Icraet.pptx

download Icraet.pptx

of 20

Transcript of Icraet.pptx

  • 7/27/2019 Icraet.pptx

    1/20

    MODELLING THE TRANSMISSIONOF VECTOR BORNE DISEASESUSING SVM-PSOSudheer Ch.

  • 7/27/2019 Icraet.pptx

    2/20

    2ICRAET

    IIT Delhi

    Synopsis

    Introduction Problem formulation

    SVM

    PSO SVM-PSO

    Application to case study

    Results and Discussions

    Conclusions

  • 7/27/2019 Icraet.pptx

    3/20

    3ICRAET

    IIT Delhi

    INTRODUCTION

    Vector : Any agent (person or animal ormicroorganism) that carries and transmits a

    disease (Lexicographic meaning)

    Mosquitoes

    Flies etc.

    Dengue

    Chickenguniy

    a

    Malaria

    Vector

    Borne

    Diseas

    es

  • 7/27/2019 Icraet.pptx

    4/20

    4ICRAET

    IIT Delhi

    Malaria: Infection

    Anophelesmosquito bites

    infected host

    Mosquito drinks parasite in

    blood

    Parasite reproduces in gut of

    mosquito

    Sporozoites travel to

    mosquitos salivary glands Mosquito transmits disease

    during next blood meal.

  • 7/27/2019 Icraet.pptx

    5/20

    5ICRAET

    IIT Delhi

    Last few

    decades there is

    an increase invector borne

    diseases.

    Researchers have attributed that one

    of the main reason for this isClimate change

  • 7/27/2019 Icraet.pptx

    6/20

    6ICRAET

    IIT Delhi

    Malaria among all vector born

    diseases is specially becoming a big

    threat to human life and health in

    the country, and hence in this

    paper study on the malaria has

    been taken up.

    Climate change or global warmingis predicted to have unexpected

    effects on the vector prevalence

    thereby affecting malaria

    incidences.

  • 7/27/2019 Icraet.pptx

    7/207ICRAET

    IIT Delhi

    Entomological and parasitological parameters

    Parameter Value

    1. Optimum Temperature for viability ofeggs

    28C

    2. Anthropophilic Index 0 to 4%

    3. Gonotrophic cycle 78hrs-132hrs (3 days) after first blood mealAverage-96hrs

    40 hrs at temp 26-32C

    4. Biting Behavior/ man landing rate 0.4/man/night in plain0.95/man /night in forest

    0.2/man/night in dam average 0.2-0.9

    5. Longevity 4 to 8 weeks -12 to 18C12to 28 days-25

    6 to 18 days-30C4 to 10 days- 35C (at 20 to 100% RH)

    >24 hours-40C

    With RH 60 -80%

    6. Daily Survival probability 0.81 to 0.89

    7. Entomological Inoculation Rate (EIR) P.v-0.06-0.12

    P.f-0.06

  • 7/27/2019 Icraet.pptx

    8/208ICRAET

    IIT Delhi

    131900 21002000

    20

    15

    14

    16

    17

    18

    19Earths AverageSurface Temp (OC)

    Year

    205019501860

    Centralestimate:

    2.5 oCincrease

    Band o f 1200-yr historicalclimatic variabil i ty

    Most of warming since1950 is due to humanactions (IPCC, 2001)

    IPCC (2001)

    estimate:

    + 1.4-5.8 oC by 2100

    Past ClimateMean surface temperature, 1855-2004

    Climate Research Unit, UEA, 2005

    Temperature variation

    from 1961-90 average oC

    Temperature increase and

    fluctuation affects the vector

    and parasite life cycle. This

    can cause reduced prevalenceof the disease in some areas,

    while it may increase in

    others [3-7].

    Do you Know?

    There were 243 millionmalaria cases reported

    in 2008 (WHO 2009).

  • 7/27/2019 Icraet.pptx

    9/20

    How to address this problem?

    If malarial incidences can be predicted in

    advance then Health dept can take various

    precautions .

    So in this study SVM-PSO is used forforecasting the malarial outbreaks.

  • 7/27/2019 Icraet.pptx

    10/2010ICRAET

    IIT Delhi

    Support Vector Machines

    Support vector machine

    constructs a hyperplane or set ofhyperplanes in a high- or infinite-

    dimensional space, which can be

    used for classification, regression,

    or other tasks.

    Intuitively, a good separation isachieved by the hyperplane that

    has the largest distance to the

    nearest training data point of any

    class (so-called functional

    margin), since in general thelarger the margin the lower

    the generalization error of the

    classifier.

  • 7/27/2019 Icraet.pptx

    11/20

    Unique Features of SVM SVM training always finds

    global minimum

    The computational

    complexity of SVMs does not

    depend on the dimensionalityof the input space.

    SVMs use structural risk

    minimization.

    SVMs are less prone tooverfitting

    Local minimaGlobal minima

    Parameter Estimation C, epsilion, gamma,

    Kernel function.- DEFECTS OF SVM

  • 7/27/2019 Icraet.pptx

    12/2012ICRAET

    IIT Delhi

    Particle Swarm Optimization (PSO) PSO was developed in 1995 by

    James Kennedy (social-

    psychologist) and RussellEberhart (electrical engineer).

    It uses a number of agents

    (particles) that constitute a

    swarm moving around in the

    search space looking for thebest solution.

    Each particle is treated as a

    point in a N-dimensional space

    which adjusts its flying

    according to its own flying

    experience as well as the flying

    experience of other particles.

  • 7/27/2019 Icraet.pptx

    13/20

    13ICRAET

    IIT Delhi

    Each particle keeps track of its coordinates in the solutionspace which are associated with the best solution (fitness)

    that has achieved so far by that particle. This value is

    called personal best , pbest.

    Another best value that is tracked by the PSO is the bestvalue obtained so far by any particle in the neighborhood

    of that particle. This value is called gbest.

    The basic concept of PSO lies in accelerating each particle

    toward its pbest and the gbest locations, with a randomweighted acceleration at each time step.

    Particle Swarm Optimization (PSO)

  • 7/27/2019 Icraet.pptx

    14/20

    14ICRAET

    IIT Delhi

    Flow Chart for computing SVM parameters

    using PSO

  • 7/27/2019 Icraet.pptx

    15/20

    15ICRAET

    IIT Delhi

    Determining SVM parameters (Hsu etal 2009)

    Coarse search Fine search

    C [10-5,105] [ 10-1,10]

    [ 0,10 ] [ 10-7,10-1 ]

    [ 0,10 ] [ 0,1 ]

  • 7/27/2019 Icraet.pptx

    16/20

    16ICRAET

    IIT Delhi

    Case Study

    Maximum and Minimum Temperatures in area of Falodi.

    Relative Humidity and malarial incidences .

    Falodi area of Jodhpur District

    Latitude 27-06 to 27-09 north

    and 72-20 to 72-23 east.

    Average elevation of 303 meters.

    Average temperature in summer

    falling in the range of 42.2 C (max)

    to 36.6 C (min).

    The average temperature in thewinter season is somewhere between

    27.5 C (max) to 15.5 C (min).

  • 7/27/2019 Icraet.pptx

    17/20

    17ICRAET

    IIT Delhi

    Comparison of SVM-PSO with other models

    0

    500

    1000

    1500

    2000

    2500

    3000

    3500

    4000

    4500

    5000

    1 2 3 4 5 6 7 8 9

    MalariaIncidences

    Months

    Actual Malaria Incidences

    SVM-PSO

    SVM-GA

    SVM

  • 7/27/2019 Icraet.pptx

    18/20

    18ICRAET

    IIT Delhi

    Conclusions

    SVM conducts structural minimization rather than the minimizationof the errors.

    The PSO selects the optimal parameters for SVM to improve the

    forecasting accuracy.

    So this unique combination of SVM and PSO has made the proposed

    SVM-PSO model to perform better compared to the other models. The performance of SVM-PSO is out performing comparing to other

    models like SVM-GA and SVM

    Also it is seen from the results that SVM performs poorly in case the

    parameters are not chosen properly.

    Furthermore this investigation demonstrates that the proposed SVM-PSO offers a valid alternative for predicting the malaria outbreaks.

  • 7/27/2019 Icraet.pptx

    19/20

  • 7/27/2019 Icraet.pptx

    20/20

    20ICRAET

    IIT Delhi