Utilisation of AI to develop Personalised Treatment Plans for cancer

利用人工智能制定个性化癌症治疗计划

• 批准号: 106152
• 金额: $ 42.12万
• 依托单位: MY PERSONAL THERAPEUTICS LTD
• 依托单位国家: 英国
• 项目类别: Collaborative R&D
• 财政年份: 2020
• 资助国家: 英国
• 起止时间: 2020 至 无数据
• 项目状态: 已结题
• 来源: https://gtr.ukri.org/projects?ref=106152
• 关键词: Utilisation; AI; develop; Personalised; Treatment

AI can improve outcomes in cancer. Oncologists can easily identify all mutations that may be driving cancer, but struggle to identify the best mixture of drugs targeted against these mutations. Only AI systems can realistically make sense of the thousands of potentially oncogenic mutations and drugs that could target them. However, data to feed such AI systems is lacking because databases are full of years of data from failed one-mutation, one-drug experiments. My Personal Therapeutics (MPT) use fruit flies engineered to carry human oncogenic mutations in a rapid, high-throughput screen of over 1,000 different cancer and non-cancer drugs. This identifies drug combinations that rescue mutant flies from lethality, which is proving to be an excellent indicator of drug combinations working in humans. They partner with Pentavere, who will apply their DARWEN™ engine to extract useful information from unstructured data held in Genomics England’s database. MPT will perform 25 screens for lung cancer patients, using data to reduce the size of further screens and populate an AI-powered tool called TuMatch, which identifies drug combinations without needing a screen in fruit flies.

人工智能可以改善癌症的治疗效果。肿瘤学家可以很容易地识别出所有可能导致癌症的突变，但很难确定针对这些突变的最佳药物组合。只有人工智能系统才能真正理解数千种潜在的致癌突变和针对它们的药物。然而，这种人工智能系统缺乏数据，因为数据库中充满了多年来失败的一种突变、一种药物实验的数据。My Personal Therapeutics （MPT）使用经过基因改造的果蝇携带人类致癌突变，对1000多种不同的癌症和非癌症药物进行快速、高通量筛选。这一发现确定了拯救突变果蝇免于死亡的药物组合，这被证明是药物组合在人类身上起作用的一个很好的指标。他们与Pentavere合作，Pentavere将使用他们的DARWEN™引擎从Genomics England数据库中的非结构化数据中提取有用的信息。MPT将对肺癌患者进行25次筛查，利用数据减少进一步筛查的规模，并填充一个名为TuMatch的人工智能工具，该工具可以识别药物组合，而无需在果蝇中进行筛查。