Decision-making in clinical drug development.

临床试验中最重要的实验类型是随机临床试验。从合成到测试一种新的治疗方法，制药公司将大量资金置于风险之中，这表明他们渴望优化决策，以实现收入最大化。目前的一个主要问题是，尽管关于疾病的分子基础的知识激增，但很大一部分新的治疗方法仍然失败。更重要的是，第三阶段试验的失败率很高，这是有问题的，也是不直观的，因为通常只有有希望的治疗方法才会进入第三阶段，而且它们是进行成本最高的。显然，像目前这样高的失败率是不可持续的。该项目的目的是在临床试验中使用算法框架和数据驱动的方法进行决策，最终目的是指导肺癌临床试验的决策。我们的重点是在任何情况下(例如，一种罕见疾病的病例)为第二阶段和第三阶段的组合找到最佳决策。这将通过创建易于临床医生使用、可解释并可有效解决的数学模型来实现。反过来，制药公司将利用这一点做出正式决定，从而在最大限度地减少错误的同时实现未来利益的最大化。

The most important type of experiment in clinical trials is the randomised clinical trial. From synthesising to testing a novel treatment pharmaceutical companies put a large amount of money at risk which makes it evident that they are eager to optimise their decisions in order to maximise their revenue. A current major problem is that although there is a surge of knowledge about the molecular basis of diseases, a significant proportion of novel treatments still fail. More importantly, the failure rate of Phase III trials is high and this is problematic and unintuitive as usually only promising treatments move to Phase III and they are the most expensive to conduct. It is evident that failure rates as high as the current ones are not sustainable. The aim of this project is to use algorithmic frameworks and data-driven approaches to decision-making in clinical trials, with an ultimate goal of guiding in decision-making of lung cancer clinical trials. Our focus is on finding the optimal decisions for the combination of Phase II and Phase III in any scenario (e.g., a case of a rare disease). This will be done by creating mathematical models that are easy to use by clinicians, are interpretable and are solved efficiently. This will, in turn, be used by pharmaceutical companies to make formal decisions and thus maximise future benefit while minimising errors.