Christopher Fanelli回答:
And certainly, both Mike and I have seen a number of our clients start to use AI tools.
当然,Mike和我都已经看到我们的一些客户开始使用AI工具。
I think the one area where AI is much more advanced is on drug discovery, is identifying and selecting candidates for study for further evaluation based on likely outcomes using artificial intelligence tools. And certainly, I've seen both friends and clients successfully use AI applications to target a specific molecule, and for that specific molecule to have success, at least in early phase in a clinic.
我认为AI发展更为先进的领域是药物发现,即利用人工智能工具根据可能的结果识别和筛选用于研究的候选药物,以便进行进一步评估。我确实看到有朋友和客户成功地使用AI应用来靶向特定的分子,并且该特定分子至少在早期临床阶段取得了成功。
And in terms of specific tools, at least in the U.S.,these are primarily proprietary tools.They are applications that companies are developing themselves. They'll hire a team of computer scientists and of mathematicians and folks who are experts in computer programming,and they'll design their own AI applications that target a specific purpose, right, a specific use case. And so these are not off-the-shelf applications. They are proprietary. And so in terms of specific applications,I would say most folks in the U.S. that I've seen, at least, are using purpose-built for a specific use case AI applications.
就具体的工具而言,至少在美国,这些主要是专有工具。它们是企业自行开发的应用程序。企业会聘请一支由计算机科学家、数学家和计算机编程专家组成的团队,设计针对特定的目标、特定使用场景的AI应用程序,所以这些不是现成的应用程序。它们是专有的。
因此,就具体的应用程序而言,我想说,至少我所见到的美国大多数企业,都在使用针对特定使用场景定制的AI应用程序。
And interestingly, one of the things that is more recent, I think the use or potential for AI in the drug discovery context, has been kind of been developing for a number of years now. But one thing that has come up more recently,is the use of AI tools in the GMP setting, for some limited amount of GMP activities, whether it's investigation writing, whether it's trending, APQRs, these are specific instances where clients have already or are in the process of rolling out AI applications
有趣的是,最近出现的一件事是,我认为AI在药物发现领域的应用或潜力,已经发展了好几年了。但最近出现的一件事是在GMP环境中使用AI工具,用于一些有限的GMP活动,无论是调查写作,还是趋势分析、年度产品质量回顾(APQR),这些都是客户已经部署或正在部署AI应用程序的具体实例,
to try to minimize the amount of, you know, the need for kind of direct human involvement, in some of these kind of more basic GMP activities.
目的是尽量减少某些较为基础的GMP活动中人工直接参与。
so like investigation writing would be one of them that I've seen. Trending, so, you know, event trending is another one that I've seen. APQRs, which are, you know, kind of big data analyses of how a product has performed over the course of the year. Those are specific applications where I've seen clients implement AI applications for GMP functions.
比如调查写作是我见过的一个应用。趋势分析,比如事件趋势分析,是我见过的另一个应用。APQR,也就是对产品在一年中的表现进行的大数据分析。这些是我见过客户为GMP功能实施AI应用程序的具体应用实例。
the part of the reason why I think AI in the GMP setting has lagged is,there's a required explainability for, you know, utilizing an AI tool in a GMP setting,where you have to understand how it achieved the result that it achieved.
我认为AI在GMP环境中发展滞后的部分原因是,在GMP环境中使用AI工具需要具备一定的可解释性,你必须理解它是如何得出它所得到的结果的。
And it takes a sophisticated understanding of an AI application, to be able to kind of break out or, kind of separate out what the justification for the decision is, because these have to be scientific justifications in a GMP setting. And so, developing a tool that's capable of that kind of analysis is extremely time consuming, and demonstrating to a regulator that it is as good or better as a human in that role is also extremely challenging.
这需要对AI应用有深入的理解,才能够拆解出或分离出决策的依据,因为在GMP环境中,这些依据必须是科学的依据。因此,开发一种能够进行此类分析的工具非常耗时,并且向监管机构证明该工具能够胜任此类工作,甚至比人类做得更好,也极具挑战性。
Michael Varrone回答:
Chris, I think that last point is important,and one that I was going to mention is,when developing these tools, in particular in the GMP space, being able to understand it,and be able to explain it to the regulator and to FDA, is going to be very important. because FDA will want to know whether it's appropriate in the particular application, in the application in which it's used. And they may not necessarily understand it, and they probably won't understand it.
Chris,我认为最后一点很重要,也是我本来想提到的,在开发这些工具时,特别是在GMP领域,能够理解这个工具,并能够向监管机构和FDA解释这个工具,将会非常重要。
因为FDA会想知道这个工具是否适用于特定的应用场景,即它所使用的应用场景。而FDA未必理解这个工具,也可能不会理解。
So it's going to be very important that the company understands it,and can explain it in clear terms,so that the agency understands, and can see that it's appropriate for the situation in which it's employed.
所以非常重要的是公司必须理解这个工具,并能用清晰的语言解释清楚,以便FDA能够理
并且能够看到它适用于其应用场景。
这{{threadTextType}}正{{isAdminText}}
为帮助审核人员更快处理,请填写举报原因:
为帮助审核人员更快处理,请填写举报原因: