Loading [MathJax]/jax/output/SVG/jax.js
Research article

Optimal feature selection using novel flamingo search algorithm for classification of COVID-19 patients from clinical text


  • Though several AI-based models have been established for COVID-19 diagnosis, the machine-based diagnostic gap is still ongoing, making further efforts to combat this epidemic imperative. So, we tried to create a new feature selection (FS) method because of the persistent need for a reliable system to choose features and to develop a model to predict the COVID-19 virus from clinical texts. This study employs a newly developed methodology inspired by the flamingo's behavior to find a near-ideal feature subset for accurate diagnosis of COVID-19 patients. The best features are selected using a two-stage. In the first stage, we implemented a term weighting technique, which that is RTF-C-IEF, to quantify the significance of the features extracted. The second stage involves using a newly developed feature selection approach called the improved binary flamingo search algorithm (IBFSA), which chooses the most important and relevant features for COVID-19 patients. The proposed multi-strategy improvement process is at the heart of this study to improve the search algorithm. The primary objective is to broaden the algorithm's capabilities by increasing diversity and support exploring the algorithm search space. Additionally, a binary mechanism was used to improve the performance of traditional FSA to make it appropriate for binary FS issues. Two datasets, totaling 3053 and 1446 cases, were used to evaluate the suggested model based on the Support Vector Machine (SVM) and other classifiers. The results showed that IBFSA has the best performance compared to numerous previous swarm algorithms. It was noted, that the number of feature subsets that were chosen was also drastically reduced by 88% and obtained the best global optimal features.

    Citation: Amir Yasseen Mahdi, Siti Sophiayati Yuhaniz. Optimal feature selection using novel flamingo search algorithm for classification of COVID-19 patients from clinical text[J]. Mathematical Biosciences and Engineering, 2023, 20(3): 5268-5297. doi: 10.3934/mbe.2023244

    Related Papers:

    [1] Kenneth H. Karlsen, Süleyman Ulusoy . On a hyperbolic Keller-Segel system with degenerate nonlinear fractional diffusion. Networks and Heterogeneous Media, 2016, 11(1): 181-201. doi: 10.3934/nhm.2016.11.181
    [2] José Antonio Carrillo, Yingping Peng, Aneta Wróblewska-Kamińska . Relative entropy method for the relaxation limit of hydrodynamic models. Networks and Heterogeneous Media, 2020, 15(3): 369-387. doi: 10.3934/nhm.2020023
    [3] Piotr Biler, Grzegorz Karch, Jacek Zienkiewicz . Morrey spaces norms and criteria for blowup in chemotaxis models. Networks and Heterogeneous Media, 2016, 11(2): 239-250. doi: 10.3934/nhm.2016.11.239
    [4] Panpan Xu, Yongbin Ge, Lin Zhang . High-order finite difference approximation of the Keller-Segel model with additional self- and cross-diffusion terms and a logistic source. Networks and Heterogeneous Media, 2023, 18(4): 1471-1492. doi: 10.3934/nhm.2023065
    [5] Steinar Evje, Kenneth H. Karlsen . Hyperbolic-elliptic models for well-reservoir flow. Networks and Heterogeneous Media, 2006, 1(4): 639-673. doi: 10.3934/nhm.2006.1.639
    [6] Raul Borsche, Axel Klar, T. N. Ha Pham . Nonlinear flux-limited models for chemotaxis on networks. Networks and Heterogeneous Media, 2017, 12(3): 381-401. doi: 10.3934/nhm.2017017
    [7] Avner Friedman . PDE problems arising in mathematical biology. Networks and Heterogeneous Media, 2012, 7(4): 691-703. doi: 10.3934/nhm.2012.7.691
    [8] Hirotada Honda . Global-in-time solution and stability of Kuramoto-Sakaguchi equation under non-local Coupling. Networks and Heterogeneous Media, 2017, 12(1): 25-57. doi: 10.3934/nhm.2017002
    [9] Caihong Gu, Yanbin Tang . Global solution to the Cauchy problem of fractional drift diffusion system with power-law nonlinearity. Networks and Heterogeneous Media, 2023, 18(1): 109-139. doi: 10.3934/nhm.2023005
    [10] Stefan Berres, Ricardo Ruiz-Baier, Hartmut Schwandt, Elmer M. Tory . An adaptive finite-volume method for a model of two-phase pedestrian flow. Networks and Heterogeneous Media, 2011, 6(3): 401-423. doi: 10.3934/nhm.2011.6.401
  • Though several AI-based models have been established for COVID-19 diagnosis, the machine-based diagnostic gap is still ongoing, making further efforts to combat this epidemic imperative. So, we tried to create a new feature selection (FS) method because of the persistent need for a reliable system to choose features and to develop a model to predict the COVID-19 virus from clinical texts. This study employs a newly developed methodology inspired by the flamingo's behavior to find a near-ideal feature subset for accurate diagnosis of COVID-19 patients. The best features are selected using a two-stage. In the first stage, we implemented a term weighting technique, which that is RTF-C-IEF, to quantify the significance of the features extracted. The second stage involves using a newly developed feature selection approach called the improved binary flamingo search algorithm (IBFSA), which chooses the most important and relevant features for COVID-19 patients. The proposed multi-strategy improvement process is at the heart of this study to improve the search algorithm. The primary objective is to broaden the algorithm's capabilities by increasing diversity and support exploring the algorithm search space. Additionally, a binary mechanism was used to improve the performance of traditional FSA to make it appropriate for binary FS issues. Two datasets, totaling 3053 and 1446 cases, were used to evaluate the suggested model based on the Support Vector Machine (SVM) and other classifiers. The results showed that IBFSA has the best performance compared to numerous previous swarm algorithms. It was noted, that the number of feature subsets that were chosen was also drastically reduced by 88% and obtained the best global optimal features.



    Predator-prey model is one of basic interspecies relations for ecological and social system [1]. The more complex biochemical network structure and food chain are based on the predator-prey model [2]. The study of Lotka and Volterra [3,4] has opened the way to study the dynamics of the predator-prey systems. After that, Gause and Smaragdova also proposed a well-known Gause-type predator-prey model. Kolmogorov first focused on the qualitative analysis of this Gause-type predator-prey model in 1972. Freedman [5] introduced the generalized autonomous Gause model, which comes from accounting for periodic changes of the environment. Gause-type predator-prey models have been widely applied to describe some population models [6,7,8,9]. For example, Hasik [6] considered the generalized Gause-type predator-prey model

    {x=xg(x)yp(x),y=y[q(x)γ], (1.1)

    here g(x) represents the increase in prey density. When the natural environment is relatively bad, the mortality rate of the population is higher than its birth rate, so the g(x) here can get a negative value. p(x) represents the amount of prey consumed by a single predator per unit time. q(x)γ represents the growth rate of the predator, and the same as g(x), q(x)γ can also be taken to a negative value. Ding et al. [7] considered the periodic Gause-type predator-prey system with delay

    {x(t)=x(t)f(t,x(tτ(t)))g(t,x(t))y(tσ1(t)),y(t)=y(t)[d(t)+h(t,x(tσ2(t)))], (1.2)

    where x(0),y(0)0 are the prey and the predator and obtained the positive periodic solution of this system (1.2) by using the continuation theorem.

    For the past few years, more and more researchers are interested in the dynamic behavior of predator-prey systems with Allee effect. The Allee effect describes that the low population is affected by the positive relationship between population growth rate and density, which increases the likelihood of their extinction. Terry [10] considered predator-prey systems with Allee effect and described how to extend the traditional definition of effective components and population Allee effect for a single species model to predators in the predator-prey model. Cui et al. [11] focused on the dynamic behavior and steady-state solutions of a class of reaction-diffusion predator-prey systems with strong Allee effect. Cai et al. [12] explored the complex dynamic behavior for a Leslie-Gower predation model with additive Allee effect on prey. Without considering the influence of Allee effect on prey, the model has a unique global asymptotically stable equilibrium point. However, considering the influence of Allee reaction on prey, the model has no definite positive equilibrium point [12]. Baisad and Moonchai [13] were interested in a Gause-type predator-prey model with Holling type-Ⅲ functional response and Allee effect on prey as follows

    {dxdt=r(1xK)(xm)xsx2x2+a2y,dydt=(px2x2+a2c)y. (1.3)

    Using the linearization method, they gave the local stability of three equilibrium types and also carried out a numerical simulation of the results. Guan and Chen [14] studied the dynamical analysis of a two species amensalism model with Beddington-DeAngelis functional response and Allee effect.

    The study of the dynamics of a harvested population is a topic studied in mathematical bioeconomics [15], inside a larger chapter dealing with optimal management of renewable resources. The exploitation of biological resources and the harvesting of interacting species is applied in fisheries, forestry and fauna management [15,16,17]. Etoua and Rousseau [16] studied a generalized Gause model with both prey harvesting and Holling response function of type Ⅲ:

    {dxdt=rx(1xK)mx2yax2+bx+1h1,dydt=y(d+cmx2ax2+bx+1),x0, y0, (1.4)

    where the eight parameters: r,k,m,a,c,d,h are strictly positive and b0. Through the following linear transformation and time scaling

    (X,Y,T)=(1kx,1cky,cmk2t).

    Laurin and Rousseau[17] transformed the model (1.4) into the simplified system with the number of parameters reduced to five

    {˙x=ρx(1x)yx2αx2+βx+1λ,˙y=y(δ+x2αx2+βx+1),x0, y0, (1.5)

    with parameters

    (ρ,α,β,δ,λ)=(rcmk2,ak2,bk,dcmk2,hcmk3).

    And the Hopf bifurcation was studied in [16,17]. Du et al. [18] considered a general predator-prey model with prey and predator harvesting and proved that the predator-prey system has at least four positive periodic solutions. In addition, some other predator-prey models have been studied widely [19,20,21,22].

    In this paper, we consider a generalized Gause-type predator-prey model with Allee effect, Holling type Ⅲ functional response and harvesting terms

    {dxdt=r(t)(1x(t)K)(x(t)m(t))x(t)s(t)x2x2(t)+a2(t)y(t)H1(t),dydt=(p(t)x2(t)x2(t)+a2(t)b(t)y(t)c(t))y(t)H2(t),x(0)>0, y(0)>0,  t[0,T], (1.6)

    where x=x(t) and y=y(t) represent the population sizes of prey and predator at time t, respectively. The size can represent numbers of individuals or density in the unit space of the population. To ensure biological significance, the parameter of K is positive, and a, b, c, H1, H2, m, p, r, s are positive Tperiodic functions. The meaning of the parameters in system (1.6) is given as follows:

     a is the amount of prey at which predation rate is maximal.

     b is the predator population decays in the competition among the predators.

     c is the natural per capita death rate of the predator.

     K is the environmental capacity of the prey.

     m is the minimum viable population.

     p is the conversion efficiency of reduction rate of the predator.

     r is the growth rate of the prey.

     s is the maximum per capita consumption rate.

    In system (1.6), the Allee effect is defined by the term r(t)(1x(t)K)(x(t)m(t))x(t) and the Holling type-Ⅲ functional response is represented by the term s(t)x2x2(t)+a2(t). This Holling type-Ⅲ functional response describes a behavior in which the number of prey consumed per predator initially increases quickly as the density of prey grows and levels off with further increase in prey density [13]. H1(t) and H2(t) describe the harvesting rate of prey and predators. We consider four important assumptions as regards the interactions between prey and predator:

    the prey population is affected by the Allee effect,

    the functional response is Holling type-Ⅲ,

    the influence of artificial harvest is considered on predator and prey, and

    the predator population decays in the competition among the predators is investigated.

    In this paper, we establish some conditions to ensure that system (1.6) has at least two positive periodic solutions. We outline the format for the rest of this paper. In Sect. 2, we describe several technical lemmas. In Sect. 3, using a systematic qualitative analysis and employing the Mawhin coincidence degree theory, we obtain that system (1.6) has at least two positive Tperiodic solutions of system (1.6).

    In this section, we will give relevant definitions of the Mawhin coincidence degree theory [23] and several technical lemmas.

    Let both X and Y be Banach spaces, L: DomLXY be a linear map and N: X×[0,1]Y be a continuous map. If ImLY is closed and dimKerL=codimImL<+, then we call the operator L is a Fredholm operator of index zero. If L is a Fredholm operator with index zero and there exists continuous projections P: XX and Q: YY such that ImP=KerL and ImL=KerQ=Im(IQ), then L|DomLKerP:(IP)XImL has an inverse function, and we set it as KP. Assume that Ω×[0,1]X is an open set. If QN(¯Ω×[0,1]) is bounded and KP(IQ)N(¯Ω×[0,1])X is relatively compact, then we say that N(¯Ω×[0,1]) is Lcompact.

    Next, we will give the Mawhin coincidence degree theorem.

    Lemma 2.1. ([23,24]) Let X and Y be two Banach spaces, L: DomLXY be a Fredholm operator with index zero, ΩY be an open bounded set, and N:¯Ω×[0,1]X be Lcompact on ¯Ω×[0,1]. If all the following conditions hold

    [C1] LxλNx, for xΩDomL, λ(0,1);

    [C2] QNx0, for every xΩKerL;

    [C3] Brouwer degree deg{JQN,ΩKerL,0}0, where J: ImQKerL is an isomorphism.

    Then the equation Lx=Nx has at least one solution on ¯ΩDomL.

    Lemma 2.2. ([19]) Let x>0, y>0, z>0 and x>2yz. For functions f(x,y,z)=x+x24yz2z and g(x,y,z)=xx24yz2z, the following assertions hold:

    (ⅰ) f(x,y,z) and g(x,y,z) are monotonically increasing and monotonically decreasing with respect to the variable x(0,);

    (ⅱ) f(x,y,z) and g(x,y,z) are monotonically decreasing and monotonically increasing with respect to the variable y(0,);

    (ⅲ) f(x,y,z) and g(x,y,z) are monotonically decreasing and monotonically increasing with respect to the variable z(0,).

    Throughout this paper, we denote by C[0,T] the space of all bounded continuous functions f: RR, and denote by C+ the set of all functions fC and f0. For the convenience of statement, we use the notations as follows

    ¯f=1TT0f(t)dt, fL=mint[0,T]f(t), fM=maxt[0,T]f(t).

    In this section, we will establish the existence results of at least two positive periodic solutions for the system (1.6).

    Theorem 2.1. Assume the following conditions hold:

    (H1) (pMu+2u2+(aL)2cL)2>4bLHL2;

    (H2) (pLu2u+2+(aM)2cM)2>4bMHM2;

    (H3) the algebra equation system

    {ˉr(1euK)(euˉm)ˉseueve2u+(ˉa)2¯H1eu=0,ˉpe2ue2u+(ˉa)2ˉbevˉc¯H2ev=0,

    has finite real-valued solutions (uk,vk), k=1, 2,..., n, satisfying

    (uk,vk)detG(uk, vk)nk=1uknk=1vk0,

    where

    G(uk,vk)=(ˉreu(ˉmK+1euK)ˉseueve2u+(ˉa)2+2ˉse3uev[e2u+(ˉa)2]2+¯H1e2uˉseueve2u+(ˉa)22ˉpe2ue2u+(ˉa)2+2ˉpe4u[e2u+(ˉa)2]2ˉbev+¯H2e2v).

    Then system (1.6) has at least two positive Tperiodic solutions.

    Proof of Theorem 2.1.. Suppose (x(t),y(t))R2 is an arbitrary positive of system (1.6). Let x=eu(t), y=ev(t), it follows from system (1.6) we can obtain

    {˙u(t)=r(t)(1eu(t)K)(eu(t)m(t))s(t)eu(t)ev(t)e2u(t)+a2(t)H1(t)eu(t),˙v(t)=p(t)e2u(t)e2u(t)+a2(t)b(t)ev(t)c(t)H2(t)ev(t), (3.1)

    where ˙=ddt.

    Let

    X=Y={z(t)=(u(t), v(t))TC(R, R2):z(t+T)z(t)},

    be equipped with the norm

    z(t)=(u(t), v(t))T=maxt[0,T]|u(t)|+maxt[0,T]|v(t)|,

    where X and Y are Banach spaces, T is the transpose.

    Taking zX and then we will define operators of L, P and Q as follows.

    Firstly, let

    L: DomLXY, Lz=dzdt.

    It is clear that

    KerL={zdomL: z=c, cR2},

    that is dimKerL=dimR2=2. Next we calculate ImL. Let

    dzdt=y(t), y(t)Y.

    Integrating both sides of this equation, we have

    T0dzdtdt=T0y(t)dt,

    thus

    T0y(t)dt=z(T)z(0)=0.

    From

    X=Y={z(t)=(u(t), v(t))TC(R, R2):z(t+T)z(t)},

    we can obtain y(t)=z(t), that is

    ImL={zY: T0z(t)dt=0}

    is closed in Y. Obviously, ImLR2={0}.

    Considering P, Q are both continuous projections satisfying

    ImP=KerL, ImL=KerQ=Im(IQ).

    Let

    P: XKerL,

    then, we get P(z) is a constant. Here, we denote it by

    P(z)=1TT0z(t)dt.

    Secondly, let

    Q: YYImL, β=T0z(t)dt and Q(z)=αβ,

    then, we have

    Q(Q(z))=Q(αβ)=αT0αβdt=α2βT0dt=α2βT=Q(z)=αβ,

    i.e.

    α=1T.

    Hence,

    Q(z)=1TT0z(t)dt.

    Thirdly, for zY, z1(t)=z(t)Q(z), we're going to verify z1(t)ImL, i.e. T0z1(t)dt=0. Here

    T0z1(t)dt=T0z(t)dtT0Q(z)dt=βαT0T0z(t)dtdt=βαβT=ββ=0,

    that is

    z1(t)ImL.

    Moreover, we can obtain

    Y=ImLR2, codimImL=dimR2=2.

    i.e.,

    dimKerL=codimImL.

    So L is a Fredholm operator with index zero, which implies L has a unique inverse. We define by KP: ImLKerPDomL the inverse of L.

    By simply calculating, we have

    KP(z)=t0z(w)dw1TT0t0z(w)dwdt.

    Define N: XY by the form

    Nz=(Δ1(z(t), t)Δ2(z(t), t)),

    where

    Δ1(z(t), t)=r(t)(1eu(t)K)(eu(t)m(t))s(t)eu(t)ev(t)e2u(t)+a2(t)H1(t)eu(t),Δ2(z(t), t)=p(t)e2u(t)e2u(t)+a2(t)b(t)ev(t)c(t)H2(t)ev(t).

    Thus

    QNz=(1TT0Δ1(z(t), t)dt1TT0Δ2(z(t), t)dt),

    and

    KP(IQ)Nz=(t0Δ1(z(w), w)dwt0Δ2(z(w), w)dw)(1TT0t0Δ1(z(w), w)dwdt1TT0t0Δ2(z(w), w)dwdt) (1Tt0T0Δ1(z(w), w)dwdw1Tt0T0Δ2(z(w), w)dwdw)+(1T2T0t0T0Δ1(z(w), w)dwdwdt1T2T0t0T0Δ2(z(w), w)dwdwdt)=(t0Δ1(z(w), w)dw1TT0t0Δ1(z(w), w)dwdt(tT12)T0Δ1(z(w), w)dwt0Δ2(z(w), w)dw1TT0t0Δ2(z(w), w)dwdt(tT12)T0Δ2(z(w), w)dw). (3.2)

    Let ΩX be bounded. For zΩ, we have that zM1, |u(t)|M1 and |v(t)|M1.

    Next, we see that QN(¯Ω) is bounded.

    |1TT0Δ1(z(t), t)dt||1TT0r(t)(1eu(t)K)(eu(t)m(t))dt| +|1TT0s(t)eu(t)ev(t)e2u(t)+a2(t)dt|+|1TT0H1(t)eu(t)dt||1TT0[r(t)eu(t)r(t)m(t)r(t)e2u(t)K+r(t)m(t)eu(t)K]dt| +|1TT0s(t)eu(t)ev(t)a2(t)dt|+|1TT0H1(t)eu(t)dt|ˉreM1+¯rm+ˉre2M1K+¯rmeM1K+¯(sa2)e2M1+¯H1eM1,

    and

    |1TT0Δ2(z(t), t)dt||1TT0p(t)e2u(t)e2u(t)+a2(t)dt|+|1TT0b(t)ev(t)dt| +|1TT0c(t)dt|+|1TT0H2(t)ev(t)dt||1TT0p(t)e2u(t)+p(t)a2(t)e2u(t)+a2(t)dt|+ˉbeM1+ˉc+¯H2eM1=|1TT0p(t)dt|+ˉbeM1+ˉc+¯H2eM1=ˉp+ˉbeM1+ˉc+¯H2eM1.

    It is immediate that QN and KP(IQ)N are continuous.

    Consider a sequence of function {z}Ω. We have the following inequality for the first function of KP(IQ)Nz.

     KP(IQ)NΔ1(z(t1), t1)KP(IQ)NΔ1(z(t2), t2)=t1t2[r(w)(1eu(w)K)(eu(w)m(w))s(w)eu(w)ev(w)e2u(w)+a2(w)H1(w)eu(w)]dw (t1t2T)T0[r(w)(1eu(w)K)(eu(w)m(w))s(w)eu(w)ev(w)e2u(w)+a2(w)H1(w)eu(w)]dw(t1t2)[rMeM1rLmLrLe2M1K+rMmMeM1K] (t1t2)[ˉreM1¯rmˉre2M1K+¯rmeM1K].

    For another function, we have similar inequalities as follows

     KP(IQ)NΔ2(z(t1), t1)KP(IQ)NΔ2(z(t2), t2)=t1t2[p(w)e2u(w)e2u(w)+a2(w)b(w)ev(w)c(w)H2(w)ev(w)]dw (t1t2T)T0[p(w)e2u(w)e2u(w)+a2(w)b(w)ev(w)c(w)H2(w)ev(w)]dw(t1t2)[pMe2M1a2LbLeM1cLH2LeM1] +(t1t2)[ˉbeM1+ˉc+¯H2eM1].

    Hence the sequence {KP(IQ)Nz} is equicontinuous. Using the periodicity of the functions, we know that the sequence {KP(IQ)Nz} is uniformly bounded.

    An application of Ascoli-Arzela's theorem shows that {KP(IQ)N(¯Ω)} is compact for any bounded set ΩX. Since QN(¯Ω) is bounded, we conclude that N is Lcompact on Ω for any bounded set ΩX.

    Then, considering the operator equation Lx=λNx, as follows

    {˙u(t)=λΔ1(z(t), t),˙v(t)=λΔ2(z(t), t), (3.3)

    where λ(0,1). Let

    u(ξ1)=maxt[0,T]u(t),  u(η1)=mint[0,T]u(t),
    v(ξ2)=maxt[0,T]v(t),  v(η2)=mint[0,T]v(t).

    Through simple analysis, we have

    ˙u(ξ1)=˙u(η1)=0, ˙v(ξ2)=˙v(η2)=0.

    From (3.3), we can find that

    r(ξ1)(1eu(ξ1)K)(eu(ξ1)m(ξ1))s(ξ1)eu(ξ1)ev(ξ1)e2u(ξ1)+a2(ξ1)H1(ξ1)eu(ξ1)=0, (3.4)
    p(ξ2)e2u(ξ2)e2u(ξ2)+a2(ξ2)b(ξ2)ev(ξ2)c(ξ2)H2(ξ2)ev(ξ2)=0, (3.5)

    and

    r(η1)(1eu(η1)K)(eu(η1)m(η1))s(η1)eu(η1)ev(η1)e2u(η1)+a2(η1)H1(η1)eu(η1)=0 (3.6)
    p(η2)e2u(η2)e2u(η2)+a2(η2)b(η2)ev(η2)c(η2)H2(η2)ev(η2)=0, (3.7)

    In view of (3.4), we have

    r(ξ1)(1eu(ξ1)K)(eu(ξ1)m(ξ1))=s(ξ1)eu(ξ1)ev(ξ1)e2u(ξ1)+a2(ξ1)+H1(ξ1)eu(ξ1)>0,

    then, we get

    rLKe2u(ξ1)(rM+rMmMK)eu(ξ1)+rLmL<0,

    which implies that

    eu=:rM+rMmMK(rM+rMmMK)24(rL)2mLK2rLK<eu(ξ1)<rM+rMmMK+(rM+rMmMK)24(rL)2mLK2rLK:=eu+.

    Similarly, we can discuss the range of eu(η1) from (3.6)

    r(η1)(1eu(η1)K)(eu(η1)m(η1))=s(η1)eu(η1)ev(η1)e2u(η1)+a2(η1)+H1(η1)eu(η1)>0,

    A direct calculation gives

    rLKe2u(η1)(rM+rMmMK)eu(η1)+rLmL<0,

    so we can obtain

    eu=:rM+rMmMK(rM+rMmMK)24(rL)2mLK2rLK<eu(η1)<rM+rMmMK+(rM+rMmMK)24(rL)2mLK2rLK:=eu+.

    From (3.5), we have

    b(ξ2)e2v(ξ2)(p(ξ2)e2u(ξ2)e2u(ξ2)+a2(ξ2)c(ξ2))ev(ξ2)+H2(ξ2)=0,

    and

    bLe2v(ξ2)(pMe2u+e2u+(aL)2cL)ev(ξ2)+H2L<0,

    then, we get

    ev=:pMe2u+e2u+(aL)2cL(pMe2u+e2u+(aL)2cL)24bLH2L2bL<ev(ξ2)<pMe2u+e2u+(aL)2cL+(pMe2u+e2u+(aL)2cL)24bLH2L2bL:=ev+.

    By (3.7), we obtain

    b(η2)e2v(η2)(p(η2)e2u(η2)e2u(η2)+a2(η2)c(η2))ev(η2)+H2(η2)=0,

    and

    bLe2v(η2)(pMe2u+e2u+(aL)2cL)ev(η2)+H2L<0,

    this implies that

    ev=:pMu2+e2u+(aL)2cL(pMu2+e2u+(aL)2cL)24bLH2L2bL<ev(η2)<pMu2+e2u+(aL)2cL+(pMe2u+e2u+(aL)2cL)24bLH2L2bL:=ev+.

    And then, in view of (3.5) and (3.7) we have

    bMe2v(ξ2)(pLe2ue2u++(aM)2cM)ev(ξ2)+H2M>0,

    that is

    ev(ξ2)>pLe2ue2u++(aM)2cM+(pLe2ue2u++(aM)2cM)24bMH2M2bM:=el+

    or

    ev(ξ2)<pLe2ue2u++(aM)2cM(pLe2ue2u++(aM)2cM)24bMH2M2bM:=el.

    From (3.7), we obtain

    bMe2v(η2)(pLe2ue2u++(aM)2cM)ev(η2)+H2M>0,

    i.e.

    ev(ξ2)>l+ or ev(ξ2)<el.

    In view of Lemma 2.2, we can find that v<l<l+<v+. Thus, we have

    {v(ξ2)>l+ or v(ξ2)<l,v<v(ξ2)<v+,

    and

    {v(η2)>l+ or v(η2)<l,v<v(η2)<v+,

    that is

    v(ξ2)(v, l)(l+, v+), v(η2)(v, l)(l+, v+).

    Similarly, we get

    u(ξ1)(u, u+), u(η1)(u, u+).

    It is easy to know that u±,v±,l± are independent of λ. Consider the following two sets

    Ω1={z=(u,v)TX|u<u< u+, v<v< l},
    Ω2={z=(u,v)TX|u<u< u+, l+<v< v+}.

    Obviously, ΩiX and ¯Ω1¯Ω2=. So Ωi's (i=1, 2) satisfy the condition [C1] of Lemma 2.1.

    Next, we show that QNz0, for zΩiKerL=ΩiR2 (i=1,2).

    If it is not true, then there exists (u,v)TΩi, such that

    {T0Δ1(z(t), t)dt=0,T0Δ2(z(t), t)dt=0.

    By virtue of the mean value theorem, there exists two points tj[0,T] (j=1, 2) satisfying

    {Δ1(z(t1), t1)=0,Δ2(z(t2), t2)=0.

    So, we obtain

    u(u, u+),

    and

    v(v, l)(l+, v+),

    which contradicts (u,v)TΩiR2. So the condition [C2] in Lemma 2.1 holds.

    Then, we check the condition [C3] in Lemma 2.1. Define the homomorphism J: ImQKerL, Jzz. From [H3], we have

    deg{JQN,ΩiKerL,0}=zkQN1(0)sgnJQN(zk)=zkQN1(0)detG(uk,vk)nk=1uknk=1vk0.

    This implies that the condition [C3] in Lemma 2.1 holds too. Note that, Ω1 and Ω2 satisfies all conditions of Lemma 2.1. Therefore, system (1.6) has at least two Tperiodic solutions.

    Here, we would like to give two remarks.

    Remark 2.1. If we take H1(t)=0,H2(t)=0 and b(t)=0, i.e., system (1.6) without considering the harvesting terms of prey and predator, as well as the predator competition, we find that the system (1.3) in [13] is the system (1.6).

    Remark 2.2. In [16] and [17], the authors only considered the Gause model (1.4) and (1.5) with prey harvesting h1, respectively, but they don't investigate the influence of the predator harvesting. In fact, the influence of the predator harvesting is very important in biological populations and bioeconomics, especially in fisheries management etc.

    In this paper, we are concerned with a Gause-type predator-prey model with Allee effect, Holling type Ⅲ functional response and the artificial harvesting terms, which are very important in biological populations and bioeconomics. Four important assumptions as regards the interactions between prey and predator is considered. By applying the Mawhin coincidence degree theory, we obtain the existence of multiple positive periodic solutions for the predator-prey model.

    We express our sincere thanks to the anonymous reviewers for their valuable comments and suggestions. This work is supported by the Natural Science Foundation of China (Grant No.11771185). The first author supported by Postgraduate Research and Practice Innovation Program of Jiangsu Province (Grant Nos. KYCX18-2091, KYCX20-2082 and KYCX20-2206).

    The authors declare no conflict of interest in this paper.



    [1] C. Li, C. Zhao, J. Bao, B. Tang, Y. Wang, B. Gu, Laboratory diagnosis of coronavirus disease-2019 (COVID-19), Clin. Chim. Acta., 510 (2020), 35–46. https://doi.org/10.1016/j.cca.2020.06.045 doi: 10.1016/j.cca.2020.06.045
    [2] Y. Guo, Q. Cao, Z. Hong, Y. Tan, S. Chen, H. Jin, et al., The origin, transmission and clinical therapies on coronavirus disease 2019 (COVID-19) outbreak- A n update on the status, Mil. Med. Res., 7 (2020), 1–10. https://doi.org/10.1186/s40779-020-00240-0 doi: 10.1186/s40779-020-00240-0
    [3] M. Rostami, M. Oussalah, A novel explainable COVID-19 diagnosis method by integration of feature selection with random forest, Inform. Med. Unlocked, 30 (2022), 100941. https://doi.org/10.1016/j.imu.2022.100941 doi: 10.1016/j.imu.2022.100941
    [4] X. Luo, P. Gandhi, S. S. KH, A deep language model for symptom extraction from clinical text and its application to extract COVID-19 symptoms from social media, IEEE J. Biomed. Heal Inform., 26 (2022), 1737–1748. https://doi.org/10.1109/JBHI.2021.3123192 doi: 10.1109/JBHI.2021.3123192
    [5] G. Saranya, A. Pravin, Feature selection techniques for disease diagnosis system: A survey, in Artificial Intelligence Techniques for Advanced Computing Applications, Springer, Singapore, 130 (2021), 249–258. https://doi.org/10.1007/978-981-15-5329-5_24
    [6] J. T. Pintas, L. A. F. Fernandes, A. C. B. Garcia, Feature selection methods for text classification: A systematic literature review, Artif. Intell. Rev., 54 (2021), 6149–6200. https://doi.org/10.1007/s10462-021-09970-6 doi: 10.1007/s10462-021-09970-6
    [7] L. M. Abualigah, A. T. Khader, E. S. Hanandeh, A new feature selection method to improve the document clustering using particle swarm optimization algorithm, J. Comput. Sci., 25 (2018), 456–466. https://doi.org/10.1016/j.jocs.2017.07.018 doi: 10.1016/j.jocs.2017.07.018
    [8] D. A. Elmanakhly, M. Saleh, E. A. Rashed, M. Abdel-Basset, BinHOA : Efficient binary horse herd optimization method for feature selection : Analysis and validations, IEEE Access., 10 (2022), 26795–26816. https://doi.org/10.1109/ACCESS.2022.3156593 doi: 10.1109/ACCESS.2022.3156593
    [9] R. Abu Khurmaa, I. Aljarah, A. Sharieh, An intelligent feature selection approach based on moth flame optimization for medical diagnosis, Neural Comput. Appl., 33 (2021), 7165–7204. https://doi.org/10.1007/s00521-020-05483-5 doi: 10.1007/s00521-020-05483-5
    [10] P. H. Prastyo, R. Hidayat, I. Ardiyanto, Enhancing sentiment classification performance using hybrid query expansion ranking and binary particle swarm optimization with adaptive inertia weights, ICT Express., 8 (2021), 189–197. https://doi.org/10.1016/j.icte.2021.04.009 doi: 10.1016/j.icte.2021.04.009
    [11] B. Ji, X. Lu, G. Sun, W. Zhang, J. Li, Y. Xiao, Bio-Inspired feature selection : An improved binary particle swarm optimization approach, IEEE Access., 8 (2020), 85989–86002. https://doi.org/10.1109/ACCESS.2020.2992752 doi: 10.1109/ACCESS.2020.2992752
    [12] H. K. H. Chantar, M. M. Mafarja, H. I. Alsawalqah, A. A. Heidari, I. Aljarah, H. Faris, Feature selection using binary grey wolf optimizer with elite-based crossover for Arabic text classification, Neural Comput. Appl., 32 (2020), 12201–12220. https://doi.org/10.1007/s00521-019-04368-6 doi: 10.1007/s00521-019-04368-6
    [13] M. H. Nadimi-Shahraki, S. Taghian, S. Mirjalili, L. Abualigah. Binary aquila optimizer for selecting effective features from medical data: A COVID-19 case study, Math. MDPI., 10 (2022), 1–24. https://doi.org/10.3390/math10111929 doi: 10.3390/math10111929
    [14] J. Piri, P. Mohapatra, B. Acharya, F. S. Gharehchopogh, V. C. Gerogiannis, A. Kanavos, et al., Feature selection using artificial gorilla troop optimization for biomedical data: A case analysis with COVID-19 data, Mathematics, 10 (2022), 1–31. https://doi.org/10.3390/math10152742 doi: 10.3390/math10152742
    [15] W. Tuerxun, X. Chang, G. Hongyu, J. Zhijie, Z. Huajian, Fault diagnosis of wind turbines based on a support vector machine optimized by the sparrow search algorithm, IEEE Power Energy Soc. Sect., 9 (2021), 69307–69315. https://doi.org/10.1109/ACCESS.2021.3075547 doi: 10.1109/ACCESS.2021.3075547
    [16] C. A. Flores, R. L. Figueroa, J. E. Pezoa, FREGEX: A feature extraction method for biomedical text classification using regular expressions, in 2019 41st Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC), (2019), 6085–6088. https://doi.org/10.1109/EMBC.2019.8857471
    [17] W. M. Shaban, A. H. Rabie, A. I. Saleh, M. A. Abo-Elsoud, Accurate detection of COVID-19 patients based on distance biased Naïve Bayes (DBNB) classification strategy, Pattern Recognit., 119 (2021), 108110–108110. https://doi.org/10.1016/j.patcog.2021.108110 doi: 10.1016/j.patcog.2021.108110
    [18] A. Singh, K. K. Singh, M. Greguš, I. Izonin, CNGOD-An improved convolution neural network with grasshopper optimization for detection of COVID-19, Math. Biosci. Eng., 9 (2022), 12518–12531. https://doi.org/10.3934/mbe.2022584 doi: 10.3934/mbe.2022584
    [19] Z. M. Fadhil, R. A. Jaleel, Multiple efficient data mining algorithms with genetic selection for prediction of SARS-CoV2, in 2022 2nd International Conference on Advance Computing and Innovative Technologies in Engineering (ICACITE), (2022). https://doi.org/10.1109/ICACITE53722.2022.9823757
    [20] I. M. El-Hasnony, M. Elhoseny, Z. Tarek, A hybrid feature selection model based on butterfly optimization algorithm: COVID‐19 as a case study, Expert Syst., 39 (2022), e12786. https://doi.org/10.1111/exsy.12786 doi: 10.1111/exsy.12786
    [21] M. A. k. alsaeedi, S. Kurnaz, Feature selection for diagnose coronavirus (COVID-19) disease by neural network and Caledonian crow learning algorithm, Appl Nanosci., (2022), 1–16. https://doi.org/10.1007/s13204-021-02159-x doi: 10.1007/s13204-021-02159-x
    [22] T. Bezdan, M. Zivkovic, N. Bacanin, A. Chhabra, M. Suresh, Feature selection by hybrid brain storm optimization algorithm for COVID-19 classification, J. Comput. Biol., 29 (2022), 515–529. https://doi.org/10.1089/cmb.2021.0256 doi: 10.1089/cmb.2021.0256
    [23] Z. Wang, J. Liu, Flamingo search algorithm and its application to path planning problem, in 2021 4th Flamingo search algorithm and its application to path planning problem, (2021), 567–573. https://doi.org/10.1145/3488933.3489011
    [24] A. Onan, M. A. Toçoğlu, A term weighted neural language model and stacked bidirectional LSTM based framework for sarcasm identification, IEEE Access, 9 (2021), 7701–7722. https://doi.org/10.1109/ACCESS.2021.3049734 doi: 10.1109/ACCESS.2021.3049734
    [25] M. Neumann, D. King, I. Beltagy W. Ammar, ScispaCy: Fast and robust models for biomedical natural language processing, in Proceedings of the 18th BioNLP Workshop and Shared Task, (2019), 319–327. https://doi.org/10.18653/v1/W19-5034
    [26] A. Y. Mahdi, S. S. Yuhaniz, Automatic diagnosis of COVID-19 patients from unstructured data based on a novel weighting scheme, C. Mater. Contin., 74 (2022), 1375–1392. https://doi.org/10.32604/cmc.2023.032671 doi: 10.32604/cmc.2023.032671
    [27] T. Parlar, S. A. Özel, F. Song, A new feature selection method for sentiment analysis, Human-centric Comput. Inf. Sci., 8 (2018), 1–19. https://doi.org/10.1515/jisys-2018-0171 doi: 10.1515/jisys-2018-0171
    [28] S. L. Marie-Sainte, N. Alalyani, Firefly algorithm based feature selection for arabic text classification, J. King Saud Univ. Comput. Inf. Sci., 32 (2020), 320–328, https://doi.org/10.1016/j.jksuci.2018.06.004 doi: 10.1016/j.jksuci.2018.06.004
    [29] W. Zhiheng, L. Jianhua, Flamingo search algorithm: A new swarm intelligence optimization algorithm, IEEE Access., 9 (2021), 88564–88582. https://doi.org/10.1109/ACCESS.2021.3090512 doi: 10.1109/ACCESS.2021.3090512
    [30] M. Abd El Aziz, A. Hassanien, Modified cuckoo search algorithm with rough sets for feature selection, Neural Comput. Appl., 29 (2018), 925–934. https://doi.org/10.1007/s00521-016-2473-7 doi: 10.1007/s00521-016-2473-7
    [31] Z. Li, Y. Zhou, S. Zhang, J. Song, Lévy-Flight Moth-Flame algorithm for function optimization and engineering design problems, Math. Probl. Eng., (2016), 1–22. https://doi.org/10.1155/2016/1423930 doi: 10.1155/2016/1423930
    [32] P. A. Digehsara, S. N. Chegini, A. Bagheri, M. P. Roknsaraei, An improved particle swarm optimization based on the reinforcement of the population initialization phase by scrambled Halton sequence, Cogent. Eng., 7 (2020), 1–29. https://doi.org/10.1080/23311916.2020.1737383 doi: 10.1080/23311916.2020.1737383
    [33] B. Kazimipour, X. Li, A. K. Qin, A review of population initialization techniques for evolutionary algorithms, 2014 IEEE Congr. Evol. Comput., (2014), 2585–2592. https://doi.org/10.1109/CEC.2014.6900618 doi: 10.1109/CEC.2014.6900618
    [34] W. H. Bangyal, A. Hameed, W. Alosaimi, H. Alyami, A new initialization approach in particle swarm optimization for global optimization problems, Comput. Intell. Neurosci., 2021 (2021), 1–17. https://doi.org/10.1155/2021/6628889 doi: 10.1155/2021/6628889
    [35] A. G. Gad, K. M. Sallam, R. K. Chakrabortty, M. J. Ryan, A. A. Abohany, An improved binary sparrow search algorithm for feature selection in data classification, Neural Comput. Appl., 34 (2022), 15705–15752. https://doi.org/10.1007/s00521-022-07546-1 doi: 10.1007/s00521-022-07546-1
    [36] P.H. Prastyo, A.S. Sumi, A.W. Dian, A. E Permanasari, Tweets responding to the Indonesian government's handling of COVID-19: Sentiment analysis using SVM with Normalized Poly Kernel, J. Inf. Syst. Eng. Bus. Intell., 6 (2020), 112–122. https://doi.org/10.20473/jisebi.6.2.112-122 doi: 10.20473/jisebi.6.2.112-122
    [37] K. Kowsari, K. Meimandi, M. Heidarysafa, S. Mendu, L. E. Barnes, D. E. Brown, Text classification algorithms : A survey, Inf. J., 10 (2019), 1–68. https://doi.org/10.3390/info10040150 doi: 10.3390/info10040150
    [38] M. Qaraad, S. Amjad, I. I. M. Manhrawy, H. Fathi, B. A. Hassan, P. E. Kafrawy, A hybrid feature selection optimization model for high dimension data classification, IEEE Access., 9 (2021), 42884–42895. https://doi.org/10.1109/ACCESS.2021.3065341 doi: 10.1109/ACCESS.2021.3065341
  • Reader Comments
  • © 2023 the Author(s), licensee AIMS Press. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0)
通讯作者: 陈斌, bchen63@163.com
  • 1. 

    沈阳化工大学材料科学与工程学院 沈阳 110142

  1. 本站搜索
  2. 百度学术搜索
  3. 万方数据库搜索
  4. CNKI搜索

Metrics

Article views(2628) PDF downloads(56) Cited by(7)

Figures and Tables

Figures(11)  /  Tables(18)

Other Articles By Authors

/

DownLoad:  Full-Size Img  PowerPoint
Return
Return

Catalog