Script Type	Simulate dosing and sampling data only (gen) [docs] Simulate dosing and sampling data, and fit model (gen + fit) [docs]
No. of Pharmacokinetic Compartments	One [docs] Two [docs] Three [docs]
Absorption Model	Intravenous (IV) dosing [docs] Oral dosing with First-order absorption [docs]
Derivatives Solver	Analytic (ADVANx) [docs] Numeric (ODEs) [docs]
Inter-Occasion Variability (IOV)	Without IOV [docs] With IOV [docs]
Comments	With comments Without comments

Edit Script

# Rules:
#   Comments are prefixed with the hash (#) character.
#   Sections must start with an indent that is a multiple of 4 spaces.
METHOD_OPTIONS:
    py_module: gen

# Uncomment the following section to add metadata to the model

# DESCRIPTION:
    # name: my_model
    # title: My PK/PD Model
    # author: J.R. Hartley
    # abstract: |

# A longer description of my model.

# keywords: ['pk', 'dep_two_cmp', 'first order']

EFFECTS:
    # Pseudocode comes in sections that have a trailing "|"
    # and is processed to become executable Python code.
    # "POP"(ulation) is the top level of the effects tree.
    POP: |

# Define covariates with
        #   c[X] = k for constants, or
        #   c[X] ~ <dist>() for a randomly sampled variable
        c[AMT] = 100.0

# Define fixed effects with
        #   f[X] = k for constants, or
        #   f[X] ~ <dist>() for a randomly sampled variable
        f[VC] = 10.0  # Volume of the central compartment
        f[CL] = 2.0  # Clearance

f[ANOISE] = 0.1  # Additive noise variance
        f[PNOISE] = 0.15  # Proportional noise variance

# The tree then branches, usually with one branch per individual.
    ID: |

# Define 20 individuals via a sequence of length 20
        # (i.e., 1, 2, 3, ..., 20)
        # Note that this covariate has the same name (ID) as the
        # level in the tree. This is important for fitting when it
        # partitions the table of data into a tree again, based on
        # a match between the tree level and the data column (e.g. ID).
        c[ID] = sequential(20)

# Define times of events with t[RESET], t[DOSE] and t[OBS]
        t[RESET] = 0.0
        t[DOSE] = 0.0
        t[OBS] = [1.0, 1.5, 2.0, 4.0, 8.0, 12.0]

# Define a covariance matrix for multiple variables
        # using the lower triangular entries.
        f[VC_isv,CL_isv] = [
            [0.15],
            [0.01, 0.15],
        ]
        # Define random effects with
        #   r[X] ~ norm() for a univariate normally distributed r[X], or
        #   r[X] ~ mnorm() for a multivariate normally distributed r[X].
        r[VC_isv,CL_isv] ~ mnorm(
            [0,0],  # mean
            f[VC_isv,CL_isv]  # covariance
        )

PREPROCESS: |

# Use this block to manipulate columns of the input data before
    # processing starts.

# For example, you may create a new column of discrete OCCASION
    # labels from a continuous TIME variable.

POSTPROCESS: |

# Use this block to manipulate columns of the output data after
    # processing has finished.

# For example, you may add a new column of discrete OCCASION
    # labels from a continuous TIME variable after generating data,
    # and avoid the need for the PREPROCESS block when fitting to
    # the generated data.

MODEL_PARAMS: |

# Model parameters that must be positive can be expressed as
    # if from a lognormal distribution, i.e.,
    #   m[X] = f[X] * exp(r[X])
    m[VC]  = f[VC] * exp(r[VC_isv])
    m[CL]  = f[CL] * exp(r[CL_isv])

m[ANOISE] = f[ANOISE]
    m[PNOISE] = f[PNOISE]

STATES: |

# You can initialize states (i.e. amounts) here, for example if
    # the model has a nonzero steady state, e.g.,
    #   s[X] = 1.0

DERIVATIVES: |

# Closed-form models are implemented using names such
    # as "dep_one_cmp_cl" to signify a one compartment model
    # plus a depot compartment, with (V)olume and
    # (CL)earance parameters specified.
    # Intravenous models start with "iv_"
    s[CENTRAL] = @iv_one_cmp_cl{dose: @bolus{amt: c[AMT]}, V: m[VC]}

PREDICTIONS: |

# Define a predicted quantity (e.g. concentration) as a function
    # of the states (i.e. amounts) and model parameters.
    p[DV_CENTRAL] = s[CENTRAL]/m[VC]
    var = m[ANOISE]**2 + (m[PNOISE]*p[DV_CENTRAL])**2

# Assume the observation (a c[X] variable) is a sample from
    # a probability distribution (e.g., normal with mean p[X]
    # and a given variance) from which we can compute the
    # likelihood of a given observation.
    c[DV_CENTRAL] ~ norm(p[DV_CENTRAL], var)

ODE_SOLVER:
    # Use the analytic (closed form) solution.
    ANALYTIC: {}

OUTPUT_SCRIPTS:
    # Plot graphs of the generated observations
    GRPH:
        grph_list:
            - "OBS_vs_TIME"  # individual plots
            - "{OBS}_vs_TIME"  # whole population plot

# Simulate smooth curves and plot
    SIM:
        output_mode: run
        sim_time_step: 1.0
        grph_list:
            - "OBS_vs_TIME"  # individual plots
            - "{OBS}_vs_TIME"  # whole population plot

# Summarize the data generation in an easy to view HTML format.
    # ("output_mode" must be set to "run" for the Online Demonstrator
    # to work.)
    GENSUM:
        output_mode: run

# Rules:
#   Comments are prefixed with the hash (#) character.
#   Sections must start with an indent that is a multiple of 4 spaces.
METHOD_OPTIONS:
    py_module: gen

# Uncomment the following section to add metadata to the model

# DESCRIPTION:
    # name: my_model
    # title: My PK/PD Model
    # author: J.R. Hartley
    # abstract: |

        # A longer description of my model.

    # keywords: ['pk', 'dep_two_cmp', 'first order']

EFFECTS:
    # Pseudocode comes in sections that have a trailing "|"
    # and is processed to become executable Python code.
    # "POP"(ulation) is the top level of the effects tree.
    POP: |

        # Define covariates with
        #   c[X] = k for constants, or
        #   c[X] ~ <dist>() for a randomly sampled variable
        c[AMT] = 100.0

        # Define fixed effects with
        #   f[X] = k for constants, or
        #   f[X] ~ <dist>() for a randomly sampled variable
        f[VC] = 10.0  # Volume of the central compartment
        f[CL] = 2.0  # Clearance

        f[ANOISE] = 0.1  # Additive noise variance
        f[PNOISE] = 0.15  # Proportional noise variance

    # The tree then branches, usually with one branch per individual.
    ID: |

        # Define 20 individuals via a sequence of length 20
        # (i.e., 1, 2, 3, ..., 20)
        # Note that this covariate has the same name (ID) as the
        # level in the tree. This is important for fitting when it
        # partitions the table of data into a tree again, based on
        # a match between the tree level and the data column (e.g. ID).
        c[ID] = sequential(20)

        # Define times of events with t[RESET], t[DOSE] and t[OBS]
        t[RESET] = 0.0
        t[DOSE] = 0.0
        t[OBS] = [1.0, 1.5, 2.0, 4.0, 8.0, 12.0]

        # Define a covariance matrix for multiple variables
        # using the lower triangular entries.
        f[VC_isv,CL_isv] = [
            [0.15],
            [0.01, 0.15],
        ]
        # Define random effects with
        #   r[X] ~ norm() for a univariate normally distributed r[X], or
        #   r[X] ~ mnorm() for a multivariate normally distributed r[X].
        r[VC_isv,CL_isv] ~ mnorm(
            [0,0],  # mean
            f[VC_isv,CL_isv]  # covariance
        )

PREPROCESS: |

    # Use this block to manipulate columns of the input data before
    # processing starts.

    # For example, you may create a new column of discrete OCCASION
    # labels from a continuous TIME variable.

POSTPROCESS: |

    # Use this block to manipulate columns of the output data after
    # processing has finished.

    # For example, you may add a new column of discrete OCCASION
    # labels from a continuous TIME variable after generating data,
    # and avoid the need for the PREPROCESS block when fitting to
    # the generated data.

MODEL_PARAMS: |

    # Model parameters that must be positive can be expressed as
    # if from a lognormal distribution, i.e.,
    #   m[X] = f[X] * exp(r[X])
    m[VC]  = f[VC] * exp(r[VC_isv])
    m[CL]  = f[CL] * exp(r[CL_isv])

    m[ANOISE] = f[ANOISE]
    m[PNOISE] = f[PNOISE]

STATES: |

    # You can initialize states (i.e. amounts) here, for example if
    # the model has a nonzero steady state, e.g.,
    #   s[X] = 1.0

DERIVATIVES: |

    # Closed-form models are implemented using names such
    # as "dep_one_cmp_cl" to signify a one compartment model
    # plus a depot compartment, with (V)olume and
    # (CL)earance parameters specified.
    # Intravenous models start with "iv_"
    s[CENTRAL] = @iv_one_cmp_cl{dose: @bolus{amt: c[AMT]}, V: m[VC]}

PREDICTIONS: |

    # Define a predicted quantity (e.g. concentration) as a function
    # of the states (i.e. amounts) and model parameters.
    p[DV_CENTRAL] = s[CENTRAL]/m[VC]
    var = m[ANOISE]**2 + (m[PNOISE]*p[DV_CENTRAL])**2

    # Assume the observation (a c[X] variable) is a sample from
    # a probability distribution (e.g., normal with mean p[X]
    # and a given variance) from which we can compute the
    # likelihood of a given observation.
    c[DV_CENTRAL] ~ norm(p[DV_CENTRAL], var)

ODE_SOLVER:
    # Use the analytic (closed form) solution.
    ANALYTIC: {}

OUTPUT_SCRIPTS:
    # Plot graphs of the generated observations
    GRPH:
        grph_list:
            - "OBS_vs_TIME"  # individual plots
            - "{OBS}_vs_TIME"  # whole population plot

    # Simulate smooth curves and plot
    SIM:
        output_mode: run
        sim_time_step: 1.0
        grph_list:
            - "OBS_vs_TIME"  # individual plots
            - "{OBS}_vs_TIME"  # whole population plot

    # Summarize the data generation in an easy to view HTML format.
    # ("output_mode" must be set to "run" for the Online Demonstrator
    # to work.)
    GENSUM:
        output_mode: run

PoPy (version 1.2.2) Online Demonstrator

Model Options

Edit Script

Results

Submitted Jobs