uation and level classification, mainly if concerned with risks disease assumes is directly proportional to fat accumulation
or prescription of treatments, bariatric surgery indication in the body – Body Fat (%BF). This fuzzification proceeding
(BSI) etc., is of great clinical interests. is extended to %BF classes. In so doing, diverse values for
Effort have been spent to classify what is overweight, BMI and %BF classify individuals in different categories
obesity etc. by using diverse anthropometric parameters when with more realistic degrees of compatibility when compared
the interest is to determine the best therapeutic indication and with those indexes for obesity evaluation and surgical treat-
treatment as well as the best appropriate mechanism for data ment with Boolean classification as usually utilized. After-
wards, the BMI and %BF classes are aggregated yielding a
Susana Miyahira is both with the Anesthesiology Dpt., Hospital Munic- new index. This new index uses the advantage of the old
ipal Dr. Jose de Carvalho Florence, Av. Saigiro Nakamura, 800, 04023- methods and aims to compensate the disadvantage of them.
062, Sao Jose dos Campos, SP, Brazil and with Universidade Federal de
Sao Paulo (UNIFESP), R. Botucatu, 862, 04023-062, Sao Paulo, SP, Brazil If as for input linguistic variable are considered the BMI and
(email: susana miyahira@uol.com.br). %BF in which the universe of discourse is parted by using
Ernesto Araujo is with Computer Science / Health Informatics Depart- fuzzy sets, as output linguistic variable is employed a obesity
ment (DIS), Universidade Federal de Sao Paulo (UNIFESP), R. Botucatu,
862, 04023-062, Sao Paulo, SP, Brazil (email: ernesto.araujo@unifesp.br), classification with entirely new classes of obesity in the fuzzy
with Integration and Testing Laboratory (LIT), Instituto Nacional de context as well as is used for BSI. The final contribution of
Pesquisas Espaciais (INPE), Av. Astronautas, 1758, 12.227-010, (email: this paper is to use the new fuzzy mechanism for getting
ernesto.araujo@lit.inpe.br), and with the Hospital Municipal Dr. Jose de
Carvalho Florence, Av. Saigiro Nakamura, 800, 04023-062, both in Sao together BMI and Body Fat (BF) for building up a Fuzzy
Jose dos Campos, SP, Brazil. Decision Support System (FDSS) for surgery indication.

2392

978-1-4244-1819-0/08/$25.002008
c IEEE
TABLE II
II. M ECHANISMS FOR B ODY M ASS E VALUATION
BARIATRIC SURGERY INDICATION (BSI) ACCORDING TO B ODY M ASS
The Body Mass Index (BMI), also denominate Quetelet I NDEX (BMI) AND COMORBIDITIES .
index [15], is a grade system for classify obesity by us-
ing a mathematical relation of proportionality given by 35 < BM I < 39 BM I > 40
IMC = P/H 2 , where the body weight, P , of the individual Without Comorbidities No Surgery Surgery
With Comorbidities Surgery Surgery
is given in Kilograms [Kg] and the square of the height, H,
express in square meters, [m2 ]. The distinct classes of BMI
for identification, evaluation and treatment of overweight
(ow) and obesity in adults is presented in Table I. accumulation and not the fat free mass that does not affect the
The BMI is an international index used in epidemiological health of the patient [16], [17], [18], [19], [20]. Additionally,
studies and it is the most known mechanism for classifying the BMI reveals itself a imprecise method and low accuracy
obesity and determining obesity treatment or BSI. According for measure BF in distinct physical classifications. Despite
to the BMI and comorbidities, the BSI classification is given these limitations, the BMI is employed to indicate obesity
in Table II. This index, however, presents severe limitations. treatment all over the world and surgical treatment [21], [22].
The main disadvantage of BMI is that the actual body There are distinct methods for BF evaluation such as, for
composition is not taken into account. The estimate excess instance, SkinFold Thickness (SKF), Densitometry (Hydro-
of weight may occur both by increasing the fat tissue and static Weight), Dual-Energy X-ray Absorptiometry (DEXA),
by the muscular hypertrophy. In this sense, a person with Bioelectrical Impedance Analysis (BIA). Each of them
BMI with a index of 30 may be classified in the obese presents inherent advantages and disadvantages. However,
group and be erroneously classified as overweight. In the all of them demonstrate good correlation among themselves,
counterclock direction, a physiculturist with an usual low thus, justifying their use interchangeably.
percentage of body fat – for instance, 9.3 – may also be The %BF is considered a better index for evaluate obesity
erroneously classified as overweight if only if it is assumed than BMI since the latter only works as an indicative of
the BMI as the unique way of obesity measurement. Another weight excess and the first one is an indicative of excess of
disadvantage is also the use of height as a main parameter fat mass and the risk of keep being obese (Table III). Thus,
of analysis due to its incorrectness and modification in the it seems that the %BF has being underexploited and so it
body structure. Consider, for instance, when the patient should be investigated to evaluate the surgical treatment.
presents anatomical alteration in backbone by having its There is, however, no consistent information in the litera-
height decreased, as it happens to old people. In this case, ture presenting which is the distributed %BF in the organism
the BMI gives an inappropriate value, more elevated than the that corresponds to the BMI, nor if it is compatible with
actual due to the bend of the backbone that usually occur with the prescription of surgical treatment. Neither there is a
old people or presenting a health backbone problem. This degree of %BF that would correspond to the development
problem is also extended to people that present low height of comorbidies that are associate to the obesity disease. In
due to some hormonal problem. Moreover, momentaneous or order to introduce a consistent mechanism for dealing with
periodic weight modification such as those of weight raise obesity in this paper both BMI and %BF are put together
in menstrual period may result in an artificial elevated BMI, to present a new obesity index by employing fuzzy sets and
influencing the final classification. Finally, the BMI does not fuzzy logic theory.
point out the body fat and becomes, thus, an restrict method
III. F UZZY O BESITY I NDEX
for evaluating obesity and for BSI.
If for one side, the determination of BMI is a mechanism Nowadays, the obesity classification and indication of
to help in determining the excess of weight world wide surgical treatment, for example, with BMI is accomplished
known, for other side, the BF measurement presents a greater by using classic (Aristotelic) sets theory, in which the values
importance than the simpler measurement of the global belongs to a set, or not (Fig. 1). In these sets an element, xi ,
body weight. The main harmful factor in obesity is the BF is a member of the universe of discourse, X, associated to
the linguistic variable BMI such that X = {x1 , . . . , xn |xi ∈
BMI} and assumes a unitary value, μ(xi ) = {1}, for each
TABLE I
element that belongs to the set or a null value, μ(xi ) = {0},
C LINICAL GUIDELINES ON THE IDENTIFICATION , EVALUATION AND
TREATMENT OF OVERWEIGHT AND OBESITY IN ADULTS .
TABLE III
Classification of Overweight Obesity Class BMI (Kg/m2 ) C LASSIFICATION OF OVERWEIGHT AND O BESITY BY %BF
Underweight (UW) – < 18, 4
Thin – 18,5 to 24,9 Obesity Women Men
Overweight (OW) – 25 to 29,9 Adequate (AF) < 25 % < 15 %
Obesity (OI) Grade I 30,0 to 34,9 Low (LF) 25–30 % 15–20 %
Obesity (OII) Grade II 35,0 to 39,9 Moderate (MOD) 30–35 % 20–25 %
Morbid Obesity (OIII) Grade III ≥ 40 Elevate (ELEV) 35–40 % 25–30 %
(Source: WHO - World Health Organization) Morbid (MOR) > 40 % > 30 %

2008 IEEE International Conference on Fuzzy Systems (FUZZ 2008) 2393


Fig. 1. Body Mass Index (BMI) as Crisp Classes.
(a) Fuzzy Membership Functions for Body Mass Index (BMI).

for each element that does not belong to the set. The same
proceeding is extended to %BF. Nevertheless, it seems a
mistake to attribute a Boolean classification as the one for
BMI and %BF. Consider, for instance, a patient with BMI =
39 and other patient with BMI = 40. According to the WHO,
these patients are classified in distinct groups, respectively,
Obesity II and Obesity III and presenting different indications
of treatment. While the first patient is out of the range of
surgical treatment, the second one is classified for treatment
even if the variation of the patients is minimal, i.e., from
BMI = 39 to BMI = 40 the variation is only ΔBMI = 1.
Usually, in these cases, both patients do not present signifi-
cant biological, anatomical, or physiopathological differences
that justify discrepancies in the indication of BSI with a (b) Fuzzy Membership Functions for Body Fat (%BF).
minimal variation, ΔBMI, as mentioned.
Fig. 2. Input Linguistic Variable for Fuzzy Obesity Index
The implementation of the proposed fuzzy obesity index
employs fuzzy grade of the values found in the BMI and
%BF and uses the logic operation of conjunction between
their partial values and linguistic terms. The obesity is, then,
given by the set of ordered pair, (x, y), such that the first
element is the member of the universe of discourse, X,
associated to the linguistic variable BMI and the second
one is member of the universe of discourse, Y , related to
linguistic variable %BF, yielding a Cartesian product in the
form , X × Y = {(x, y)|x ∈ X, y ∈ Y }, such that X =
{x1 , . . . , xn |xi ∈ BMI} and Y = {y1 , . . . , yn |yj ∈ %BF}.
Since in classical set theory the parting of the universe of
discourse is such that the ordered pair assumes either a Fig. 3. Output Linguistic Variable for Fuzzy Obesity Index
unitary value, (x, y) = 1, for each pair that belongs to the
relation or a null value, (x, y) = 0, for each par that does not
belong to the relation, i.e., μ(x,y) = {0, 1}, it seems to be classified as OII and OIII in the same time. The difference
more appropriate to part the universe of discourse of BMI and now exists in the fact that the first patient is classified in the
%BF by using the fuzzy set theory. In doing so, each ordered OIII group with a degree of activation higher than that one
pair, (x, y) assumes an intermediary value between 0 and 1, in the OII group – the contrary is true for the second patient.
i.e., μ(x,y) = [0, 1], being able to yield an overlapping of The use of fuzzy sets theory and fuzzy sets allows a patient
classes. The patient can now be classified in complementary with BMI = 39kg/m2 be classified according to the use of
and distinct manners as depicted in Fig. 2(a) and Fig. 2(b). fuzzy sets in the group of surgical treatment with a degree
The advantage of using fuzzy sets for classes of BMI and of pertinence concurrently being in the group of no surgical
%BF is that, following the same example used before, the indication but with another degree of pertinence. Moreover,
patients with BMI = 39 and BMI = 40 are both to be both patients would be able to be indicate to BSI, or not, for

2394 2008 IEEE International Conference on Fuzzy Systems (FUZZ 2008)


example, depending on additional factors and not exclusively
related to BMI as currently accomplished. Due to that, the
fuzzy approach allows to each patient to be classified in a
compatible manner with the various available categories but
with different degrees of pertinence with the advantage of
considering the more realistic classification for BSI related
to the variables of analysis employed.
The fuzzy logic has its core based on inference rules.
This fuzzy inference mechanism uses the logical principles to
determine how facts and rules must be combined to derive
other facts. An important concept is the fuzzy conditional
proposition. When dealing with two inputs and one output –
as it is the case in this paper – it assumes the form, IF: < x
is A > AND < y is B > THEN < z is C >, where x and
y are input linguistic variables, A and B are input linguistic
terms, z is the output linguistic variable, and C is the output
linguistic term. The terms (x is A) and (y is B) form the Fig. 4. Porcentage of Body Fat (%BF) as proposed Fuzzy Membership
Functions.
antecedent while (z is C), the consequent of the rule. In
this paper, the input linguistic variables are the BMI and
%BF. The output linguistic variable is the obesity evaluation
labeled as classes or linguistic terms, for example, as
and its correlated to BSI. This relationship is associate to
%BF = {AF, LF, MOD, ELEV, MOR} (Fig. 2(b)).
the estimative of obesity (input) and obesity classification or
The output in the consequent of the rule is given by obesity
BSI (output).
evaluation also related to BSI (consequent of the rule). The
set of linguistic terms are chronic deficit of energy (CDE),
A. Preliminary Parameters of the Fuzzy Obesity Index
Thin (TH), muscular hypertrophy (MUH), excess of weight
A fuzzy set, A, in the universe of discourse BMI is (EW), sutomori (SUT), fuzzy obesity (FZOB), morbid obe-
defined as a membership function, μA (x), mapping each sity (MOB). The output linguistic variable for obesity index
element to a number (degree) in the interval [0, 1]. This or BSI is given in Fig. 3.The elements, zi , of %BF, are
membership function, μA (x), can be understood as de degree distributed in the universe of discourse, Z, and grouped
of compatibility for the linguistic terms. Thus, the proposed and labeled as classes or linguistic terms, for example,
fuzzy obesity index establish an arbitrary value in the interval as Obesity = {CDE, TH, MUH, EW, SUT, FZOB, MOB}
of [0, 1] yielding a graduated and smooth surface in the (Fig. 3). The sutomori fuzzy set for obesity is introduced
classification of BMI. The BMI input linguistic variable was by the authors an there is no similar in literature. The sumo
built up based on World Health Organization (WHO) as wrestlers are classified apart of the other categories since they
shown in Table I. The linguistic terms for parting the universe present unique characteristics. These athletes have a muscular
of discourse are lower weight (LW), thin (T), overweight mass and presents a high level of %BF and due to that are
(OW), Obesity grade I (OI), Obesity grade II (OII), and usually considered as obese. However, when compared with
Obesity grade III (OIII). In doing so, the elements, xi , of the individuals with equivalent BMI, they present lower values
universe of discourse BMI are distributed in the universe of of %BF.
discourse, X, and grouped and labeled as classes or linguistic In order to find out a mechanism of classification that takes
terms as BMI = {LW, T, OW, OI, OII, OIII} (Fig. 2(a)). into account the advantage of distinct method already em-
When building the %BF input linguistic variable, new ployed but working in synergy to suppress their disadvantage,
overweight and obesity values were adapted from Ta- here, the BMI and %BF are combined by using the fuzzy
ble III and the selected linguistic terms are adequate sets and fuzzy logic and form the premises related to the
fat (AF), low fat(LF), moderate (MOD), elevate (ELEV) antecedent of the rule. The rules for the base of knowledge
and morbid (MOR). The elements, yi , of %BF, are dis- is given according to the fuzzy matrix presented in Table IV.
tributed in the universe of discourse, Y , and grouped and The rules herewith are restrict to those relevant and actual
to life existence as well as practical occurrence, so, some
TABLE IV possible rules are not taken into account. In this paper, obese
F UZZY M ATRIX FOR O BESITY C LASSIFICATION for BSI are considered those into FZOB class in conjunction
with comorbidities and those classified as MOB, independent
UW T OW OI OII OIII of the presence of comorbidities. The mapping surface for
AF CDE TH MUH MUH MUH X
the fuzzy model related to the new obesity index and surgical
LF X TH MUH MUH MUH X
MOD X EW EW SUT SUT MOB treatment is depicted in Fig. 4.
ELEV X EW FZOB FZOB FZOB MOB The inference mechanism for decision-making chosen is
MOR X X FZOB FZOB MOB MOB the Mamdani approach. The defuzzification method is the

2008 IEEE International Conference on Fuzzy Systems (FUZZ 2008) 2395


TABLE V
C OMPARATIVE E XAMPLES OF B ODY M ASS I NDEX (BMI), B ODY FAT (%BF) AND M IYAHIRA -A RAUJO F UZZY O BESITY I NDEX (MAFOI)

Classification (Men) BMI Crisp Fuzz %BF Crisp Fuzz MAFOI∗


1 Control [35] 20.48 T UW, T 12.9 LF AF,LF 14.2 TH
2 Body building [36] 26.32 OW T, EW 9.3 AF AF 17.6 MUH
3 Sarcopenic Obesity [37] 24.00 T T, EW 19.9 LF LF,MOD 20.2 EW
4 USA Men [38] 34.30 OI OI, OII 34.7 MOR MOR 28 FZOB
5 GrH Treated [39] 39.20 OII OII, OIII 28 ELEV ELEV, MOR 29 FZOB
6 Sumo Wrestlers [40] 36.50 OII OI, OII 26.5 ELEV ELEV, MOR 28.1 SUT
7 Sumo Wrestlers [41] 36.84 OII OI, OII 29.6 MOR ELEV, MOR 27.9 SUT
8 Control Sumo Wrestlers [41] 36.50 OII OI, OII 35.4 MOR MOR 31.8 MOB

center of area. It is worthing mentioning that prevents the use This paper, initially, uses data of men population, further
of other inference mechanisms and defuzzification methods. paper will consider women and children, as well. For in-
stance, in another example 8 in table V, the MAFOI indicates
IV. A NALYSIS AND I LLUSTRATIVE E XAMPLES
MOB. According to the set of rules and the membership the
At the upper extreme of the bidimensional graphic in Fig. 4 output value is 31.8 and absolutely indicates a obesity for
there is no difference when using the proposed method or bariatric surgical indication. In the examples 4 and 5, the
the classical approaches. Nevertheless, in the intermediary MAFOI is FZOB with obesity value of 29 meaning that
regions the fuzzy obesity index shows its value since it allows there is BSI only in the presence of some commorbities.
to represent degrees of composed obesity when using two Other examples can directly be observed in table V.
important indexes in medicine – BMI and %BF – that was Results indicate a gradual, smooth obesity classification
not possible before without employing fuzzy set theory. and BSI when using the proposed fuzzy obesity index when
Although not being the focus of this paper, another region compared with other traditional methods for dealing with
that is worth to point out is the inferior part of the Cartesian obesity.
product (fuzzy obesity index) labeled CDE. It indicates
a mathematical and formal mechanism for represent and
understand anorexy. V. C ONCLUSIONS
It is also important to mention that when the %BF is in the
The use of fuzzy sets and fuzzy logic allows the conception
lower limit this obesity indicative is more important than the
of a new model for obesity classification were individuals
BMI and the latter would be disconsidered, however, being
were classified in distinct categories according to conjunct
important in the other regions. The HMU class means that
fuzzy values of Body Mass Index (BMI) and Porcentage of
there is an increasing weight due to muscular mass and not
Body Fat (%BF) in a gradual, smooth and complementary
to fat excess.
manner. In this way, this paper preconize a novel index that
There is overlapping classification of EW, SUT, and
relate the obesity indicators previously known in literature,
FZOB, however, the sumo wrestlers (SUT) are put apart
in particular, the BMI and %BF by using fuzzy sets and
since they present special characteristics. These athletes have
fuzzy logic, making them work in synergy instead of in a
weight increasing both in muscular mass as also of body fat
conflicting, competitive manner.
and may be classified as obese in the EW category. However,
The original approach herewith is more realistic and ap-
when compared with individuals with BMI equivalents, they
propriate to the practical necessities to obesity classification
have lower %BF, present physiological adaptation, and do
as well as indication of bariatric surgery.
not present commorbities. Such a fact can be noticed in the
examples 4 and 7 in the illustrative examples of table V This study conclude that the BMI is not adequate for
for what the values obtained with the proposed MAFOI are surgical indication in all the conditions and that the fuzzy set
coincident for both groups of analysis. theory and fuzzy logic become an alternative for decision-
In order to validate the fuzzy obesity model proposed, this making in bariatric surgery indication based on the Miyahira–
paper selected different obesity evaluations in the Medline Araujo Fuzzy Obesity Index (MAFOI). The novel index can
and Medscape Data Base for analyzing both BMI and %BF. assume an important whole in medicine by changing the way
Since the proposed fuzzy obesity index can interchangeably obesity is perceived and treated.
be employed with all of the other BF measurements. The
%BF values obtained with SkinFold Thickness (SKF), Den- R EFERENCES
sitometry (Hydrostatic Weight), Dual-Energy X-ray Absorp-
tiometry (DEXA), as well. These data are compared and [1] G. Kolata, “Obesity declared a disease”, Science, vol. 227, pp. 1019–
1020, 1985.
analysed as shown in Table V according to literature [35], [2] W. P. T James and A. Ralph, “New understanding in obesity research”,
[36], [37], [38], [39], [40], [41]. Proc. Nutr Soc., vol. 58, 1999, pp. 385–393.

2396 2008 IEEE International Conference on Fuzzy Systems (FUZZ 2008)


[3] T. L. Visscher and J. C. Seidell and A. Menotti and H. Blackburn and [22] F. Curtin and A. Morabia and C. Pichard and D. O. Slosman, “Body
A. Nissinem and E. J. Feskens and D. Kromhout, “Underweight and mass index compared to dual-energy x-ray absorptiometry: evidence for
overweight in relation to mortality among men aged 40-59 and 50-69y: a spectrum bias”, J Cin Epidemiol., vol. 50, pp. 837–843, 1997.
the Seven Countries Study.”, Am J Epidemiol., vol. 151, pp. 660–666, [23] H. C. Lukaski, “ Validation of tetrapolar biolelectrica impedance
2000. method to assess human body composition.”, J Applied Physiol., vol. 60,
[4] J. T. Fine and G. A. Colditz and E. H. Coakley and G. Moseley and pp. 1127–1132, 1986.
J. E. Manson and W. C. Willett and I. Kawachi, “A prospective study [24] G. Sum and C. French and C. Martin, “ Comparison o multifrequency
of weight change and health-relates quality of life in women.”, JAMA., bioelectrical impedance analysis with absorptometry for assessment of
vol. 282, pp. 2136–2142, 1999. percentage body fat in a large healthy population.”, Am J Clin Nutr.,
[5] M. F. Abbod and K. von Keyserling and S. Linken and M. Mahfouf, vol. 81, pp. 74–78, 2005.
“Survey of utilization of fuzzy technology in Medicine and Healthcare. [25] K. R. Segal and M. VanLoan and P. I. Fitzgerald and J. A. Hodgdon
”, Fuzzy Sets and Systems., vol. 120 pp.331-349, 2001 and T. B. VanItallie, “ Lean body mass estimation by bioelectrical
[6] Y. Miki and R. I. Grossman and J.K. Udupa and S. Samarasekera and impedance analysis: a four-site cross-validation study.”, Am J Clin Nutr.,
M. A. van Buchem and B. S. Cooney and S. N. Pollack and D. L. vol. 47, pp. 7–14, 1988.
Kolson and C. Constantinescu and M. Polansky and L. J. Mannon, [26] E. C. Rush and J. Crowley and I. F. Freita and A. Luke, “ Valid-
“Computer-assisted quantitation of enhancing lesions in multiple scle- ity of hand-to-foot measurement of bioelectrical impedance: standing
rosis: correlation with clinical classification. ”, Am J Neuroradiology., compared with lying positon.”, Obesity., vol. 14, pp. 252–257, 2006.
vol. 18, pp.705-710, 1997. [27] S. Vasudev and A. Mohan and D. Mohan and S. Farooq and D. Raj and
[7] L. F. C. Nascimento and N. R. Ortega , “Fuzzy linguistic model for V. Mohan, “ Validation of body fat measureament by skinfolds and two
evaluating the risk of neonatal death.”, Rev Saúde Pública., vol. 36, bioelectrical impedance methods with DEXA -the Chennai Urban Rural
pp.686-692, 2002. Epidemiology Study [CURES].”, J Assoc Physicians India., vol. 52, pp.
[8] E. W. Jensen and H. Litvan and M. Revuelta and B. E. Rodriguez and 877–881, 2004.
P. Carminal and P. Martinez and H. Vereecke and M. R. F. Struys, [28] R. L. Newton and A. Alfonso and M. A. White and E. York-Crowe
“Cerebral state index during propofol anesthesia. ”, Anesthesiology., and H. Walden and D. Ryan and G. A. Brav and D. A. Williamson,
vol. 105, pp.28-36, 2006. “ Percent body fat measured by BIA and DEXA in obese, African-
[9] E. P. Martinioni and C. A. Pfister and K. S. Stadler and P. M. American adolescent girls.”, Int J Obes., vol. 29, pp. 549–602, 2005.
Schumacher and D. Leibundgut and T. Bouillon and T. Bohlen and A. [29] J. M. Jakicic And R. R. Wing and W. Lang, “Bioelectrical impedance
M. Zbinden, “Model-based control of mechanical ventilation: design analysis to assess body composition in obese adult women: the effect of
and clinical validation. ”, Br J Anaesth., vol. 92, pp.800-807, 2004. ethnicity.”, Int J Obes Relat Metab Disord., vol. 22, pp.243-249, 1998.
[10] F. J. Lloyd and V. F. Reynad, “A web exercise in evidence-based [30] R. Roubenoff, “Applications of bioelectrical impedance analysis for
medicine using cognitive theory. ”, J Gen Intern Med., vol. 16, pp.94- body composition to epidemiologic studies.”, Am J Clin Nutr., vol. 644,
99, 2001. pp.459S-462S, 1996.
[11] R. S. Litman and N. Wake and L. M. L. Chan and J. M. McDonough [31] R. L. Newton and A. Alfonso and E. York-Crowe and H. Walden and
and S. Sin and S. Mahboubi and R. Arens, “Effect of lateral positioning M. A. White and D. Ryan and D. A. Williamson, “Comparison of body
on upper airway size and morphology in sedated children. ”, Anesthe- composition methods in obese African-American women.”, Obesity.,
siology., vol. 103 pp.484-489, 2005. vol. 14, pp.415-422, 2006.
[12] V. H. Heyward and L. M. Stolarczyk and I. L. Champaign, “Applied [32] J. C. Wells and J. E. Williams and M. Fewtrell and A. Singhal and
body composition assessment. ”, Human Kinetics., vol. 12 pp.88, 1996 A. Lucas and T. J. Cole, “A simplified approach to analysing bio-
[13] I. Virant-Klun and J. Virant, “Fuzzy logic alternative for analysis in electrical impedance data in epidemiological surveys.”, Int J Obes.,
the biomedical sciences. ”, Computers Biomedical Research., vol. 32 vol. 31, pp.507-514, 2007.
pp.305-321, 2006. [33] K. Willet and R. Jiang and E. Lenart and D. Spiegelman and W. Willet,
[14] A. Torres and J. J. Nieto, “Fuzzy logic in medicine and bioinformatics. “Comparison of bioelectrical impedance and BMI in predicting obesity-
”, J Biomed Biotech., vol. 2006 pp.1-7, 2006 related medical conditions.”, Obesity., vol. 14, pp.480-490, 2006.
[15] E. E. Calle and M. J. Thun and J. M. Petrelli, “Body-mass index [34] B. Guida and R. Trio and P. Pecoraro and M. C. Gerardi and R. Laccetti
and mortality in a prospective cohort of U.S. adults.”, N Engl J Med., and A. Nastasi and C. Falconi, “Impedance vector distribution by body
vol. 341, pp. 1097–1105, 1999. mass index and convertional bioelectrical impedance analysis in obese
[16] T. Chin-Hisao, “Body composition as a risk factor for coronary artery womem ”, Nutr Metab Cardiovasc Dis., vol. 13, pp.79-89, 2003.
disease in Chinese type2 diabetic patients in Taiwan”, Circ J., vol. 67, [35] H. Kanehisa and M. Kondo and S. Ikegawa and Fukuaga , “Char-
pp. 479–484, 2003. acteristcs of body composition and muscle strength in college sumo
[17] R. B. Singh and M. A. Niaz and R. Beegom and G. S. Wander and wrestlers. ”, Int J Sports Med., vol. 18 pp.510-515, 1997
A. S. Thakur, “Body fat percent by bioelectrical impedance analisys [36] V. L. Katch, “Muscular development and lean body weight in body-
and risk of coronary artery disease among urban men with low rates buiders and weight-lifters. ”, Med Sci Sports., vol. 12 pp.340, 1998
of obesity: the Indian paradox”, J Am Coll Nutr., vol. 18, pp. 268–273, [37] Y. Shultz and L. V. G. Kyle and C. Pichard, “Fat-free mass index and
1999. fat mass index percentiles in Caucasian aged 18-48y. ”, Int J Obesity.,
[18] T. Nagaya and H. Yoshida and H. Takahashi and Y. Matsuda and M. vol. 26 pp.953-960, 2002
Kawai, “Body mass index (weight/height2) or percentage body fat by [38] I. R. Pateyjohns and D. Brinkworth and J. D. Bucley and M. Noakes
bioelectrical impedance analisys: which variable better reflects serum and P. M. Clifton, “Comparison of three bioelectrical impedance
lipid profile?”, Int J Obes Relat Metab Disord., vol. 23, pp. 771–774, methods with DXA in overweight and obese men. ”, Obesity., vol. 14
1999. pp.2067-2070, 2006
[19] C. R. McLachlan and R. Poulton and G. Car and J. Cowan and [39] A. Halpern and M. C. Mancini and C. Cercato and S. M. Villares and
S. Filsell and J. M. Greene and D. R. Taylor and D. Welch and A. A. P. A. C. Costa, “[Effects of growth hormone on anthropometric
Williamson and M. R. Sears and R. J. Hancox, “Adiposity, asthma, and parameters in android obesity] Portuguese. ”, Arq Bras Endocrinol
airway inflammation”, J Allergy Clin Immunol., vol. 119, pp. 643–649, Metabol., vol. 50 pp.68-72, 2006
2007. [40] K. Hattori and M. Kondo and T. Abe and S. Tanaka and T. Fukunaga,
[20] T. D. Adams and E. M. Heath and M. J. LaMonte and R. E. Gress “Hierarchical differences in body composition of professional sumo
and R. Pendleton and M. Strong and S. C. Smith and S. C. Hunt, “The wrestlers. ”, Ann Human Biol., vol. 26 pp.179-184, 1999
relationship between body mass index and per cent body fat in the [41] S. Kinoshita and S. Onishi and S. Yamamoto and K. Yamada and Y.
severily obese”, Diabetes Obes Metab., vol. 9, pp. 498–505, 2007. Ogume and F. Katsukawa and H. Yamazaki, “Unusual left ventricular
[21] R. Ricciardi and L. A. Talbot, “Use of bioelectrical impedance analisys dilatation without functional or biochemical impairment in normoten-
in the evaluation, treatment, and prevention of overweight and obesity”, sive extremely overweight Japanese professional sumo wrestlers. ”, Am
J Am Acad Nurse Pract., vol. 19, pp. 235–241, 2007. J Cardiol., vol. 91 pp.699-702, 2003

2008 IEEE International Conference on Fuzzy Systems (FUZZ 2008) 2397

View publication stats

You might also like