Early-Stage Clinical Trials with Patient Choice

本项目探讨设计早期临床试验有效机制的数学方法。早期临床试验的目的是为后续临床试验阶段确定研究药物的安全和治疗剂量。由于事先对研究药物的毒性特征知之甚少，目前的早期试验方法采用剂量递增方法，即连续组患者接受增加剂量的治疗，直到观察到剂量限制性毒性。这种做法往往导致招募和留住患者的重大挑战，因为许多患者认为参与的好处很低。例如，穷尽所有其他选择的绝症患者通常自愿参加早期试验，其中规定的剂量低且无治疗作用，而相对健康的患者可能会推迟参与试验，以等待有关剂量安全性和有效性的更多信息。过于保守的剂量递增方法往往导致大量的试验延误和费用，在患者安全方面收效甚微。本项目研究允许基于个体患者风险偏好的有限患者输入来指导剂量决定对试验安全性和有效性的影响。PIs的目标是通过课程作业和研究支持，让本科生（通过克莱姆森大学的创造性探究学习计划）和研究生参与研究。该项目引入了一类新的具有学习能力的动态剂量选择游戏，将不断变化的患者偏好和对剂量效率和剂量毒性的信念与最佳停止规则结合起来，以确定证据何时汇聚到可接受的剂量，以便进行后续试验。该研究计划采用了一种伪市场机制，通过将集中剂量分配决策权委托给自利主体（患者及其医生），而中央机构评估研究数据的结果，解决了一种新的探索和效率权衡问题。研究如何解决分配效率（当前参与者效用最大化）与探索（总体效用最大化）之间的潜在冲突，实现学习最大化，统一探索剂量范围，避免集中在小剂量范围内。该奖项反映了美国国家科学基金会的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

This project investigates mathematical approaches for designing efficient mechanisms for early-stage clinical trials. Early-stage clinical trials aim to establish safe and therapeutic doses of an investigational drug for subsequent clinical trial phases. Because little is known in advance about the toxicity profile of investigational drugs, current approaches to early-stage trials employ dose-escalation methods, in which consecutive groups of patients receive increasing doses until dose-limiting toxicities are observed. This practice often results in significant challenges in recruiting and retaining patients, as many patients perceive the benefits of participation to be low. For example, terminally ill patients who have exhausted all other options often volunteer for early trials, in which the doses prescribed are low and nontherapeutic, whereas relatively healthy patients may postpone their involvement to wait for more information on dosing safety and efficacy. Overly conservative dose-escalation methods often result in substantial trial delays and expense, with little gain in patient safety. This project studies the effect on trial safety and efficacy of allowing limited patient input based on individual patient risk preferences to guide dosage decisions. The PIs aim to involve both undergraduate students (through Clemson's Creative Inquiry learning program) and graduate students in the research, through both coursework and research support. The project introduces a new class of dynamic dose-selection games with learning that integrate evolving patient preferences and beliefs about dose efficiency and dose toxicity with optimal stopping rules to determine when evidence converges on acceptable dosage to move to later trials. The research plan employs a pseudo-market mechanism that addresses a novel exploration-and-efficiency tradeoff created by delegating the centralized dose allocation decision to self-interested agents (patients and their physicians), while the central authority evaluates the results of study data. Methods are investigated to resolve potential conflicts between allocation efficiency (maximizing current participants’ utility) and exploration (maximizing the general population utility) so as to maximize learning, explore the dose-range uniformly, and avoid concentration within a small dosage range.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.