Dismiss Notice
Join Physics Forums Today!
The friendliest, high quality science and math community on the planet! Everyone who loves science is here!

Logistic growth model for cell proliferation with agents

  1. Nov 9, 2011 #1
    Hello community,

    in project group im doing some in which I definately need your help.
    Basically what Im intending to do is to examine the cell proliferation
    of cell of the colorectal carcinoma. The data I will receive is from
    the xCELLigence system and generates so called cell indices, basically over
    a resistence measurement. These are values for approximately 72 hrs with
    the intervall of 5 minues. Meaning I would have about 864 cell indices.

    In addition during the experiments inhibitors or stimulators will be added
    on to the cells, resulting in a change of the growth.

    However for the modelling via differential equation with the logistic model,
    I would further need to include parameters for the strength of the agent
    and maybe taking the half-life into consideration. Such that the growth can
    be identified at every time.

    So basically after treatment the growth curve could show 4 characteristics
    in respect to the control:
    1) curve shifts on the x-axis to the right (inhibition),
    2) curve shifts on the x-axis to the left (stimulation), which is basically
    a delayed growth and accerlerated grwoth.
    -curve shifts on the y-axis up (stimulation)
    -curve shifts on the y-axis down (inhibition)
    resulting in a higher and lower growth rate during a curve.

    Could I now just express a value for the agent and the half-life as part of an
    exponential grwoth into the equation and do I need to distinguish between
    inhibitor and stimulator? Or how can i express stagnation.
    I have right now no clue how to approach this task. Hope somebody can give me hint
    or literature?

  2. jcsd
  3. Nov 10, 2011 #2
    Or maybe it is possible to ignore inhibition and stimulation?
    After all I only need to derive a growth factor over a certain time intervall.
    The difference I could calculate in reference to the control.
    In this case the logistic function should be sufficient?
Share this great discussion with others via Reddit, Google+, Twitter, or Facebook