Fitting error of a tumor growth model

Aleem · March 4, 2025, 4:09pm

Hi I am tryting to recaptulate a tumour model of a published study as a practice. But I encounted

ERROR: MethodError: no method matching fit(::PumasModel{…}, ::Vector{…}, ::FOCE{…})

using Pumas
using CSV
using DataFrames
using DataFramesMeta

# Define the model
model = @model begin
    @param begin
        TRF ∈ RealDomain(lower=-Inf, upper=Inf, init = -2.0)
        TG ∈ RealDomain(lower=-Inf, upper=Inf, init = -7.0)
        TD ∈ RealDomain(lower=-Inf, upper=Inf, init = -4.0)
        add_err ∈ RealDomain(lower=0, upper=10, init = 0.1)
        prop_err ∈ RealDomain(lower=0, upper=10, init = 0.1)
        Ω ∈ PDiagDomain(init=[0.1])
    end

    @random begin
        η_RF ~ MvNormal(Ω)
    end

    @pre begin
        RF = exp(TRF + η_RF) / (1 + exp(TRF + η_RF))
        G = exp(TG)
        D = exp(TD)
        CL = 92.2
        Q = 125
        V1 = 17.5
        V2 = 47.4
        Vmax = 2
        Km = 0.0075
        k30 = log(2) / 12
        k10 = CL / V1
        k12 = Q / V1
        k21 = Q / V2
    end

    @init begin
        C = 0
        P = 0
        Cin = 0
        Lsen = ILS - ILS * RF  # Use ILS from the dataset
    end

    @dynamics begin
        C' = -(k12 + k10) * C + k21 * P
        P' = k12 * C - k21 * P
        Cin' = Vmax * (C / V1) / (Km + C / V1) - k30 * Cin
        Lsen' = G * Lsen - D * Lsen * Cin
    end

    @derived begin
        LS = @. Lsen + ILS * RF * exp(G * time)  # Compute LS (lesion size)
        dv ~ @. Normal(LS, sqrt(add_err^2 + (prop_err * LS)^2))
    end
end

# Create the population object
pop = read_pumas(Dat;
                 id = :ID,
                 time = :TIME,
                 observations = [:DV],
                 covariates = [:ILS],  # Include ILS (initial lesion size) as a covariate
                 amt = :AMT,
                 evid = :EVID,
                 cmt = :CMT,
                 rate = :RATE
)

# Fit the model
fit1 = fit(model, pop, FOCE())

ERROR: MethodError: no method matching fit(::PumasModel{…}, ::Vector{…}, ::FOCE{…})

Thanks for your help.

vijay · March 4, 2025, 4:31pm

This needs to be

fit1 = fit(model, pop, init_params(model), FOCE())

you forgot to pass in the initial parameters of the model

Aleem · March 4, 2025, 7:06pm

Hi, it failed with the same error after pass in the inits directly. I have defined the initial values in the model structure before.

vijay · March 4, 2025, 7:53pm

Can you please paste the entire error message here?

Aleem · March 4, 2025, 8:30pm

Sure.

ERROR: MethodError: no method matching fit(::PumasModel{…}, ::@NamedTuple{…}, ::Vector{…}, ::FOCE{…})

Closest candidates are:
  fit(::Type{E}, ::Any...) where E<:Survival.NonparametricEstimator
   @ Survival /Users/build/self-hosted-runners/macos_intel/on-the-machine/runner/_work/PumasSystemImages/PumasSystemImages/julia_depot/packages/Survival/ciOeA/src/estimator.jl:25
  fit(::Type{D}, ::Any...) where D<:Distribution
   @ Distributions /Users/build/self-hosted-runners/macos_intel/on-the-machine/runner/_work/PumasSystemImages/PumasSystemImages/julia_depot/packages/Distributions/uuqsE/src/genericfit.jl:47
  fit(::Type{StatsBase.Histogram}, ::Any...; kwargs...)
   @ StatsBase /Users/build/self-hosted-runners/macos_intel/on-the-machine/runner/_work/PumasSystemImages/PumasSystemImages/julia_depot/packages/StatsBase/ebrT3/src/hist.jl:413
  ...

Some type information was truncated. Use `show(err)` to see complete types.

andreasnoack · March 4, 2025, 8:35pm

It looks like you might have called fit(model, init_params(model), pop, FOCE()) instead of fit(model, pop, init_params(model), FOCE()). Try to reverse the two middle arguments.

Aleem · March 4, 2025, 9:29pm

Hi,

After changing the order , it became

ERROR: PumasDataError: Subjects with ids 1, 2, ...., 367 have no modeled observations

Then I realized in the model structure dv should be DV to match my read_pumas function.

After this change, the error is

PumasModel
  Parameters: TRF, TG, TD, add_err, prop_err, Ω
  Random effects: η_RF
  Covariates:
  Dynamical system variables: C, P, Cin, Lsen
  Dynamical system type: Nonlinear ODE
  Derived: LS, DV
  Observed: LS, DV

Population
  Subjects: 367
  Covariates: ILS
  Observations: DV

[ Info: Checking the initial parameter values.
ERROR: MethodError: no method matching +(::Float64, ::Vector{ForwardDiff.Dual{ForwardDiff.Tag{Pumas.Tag, Float64}, Float64, 1}})
For element-wise addition, use broadcasting with dot syntax: scalar .+ array

Closest candidates are:
  +(::Any, ::Any, ::Any, ::Any...)
   @ Base operators.jl:587
  +(::MutableArithmetics.Zero, ::Any)
   @ MutableArithmetics /Users/build/self-hosted-runners/macos_intel/on-the-machine/runner/_work/PumasSystemImages/PumasSystemImages/julia_depot/packages/MutableArithmetics/6jxVC/src/rewrite.jl:64
  +(::Any, ::ChainRulesCore.NoTangent)
   @ ChainRulesCore /Users/build/self-hosted-runners/macos_intel/on-the-machine/runner/_work/PumasSystemImages/PumasSystemImages/julia_depot/packages/ChainRulesCore/6Pucz/src/tangent_arithmetic.jl:60
  ...

Stacktrace:
  [1] (::var"#88#95")(_param#529::@NamedTuple{…}, _random#530::@NamedTuple{…}, _subject#528::Subject{…})
    @ Main /Users/build/self-hosted-runners/macos_intel/on-the-machine/runner/_work/PumasSystemImages/PumasSystemImages/julia_depot/packages/Pumas/GJfPM/src/dsl/model_macro.jl:864
  [2] #_derived#328
    @ /Users/build/self-hosted-runners/macos_intel/on-the-machine/runner/_work/PumasSystemImages/PumasSystemImages/julia_depot/packages/Pumas/GJfPM/src/models/model_api.jl:1214 [inlined]
  [3] _derived
    @ /Users/build/self-hosted-runners/macos_intel/on-the-machine/runner/_work/PumasSystemImages/PumasSystemImages/julia_depot/packages/Pumas/GJfPM/src/models/model_api.jl:1192 [inlined]
  [4] (::Pumas.var"#569#570"{…})(vηorth::Vector{…})
    @ Pumas /Users/build/self-hosted-runners/macos_intel/on-the-machine/runner/_work/PumasSystemImages/PumasSystemImages/julia_depot/packages/Pumas/GJfPM/src/estimation/likelihoods.jl:2873
  [5] chunk_mode_jacobian(f::Pumas.var"#569#570"{…}, x::Vector{…}, cfg::ForwardDiff.JacobianConfig{…}, ::Val{…})
    @ Pumas /Users/build/self-hosted-runners/macos_intel/on-the-machine/runner/_work/PumasSystemImages/PumasSystemImages/julia_depot/packages/Pumas/GJfPM/src/estimation/likelihoods.jl:2806
  [6] _derived_vηorth_gradient(m::PumasModel{…}, subject::Subject{…}, param::@NamedTuple{…}, vrandeffsorth::Vector{…}, diffeq_options::@NamedTuple{})
    @ Pumas /Users/build/self-hosted-runners/macos_intel/on-the-machine/runner/_work/PumasSystemImages/PumasSystemImages/julia_depot/packages/Pumas/GJfPM/src/estimation/likelihoods.jl:2877
  [7] _∂²l∂η²(m::PumasModel{…}, subject::Subject{…}, param::@NamedTuple{…}, vrandeffsorth::Vector{…}, approx::FOCE{…}, diffeq_options::@NamedTuple{})
    @ Pumas /Users/build/self-hosted-runners/macos_intel/on-the-machine/runner/_work/PumasSystemImages/PumasSystemImages/julia_depot/packages/Pumas/GJfPM/src/estimation/likelihoods.jl:2922
  [8] (::Pumas.var"#489#490"{…})(F::Float64, G::Vector{…}, H::Matrix{…}, x::Vector{…})
    @ Pumas /Users/build/self-hosted-runners/macos_intel/on-the-machine/runner/_work/PumasSystemImages/PumasSystemImages/julia_depot/packages/Pumas/GJfPM/src/estimation/likelihoods.jl:1204
  [9] (::NLSolversBase.var"#98#104"{…})(G::Vector{…}, H::Matrix{…}, x::Vector{…})
    @ NLSolversBase /Users/build/self-hosted-runners/macos_intel/on-the-machine/runner/_work/PumasSystemImages/PumasSystemImages/julia_depot/packages/NLSolversBase/kavn7/src/objective_types/incomplete.jl:115
 [10] value_gradient_hessian!!(obj::NLSolversBase.TwiceDifferentiable{…}, x::Vector{…})
    @ NLSolversBase /Users/build/self-hosted-runners/macos_intel/on-the-machine/runner/_work/PumasSystemImages/PumasSystemImages/julia_depot/packages/NLSolversBase/kavn7/src/objective_types/incomplete.jl:138
 [11] initial_state(method::Optim.NewtonTrustRegion{…}, options::Optim.Options{…}, d::NLSolversBase.TwiceDifferentiable{…}, initial_x::Vector{…})
    @ Optim /Users/build/self-hosted-runners/macos_intel/on-the-machine/runner/_work/PumasSystemImages/PumasSystemImages/julia_depot/packages/Optim/ZhuZN/src/multivariate/solvers/second_order/newton_trust_region.jl:310
 [12] optimize
    @ /Users/build/self-hosted-runners/macos_intel/on-the-machine/runner/_work/PumasSystemImages/PumasSystemImages/julia_depot/packages/Optim/ZhuZN/src/multivariate/optimize/optimize.jl:36 [inlined]
 [13] _orth_empirical_bayes!(vrandeffsorth::Vector{…}, m::PumasModel{…}, _subject::Subject{…}, param::@NamedTuple{…}, approx::FOCE{…}, diffeq_options::@NamedTuple{})
    @ Pumas /Users/build/self-hosted-runners/macos_intel/on-the-machine/runner/_work/PumasSystemImages/PumasSystemImages/julia_depot/packages/Pumas/GJfPM/src/estimation/likelihoods.jl:1231
 [14] _update_ebes_and_evaluate_marginal_nll!(m::PumasModel{…}, subject::Subject{…}, param::@NamedTuple{…}, vrandeffsorth::Vector{…}, approx::FOCE{…}, diffeq_options::@NamedTuple{})
    @ Pumas /Users/build/self-hosted-runners/macos_intel/on-the-machine/runner/_work/PumasSystemImages/PumasSystemImages/julia_depot/packages/Pumas/GJfPM/src/estimation/likelihoods.jl:3357
 [15] #590
    @ /Users/build/self-hosted-runners/macos_intel/on-the-machine/runner/_work/PumasSystemImages/PumasSystemImages/julia_depot/packages/Pumas/GJfPM/src/estimation/likelihoods.jl:3515 [inlined]
 [16] next
    @ /Users/build/self-hosted-runners/macos_intel/on-the-machine/runner/_work/PumasSystemImages/PumasSystemImages/julia_depot/packages/Transducers/IAWgA/src/library.jl:54 [inlined]
 [17] next
    @ /Users/build/self-hosted-runners/macos_intel/on-the-machine/runner/_work/PumasSystemImages/PumasSystemImages/julia_depot/packages/Transducers/IAWgA/src/core.jl:777 [inlined]
 [18] next
    @ /Users/build/self-hosted-runners/macos_intel/on-the-machine/runner/_work/PumasSystemImages/PumasSystemImages/julia_depot/packages/Transducers/IAWgA/src/simd.jl:8 [inlined]
 [19] macro expansion
    @ /Users/build/self-hosted-runners/macos_intel/on-the-machine/runner/_work/PumasSystemImages/PumasSystemImages/julia_depot/packages/Transducers/IAWgA/src/core.jl:181 [inlined]
 [20] __foldl__
    @ /Users/build/self-hosted-runners/macos_intel/on-the-machine/runner/_work/PumasSystemImages/PumasSystemImages/julia_depot/packages/Transducers/IAWgA/src/processes.jl:237 [inlined]
 [21] foldl_basecase
    @ /Users/build/self-hosted-runners/macos_intel/on-the-machine/runner/_work/PumasSystemImages/PumasSystemImages/julia_depot/packages/Transducers/IAWgA/src/processes.jl:361 [inlined]
 [22] _reduce_basecase(rf::Transducers.Reduction{…}, init::InitialValues.InitialValueOf{…}, reducible::Transducers.SizedReducible{…})
    @ Transducers /Users/build/self-hosted-runners/macos_intel/on-the-machine/runner/_work/PumasSystemImages/PumasSystemImages/julia_depot/packages/Transducers/IAWgA/src/threading_utils.jl:58
 [23] _reduce(ctx::Transducers.NoopDACContext, rf::Transducers.Reduction{…}, init::InitialValues.InitialValueOf{…}, reducible::Transducers.SizedReducible{…})
    @ Transducers /Users/build/self-hosted-runners/macos_intel/on-the-machine/runner/_work/PumasSystemImages/PumasSystemImages/julia_depot/packages/Transducers/IAWgA/src/reduce.jl:139
 [24] _reduce(ctx::Transducers.NoopDACContext, rf::Transducers.Reduction{…}, init::InitialValues.InitialValueOf{…}, reducible::Transducers.SizedReducible{…}) (repeats 3 times)
    @ Transducers /Users/build/self-hosted-runners/macos_intel/on-the-machine/runner/_work/PumasSystemImages/PumasSystemImages/julia_depot/packages/Transducers/IAWgA/src/reduce.jl:148
 [25] _reduce(ctx::Transducers.NoopDACContext, rf::Transducers.Reduction{…}, init::InitialValues.InitialValueOf{…}, reducible::Transducers.SizedReducible{…})
    @ Transducers /Users/build/self-hosted-runners/macos_intel/on-the-machine/runner/_work/PumasSystemImages/PumasSystemImages/julia_depot/packages/Transducers/IAWgA/src/reduce.jl:148
 [26] _transduce_assoc_nocomplete
    @ /Users/build/self-hosted-runners/macos_intel/on-the-machine/runner/_work/PumasSystemImages/PumasSystemImages/julia_depot/packages/Transducers/IAWgA/src/reduce.jl:131 [inlined]
 [27] transduce_assoc(xform::Transducers.Map{…}, step::Transducers.Completing{…}, init::InitialValues.InitialValueOf{…}, coll0::Base.Iterators.Zip{…}; simd::Val{…}, basesize::Int64, stoppable::Nothing, nestlevel::Nothing)
    @ Transducers /Users/build/self-hosted-runners/macos_intel/on-the-machine/runner/_work/PumasSystemImages/PumasSystemImages/julia_depot/packages/Transducers/IAWgA/src/reduce.jl:108
 [28] transduce_assoc
    @ /Users/build/self-hosted-runners/macos_intel/on-the-machine/runner/_work/PumasSystemImages/PumasSystemImages/julia_depot/packages/Transducers/IAWgA/src/reduce.jl:84 [inlined]
 [29] #foldxt#184
    @ /Users/build/self-hosted-runners/macos_intel/on-the-machine/runner/_work/PumasSystemImages/PumasSystemImages/julia_depot/packages/Transducers/IAWgA/src/reduce.jl:235 [inlined]
 [30] mapreduce(f::Function, op::Function, itr::Base.Iterators.Zip{Tuple{…}}; kw::@Kwargs{simd::Val{…}})
    @ ThreadsX.Implementations /Users/build/self-hosted-runners/macos_intel/on-the-machine/runner/_work/PumasSystemImages/PumasSystemImages/julia_depot/packages/ThreadsX/Bml38/src/reduce.jl:19
 [31] mapreduce
    @ /Users/build/self-hosted-runners/macos_intel/on-the-machine/runner/_work/PumasSystemImages/PumasSystemImages/julia_depot/packages/ThreadsX/Bml38/src/reduce.jl:17 [inlined]
 [32] sum
    @ /Users/build/self-hosted-runners/macos_intel/on-the-machine/runner/_work/PumasSystemImages/PumasSystemImages/julia_depot/packages/ThreadsX/Bml38/src/reduce.jl:48 [inlined]
 [33] _update_ebes_and_evaluate_marginal_nll_gradient_threads!(g::Vector{…}, m::PumasModel{…}, population::Vector{…}, param::@NamedTuple{…}, vparam::Vector{…}, vvrandeffs::Vector{…}, vvrandeffs_tmp::Vector{…}, approx::FOCE{…}, trf::TransformVariables.TransformTuple{…}, diffeq_options::@NamedTuple{})
    @ Pumas /Users/build/self-hosted-runners/macos_intel/on-the-machine/runner/_work/PumasSystemImages/PumasSystemImages/julia_depot/packages/Pumas/GJfPM/src/estimation/likelihoods.jl:3510
 [34] (::Pumas.var"#611#615"{…})(f::Float64, g::Vector{…}, _vparam::Vector{…})
    @ Pumas /Users/build/self-hosted-runners/macos_intel/on-the-machine/runner/_work/PumasSystemImages/PumasSystemImages/julia_depot/packages/Pumas/GJfPM/src/estimation/likelihoods.jl:4556
 [35] _optim_check_initial_values(costf::Pumas.var"#611#615"{…}, vparam::Vector{…}, verbose::Bool)
    @ Pumas /Users/build/self-hosted-runners/macos_intel/on-the-machine/runner/_work/PumasSystemImages/PumasSystemImages/julia_depot/packages/Pumas/GJfPM/src/estimation/likelihoods.jl:4449
 [36] _fit(m::PumasModel{…}, population::Vector{…}, param::@NamedTuple{…}, approx::FOCE{…}, ensemblealg::EnsembleThreads, optimize_fn::Pumas.DefaultOptimizeFN{…}, fixedparamset::ParamSet{…}, fixedparam::@NamedTuple{…}, checkidentification::Bool, diffeq_options::@NamedTuple{}, init_vrandeffsorth::Vector{…}, verbose::Bool, ignore_numerical_error::Bool, optim_state::Nothing)
    @ Pumas /Users/build/self-hosted-runners/macos_intel/on-the-machine/runner/_work/PumasSystemImages/PumasSystemImages/julia_depot/packages/Pumas/GJfPM/src/estimation/likelihoods.jl:4622
 [37] __fit
    @ /Users/build/self-hosted-runners/macos_intel/on-the-machine/runner/_work/PumasSystemImages/PumasSystemImages/julia_depot/packages/Pumas/GJfPM/src/estimation/likelihoods.jl:4331 [inlined]
 [38] __fit
    @ /Users/build/self-hosted-runners/macos_intel/on-the-machine/runner/_work/PumasSystemImages/PumasSystemImages/julia_depot/packages/Pumas/GJfPM/src/estimation/likelihoods.jl:4318 [inlined]
 [39] fit(m::PumasModel{…}, population::Vector{…}, param::@NamedTuple{…}, approx::FOCE{…}; optim_alg::Nothing, optim_options::Nothing, optimize_fn::Nothing, constantcoef::Tuple{}, omegas::Tuple{}, ensemblealg::EnsembleThreads, checkidentification::Bool, diffeq_options::@NamedTuple{}, init_randeffs::Nothing, init_vrandeffsorth::Nothing, verbose::Bool, ignore_numerical_error::Bool)
    @ Pumas /Users/build/self-hosted-runners/macos_intel/on-the-machine/runner/_work/PumasSystemImages/PumasSystemImages/julia_depot/packages/Pumas/GJfPM/src/estimation/likelihoods.jl:4285
 [40] fit(m::PumasModel{…}, population::Vector{…}, param::@NamedTuple{…}, approx::FOCE{…})
    @ Pumas /Users/build/self-hosted-runners/macos_intel/on-the-machine/runner/_work/PumasSystemImages/PumasSystemImages/julia_depot/packages/Pumas/GJfPM/src/estimation/likelihoods.jl:4231
Some type information was truncated. Use `show(err)` to see complete types.

Best regards

patrick · March 6, 2025, 12:21pm

This is because you are summing TRF and η_RF which is a number and a vector respectively. Either write η_RF[1] or rewrite η_RF to be a scalar distribution (Normal instead of MvNormal).

Aleem · March 6, 2025, 2:12pm

Hi Patrick,

Thanks for your help!

Then the error became

ERROR: UndefVarError: `ILS` not defined

I tried define it in the @coviariates, but it did not work.

The model structure now is ,


# Define the model
model = @model begin
    @param begin
        TRF ∈ RealDomain(lower=-Inf, upper=Inf, init = -2.0)
        TG ∈ RealDomain(lower=-Inf, upper=Inf, init = -7.0)
        TD ∈ RealDomain(lower=-Inf, upper=Inf, init = -4.0)
        add_err ∈ RealDomain(lower=0, upper=10, init = 0.1)
        prop_err ∈ RealDomain(lower=0, upper=10, init = 0.1)
    end

    @random begin
        η_RF ~ Normal(0, 0.1)
    end

    @covariates ILS

    @pre begin
        RF = exp(TRF + η_RF) / (1 + exp(TRF + η_RF))
        G = exp(TG)
        D = exp(TD)
        CL = 92.2
        Q = 125
        V1 = 17.5
        V2 = 47.4
        Vmax = 2
        Km = 0.0075
        k30 = log(2) / 12
        k10 = CL / V1
        k12 = Q / V1
        k21 = Q / V2
    end

    @init begin
        C = 0
        P = 0
        Cin = 0
        Lsen = ILS - ILS * RF  # Use ILS from the dataset
    end

    @dynamics begin
        C' = -(k12 + k10) * C + k21 * P
        P' = k12 * C - k21 * P
        Cin' = Vmax * (C / V1) / (Km + C / V1) - k30 * Cin
        Lsen' = G * Lsen - D * Lsen * Cin
    end

    @derived begin
        LS = @. Lsen + ILS * RF * exp(G * time)  # Compute LS (lesion size)
        DV ~ @. Normal(LS, sqrt(add_err^2 + (prop_err * LS)^2))
    end
end


# Create the population object
pop = read_pumas(Dat;
                 id = :ID,
                 time = :TIME,
                 observations = [:DV],
                 covariates = [:ILS],  # Include ILS (initial lesion size) as a covariate
                 amt = :AMT,
                 evid = :EVID,
                 cmt = :CMT,
                 rate = :RATE
)

# Fit the model
fit1 = fit(model, pop, init_params(model), FOCE())

Best wishes

patrick · March 6, 2025, 7:43pm

# Define the model
model = @model begin
    @param begin
        TRF ∈ RealDomain(lower=-Inf, upper=Inf, init = -2.0)
        TG ∈ RealDomain(lower=-Inf, upper=Inf, init = -7.0)
        TD ∈ RealDomain(lower=-Inf, upper=Inf, init = -4.0)
        add_err ∈ RealDomain(lower=0, upper=10, init = 0.1)
        prop_err ∈ RealDomain(lower=0, upper=10, init = 0.1)
    end

    @random begin
        η_RF ~ Normal(0, 0.1)
    end

    @covariates ILS

    @pre begin
        RF = exp(TRF + η_RF) / (1 + exp(TRF + η_RF))
        G = exp(TG)
        D = exp(TD)
        CL = 92.2
        Q = 125
        V1 = 17.5
        V2 = 47.4
        Vmax = 2
        Km = 0.0075
        k30 = log(2) / 12
        k10 = CL / V1
        k12 = Q / V1
        k21 = Q / V2
        ils = ILS
    end

    @init begin
        C = 0
        P = 0
        Cin = 0
        Lsen = ils*(1 - RF)  # Use ILS from the dataset
    end

    @dynamics begin
        C' = -(k12 + k10) * C + k21 * P
        P' = k12 * C - k21 * P
        Cin' = Vmax * (C / V1) / (Km + C / V1) - k30 * Cin
        Lsen' = G * Lsen - D * Lsen * Cin
    end

    @derived begin
        LS = @. Lsen + ILS * RF * exp(G * time)  # Compute LS (lesion size)
        DV ~ @. Normal(LS, sqrt(add_err^2 + (prop_err * LS)^2))
    end
end

Please try the above where the covariate is defined in @pre with a new name. Covariates are currently not automatically available in @init, but we will track this as a potential improvement to Pumas.

Aleem · March 6, 2025, 9:22pm

Thanks. Then the error became

[ Info: Checking the initial parameter values.
ERROR: MethodError: no method matching *(::Float64, ::typeof(time))

Closest candidates are:
  *(::Any, ::Any, ::Any, ::Any...)
   @ Base operators.jl:587
  *(::Real, ::Symbolics.ComplexTerm)
   @ Symbolics /Users/build/self-hosted-runners/macos_intel/on-the-machine/runner/_work/PumasSystemImages/PumasSystemImages/julia_depot/packages/SymbolicUtils/c0xQb/src/methods.jl:74
  *(::Number, ::Symbolics.ComplexTerm)
   @ Symbolics /Users/build/self-hosted-runners/macos_intel/on-the-machine/runner/_work/PumasSystemImages/PumasSystemImages/julia_depot/packages/SymbolicUtils/c0xQb/src/methods.jl:76
  ...

My code now is:

# Define the model
model = @model begin
    @param begin
        TRF ∈ RealDomain(lower=-10, upper=10, init = -2.0)
        TG ∈ RealDomain(lower=-10, upper=10, init = -7.0)
        TD ∈ RealDomain(lower=-10, upper=10, init = -4.0)
        add_err ∈ RealDomain(lower=0, upper=10, init = 0.1)
        prop_err ∈ RealDomain(lower=0, upper=10, init = 0.1)
    end

    @random begin
        η_RF ~ Normal(0, 0.1)
    end

    @covariates ILS

    @pre begin
        RF = exp(TRF + η_RF) / (1 + exp(TRF + η_RF))
        G = exp(TG)
        D = exp(TD)
        CL = 92.2
        Q = 125
        V1 = 17.5
        V2 = 47.4
        Vmax = 2
        Km = 0.0075
        k30 = log(2) / 12
        k10 = CL / V1
        k12 = Q / V1
        k21 = Q / V2
        ils = ILS
    end

    @init begin
        C = 0
        P = 0
        Cin = 0
        Lsen = ils * (1 - RF)  # Use ILS from the dataset
    end

    @dynamics begin
        C' = -(k12 + k10) * C + k21 * P
        P' = k12 * C - k21 * P
        Cin' = Vmax * (C / V1) / (Km + C / V1) - k30 * Cin
        Lsen' = G * Lsen - D * Lsen * Cin
    end

    @derived begin
        LS = @. Lsen + ils * RF * exp(G * time)  # Compute LS (lesion size)
        DV ~ @. Normal(LS, sqrt(add_err^2 + (prop_err * LS)^2))
    end
end

# Create the population object
pop = read_pumas(pancreasDat;
                 id = :ID,
                 time = :TIME,
                 observations = [:DV],
                 covariates = [:ILS],  # Include ILS (initial lesion size) as a covariate
                 amt = :AMT,
                 evid = :EVID,
                 cmt = :CMT,
                 rate = :RATE
)

# Fit the model
fit1 = fit(model, pop, init_params(model), FOCE())

I changed all the ILS to ils here. I changed time to TIME but still got error, so then I changed it back. it seems that it cant detect time and then mutiply still?

Best wishes

patrick · March 6, 2025, 9:48pm

You are looking for the reserved keyword t I believe. Please find the information on how to use time in the documentation of @pre ! Defining NLME models in Pumas · Pumas

patrick · March 7, 2025, 5:41am

Actually, as we went through the process internally, we came to the conclusion that this already works on the latest version of Pumas. Can you verify what version of Pumas you are using?

Aleem · March 7, 2025, 2:31pm

Thanks a lot! It worked and fast enough

Topic		Replies	Views
Fitting Error with Pumas 2.1 Estimation	13	271	May 17, 2022
Error with fitting using simulated data Estimation	2	276	March 15, 2022
Error : when i try to use fit Estimation	16	967	October 27, 2020
Error with Pumas v 2.2 maximum-likelihood	4	272	November 18, 2022
Error with M3 (Censored method) maximum-likelihood	2	323	June 29, 2021

Fitting error of a tumor growth model

Related topics