A wall-time minimizing parallelization strategy for approximate Bayesian computation.

Approximate Bayesian Computation (ABC) is a widely applicable and popular approach to estimating unknown parameters of mechanistic models. As ABC analyses are computationally expensive, parallelization on high-performance infrastructure is often necessary. However, the existing parallelization strategies leave computing resources unused at times and thus do not optimally leverage them yet. We present look-ahead scheduling, a wall-time minimizing parallelization strategy for ABC Sequential Monte Carlo algorithms, which avoids idle times of computing units by preemptive sampling of subsequent generations. This allows to utilize all available resources. The strategy can be integrated with e.g. adaptive distance function and summary statistic selection schemes, which is essential in practice. Our key contribution is the theoretical assessment of the strategy of preemptive sampling and the proof of unbiasedness. Complementary, we provide an implementation and evaluate the strategy on different problems and numbers of parallel cores, showing speed-ups of typically 10-20% and up to 50% compared to the best established approach, with some variability. Thus, the proposed strategy allows to improve the cost and run-time efficiency of ABC methods on high-performance infrastructure.

New In Vitro Interaction-Parasite Reduction Ratio Assay for Early Derisk in Clinical Development of Antimalarial Combinations.

The development and spread of drug-resistant phenotypes substantially threaten malaria control efforts. Combination therapies have the potential to minimize the risk of resistance development but require intensive preclinical studies to determine optimal combination and dosing regimens. To support the selection of new combinations, we developed a novel in vitro-in silico combination approach to help identify the pharmacodynamic interactions of the two antimalarial drugs in a combination which can be plugged into a pharmacokinetic/pharmacodynamic model built with human monotherapy parasitological data to predict the parasitological endpoints of the combination. This makes it possible to optimally select drug combinations and doses for the clinical development of antimalarials. With this assay, we successfully predicted the endpoints of two phase 2 clinical trials in patients with the artefenomel-piperaquine and artefenomel-ferroquine drug combinations. In addition, the predictive performance of our novel in vitro model was equivalent to that of the humanized mouse model outcome. Last, our more informative in vitro combination assay provided additional insights into the pharmacodynamic drug interactions compared to the in vivo systems, e.g., a concentration-dependent change in the maximum killing effect (Emax) and the concentration producing 50% of the killing maximum effect (EC50) of piperaquine or artefenomel or a directional reduction of the EC50 of ferroquine by artefenomel and a directional reduction of Emax of ferroquine by artefenomel. Overall, this novel in vitro-in silico-based technology will significantly improve and streamline the economic development of new drug combinations for malaria and potentially also in other therapeutic areas.