TY - JOUR

T1 - Bayesian inference for stochastic differential equation mixed effects models of a tumour xenography study

AU - Picchini, Umberto

AU - Forman, Julie Lyng

PY - 2019

Y1 - 2019

N2 - We consider Bayesian inference for stochastic differential equation mixed effects models (SDEMEMs) exemplifying tumour response to treatment and regrowth in mice. We produce an extensive study on how an SDEMEM can be fitted by using both exact inference based on pseudo-marginal Markov chain Monte Carlo sampling and approximate inference via Bayesian synthetic likelihood (BSL). We investigate a two-compartments SDEMEM, corresponding to the fractions of tumour cells killed by and survived on a treatment. Case-study data consider a tumour xenography study with two treatment groups and one control, each containing 5–8 mice. Results from the case-study and from simulations indicate that the SDEMEM can reproduce the observed growth patterns and that BSL is a robust tool for inference in SDEMEMs. Finally, we compare the fit of the SDEMEM with a similar ordinary differential equation model. Because of small sample sizes, strong prior information is needed to identify all model parameters in the SDEMEM and it cannot be determined which of the two models is the better in terms of predicting tumour growth curves. In a simulation study we find that with a sample of 17 mice per group BSL can identify all model parameters and distinguish treatment groups.

AB - We consider Bayesian inference for stochastic differential equation mixed effects models (SDEMEMs) exemplifying tumour response to treatment and regrowth in mice. We produce an extensive study on how an SDEMEM can be fitted by using both exact inference based on pseudo-marginal Markov chain Monte Carlo sampling and approximate inference via Bayesian synthetic likelihood (BSL). We investigate a two-compartments SDEMEM, corresponding to the fractions of tumour cells killed by and survived on a treatment. Case-study data consider a tumour xenography study with two treatment groups and one control, each containing 5–8 mice. Results from the case-study and from simulations indicate that the SDEMEM can reproduce the observed growth patterns and that BSL is a robust tool for inference in SDEMEMs. Finally, we compare the fit of the SDEMEM with a similar ordinary differential equation model. Because of small sample sizes, strong prior information is needed to identify all model parameters in the SDEMEM and it cannot be determined which of the two models is the better in terms of predicting tumour growth curves. In a simulation study we find that with a sample of 17 mice per group BSL can identify all model parameters and distinguish treatment groups.

KW - Intractable likelihood

KW - Pseudo-marginal Markov chain Monte Carlo sampling

KW - Repeated measurements

KW - State space model

KW - Synthetic likelihood

U2 - 10.1111/rssc.12347

DO - 10.1111/rssc.12347

M3 - Journal article

AN - SCOPUS:85063353705

VL - 68

SP - 887

EP - 913

JO - Journal of the Royal Statistical Society, Series C (Applied Statistics)

JF - Journal of the Royal Statistical Society, Series C (Applied Statistics)

SN - 0035-9254

IS - 4

ER -