Regulatory Sciences Community

Dear colleagues -
the FDA published several discussion papers on the use of AI in drug development and manufacturing. The development paper is attached. FDA has a lot of questions for input that span a wide area of the drug development trajectory. I am wondering how we can contribute to the conversation at AAPS. I am going to tag @Mark Arnold for  bioanalytical and @Victoria Demby for clinical/reg thoughts but would love to hear what everybody is thinking about this.
Thanks,
Tina

Tina,
Thanks for bringing up this current topic.  At Regulatory Sciences community we keep this topic ongoing in our monthly meetings. At the end, all the AI/ML efforts results in being part of regulatory submissions.  What is acceptable and what is not and why, is of paramount importance to Regulatory Sciences community.  We will be very interested in actively participating in these discussions.  
-Raja
Chair, RS Community

Dear colleagues -
the FDA published several discussion papers on the use of AI in drug development and manufacturing. The development paper is attached. FDA has a lot of questions for input that span a wide area of the drug development trajectory. I am wondering how we can contribute to the conversation at AAPS. I am going to tag @Mark Arnold for  bioanalytical and @Victoria Demby for clinical/reg thoughts but would love to hear what everybody is thinking about this.
Thanks,
Tina

Dear colleagues -
the FDA published several discussion papers on the use of AI in drug development and manufacturing. The development paper is attached. FDA has a lot of questions for input that span a wide area of the drug development trajectory. I am wondering how we can contribute to the conversation at AAPS. I am going to tag @Mark Arnold for  bioanalytical and @Victoria Demby for clinical/reg thoughts but would love to hear what everybody is thinking about this.
Thanks,
Tina

Tina- 

Great topic.  I had seen the FDA info release and downloaded the one file "Using Artificial Intelligence & Machine Learning in the Development of Drug and Biological Products". I was, in fact, reading it when the notification came in that you had tagged me in this post. I've been following when I can what has been going on with artificial intelligence, machine learning, and natural language processing for a few years now.  I started going down the rabbit hole with natural language processing at a prior company when internet search engines were missing publications and information related to biomarkers of specific disease.  Getting back 10,000 hits is not effective! I wanted a better search to refine things and spoke to a number of experts and management - the experts were willing to do the development but management wouldn't prioritize.  I've kept up with what I can since then. In the past 3 years there has been such an explosion of tools, and many are now available off the shelf. They're designed to take your questions, but I'm still not good at those inquiries.

Last year's PharmSci 360 had a few talks, but I was only able to attend one Rapid Fire that @Zachary Greenberg  from @Mei He 's lab presented that used AI to assess complex data sets to derive relationships for using exosomes to deliver vaccines. At the NBC last month there were three talks that included AI as part of drug development, two of which were related to bioanalysis ( @Stephanie Pasas-Farmer and Mohammed Moin) and the third to metabolomics ( @Timothy Garrett from Univ. of Florida).  At least one of the vendors, BioData Solutions, who presented at AAPS's Land O' Lakes conference in 2018, has continued to advance their software to assess the quality of bioanalytical reports (i.e., finding errors and compliance violations) and is innovating into other bioanalytical applications.  

As someone with a chromatography background, the Holy Grail has always been software to assess chromatograms and set baselines.  Many instrument manufacturers have improved their algorithms, but I've not yet heard that they can handle everything that the real world of drug and biomarkers in bio fluid can throw at them. I'm hoping that someone is developing an AI to evaluate chromatograms -- it is sorely needed to reduce regulator concerns over manually drawing baselines. It was a topic of concern at many of the AAPS-FDA Crystal City meetings over the past 3 decades, not to mention with regulators from around the world.  Other areas where AI in bioanalysis would help is with electronic notebooks and LIMS.  Being able to have an AI look at the entirety of the data related to an assay over multiple studies and identify the causation for when the assay starts to fail could flag situations in advance, prevent failures and thereby reduce cost and speed drug development. 

I'm sure that there are many more AI applications that AAPS members are involved with that I'm hoping people will respond to this post with.