The DEDUCTION Project

The DEDUCTION Project is the latest—and possibly conclusive—development within the DTAT research programme. This post explains its origins, motivation and current outlook.
Published

November 24, 2025

Modified

November 25, 2025

Among the numerous aspects of cancer therapy that are subject to individualization, the Dose Titration Algorithm Tuning (DTAT) programme has focused chiefly on dose individualization. From the outset, DTAT’s premise has been that the most effective way to promote dose individualization in oncology clinical practice is to introduce it starting from the earliest stages of oncology drug development — phase 1 trials.

As a methodologic framework, DTAT conceives patient-centered dose optimization as a continuous learning process, putting forth recursive filtering as a suitable metaphor and applicable technique (). While this idea remains important to DTAT’s conceptual coherence, recursive filtering (like many other time-series concepts) lies generally outside the repertoire of most biostatisticians involved in phase 1 trial design. Thus, a longstanding open problem within the DTAT programme has been how to render dose-titration protocols accessible to oncology clinical trialists accustomed to inflexible dose-escalation designs with much smaller state-spaces. Whereas early efforts in this direction () involved ad hoc protocol modifications, logic programming — in particular, Prolog — has since emerged as a fruitful setting for the comprehensive analysis and principled extension of dose-escalation protocols (). Recently, this effort has culminated in a generally applicable approach based on the Kan extension of category theory (). From an accessibility perspective, replacing recursive filtering with the Kan extension may seem a step in the wrong direction! But the latter mercifully lends itself to a ‘dashboard’ visualization (under active development) that situates dose titration in a familiar dose-escalation context.

Oncology clinical trialists have recently begun to articulate an interest in the development of “flexible and adaptive” dose-finding designs including allowances for dose titration (). Transforming such expressed interest into widespread, effective demand likely requires availability of robust dose-titration trial design software.

Accordingly, the DEDUCTION project is now turning its attention to the development of a free, open-source software suite for dose-titration trial design and analysis, implemented in the ISO Standard conforming Scryer Prolog. We are actively seeking funding to support this effort.

References

Araujo, D., A. Greystoke, S. Bates, A. Bayle, E. Calvo, L. Castelo-Branco, J. De Bono, et al. 2023. “Oncology Phase I Trial Design and Conduct: Time for a Change - MDICT Guidelines 2022.” Annals of Oncology 34 (1): 48–60. https://doi.org/10.1016/j.annonc.2022.09.158.
Norris, David C. 2017a. “Dose Titration Algorithm Tuning (DTAT) Should Supersede ‘the’ Maximum Tolerated Dose (MTD) in Oncology Dose-Finding Trials.” F1000Res 6 (July): 112. https://doi.org/10.12688/f1000research.10624.3.
———. 2017b. “Precautionary Coherence Unravels Dose Escalation Designs.” bioRxiv, December, 240846. https://doi.org/10.1101/240846.
———. 2025. “Dose-Escalation Trial Protocols That Extend Naturally to Admit Titration.” arXiv. https://doi.org/10.48550/arXiv.2507.01370.
Norris, David C., and Markus Triska. 2024. “An Executable Specification of Oncology Dose-Escalation Protocols with Prolog.” arXiv. https://arxiv.org/abs/2402.08334.

Footnotes

  1. These include its cadence, as considered for example in metronomic therapy, as well as its duration, and indeed even its spatial aspects as considered in radiation oncology.↩︎