startdoser

Startdoser is a software framework designed to assist researchers in planning and simulating initial dosing regimens for drug development. It provides a platform to model pharmacokinetics and pharmacodynamics, perform dose-escalation designs, and evaluate safety signals in a controlled, simulated environment before human trials. The aim is to optimize starting doses and escalation steps while reducing participant risk and accelerating study timelines.

Key features include dose-escalation planning, PK/PD modeling, Bayesian adaptive decision rules, and both prospective and retrospective simulation modes. It supports common escalation designs such as traditional 3+3, model-based approaches like continuous reassessment methods, and Bayesian optimal interval strategies. The framework also offers data visualization, audit trails, reproducible workflows, and interoperability with electronic data capture and analysis tools. It is designed to be extensible through plug-ins and custom modeling modules.

Implementation typically involves languages and tools common in pharmacometrics, such as Python and R, with interfaces to PK/PD libraries and Bayesian inference engines. Startdoser accepts standard data formats and provides a scripting interface to customize models, escalation rules, and decision criteria. The project emphasizes transparent reporting, traceable simulations, and versioning of models and datasets.

History and reception: the concept emerged within the pharmacometrics community in the late 2010s as a way to formalize in silico dose-finding. First public releases appeared in the early 2020s and have since seen growing adoption in academic and industry settings for planning phase I studies. Limitations include reliance on accurate models and data quality, as well as varying degrees of regulatory acceptance across regions.

See also: pharmacometrics, dose-escalation study, Bayesian adaptive design, PK/PD modeling.