The DrugBot

RoeyRoey Posts: 160 XPRIZE
edited September 2019 in Health
Could we automate nearly every part of the preliminary drug development and approval process? Do you think we have the technologies in place for that kind of a breakthrough? Let us know what you think!

The DrugBot - Automating Drug Development: A system that will develop and conduct a preliminary in-vitro and in-vivo testing of new drugs for various purposes.

Stipulations: The winning team will come up with an integrated system that will come up with ideas and schematics for new drugs, will rapidly test them in-vitro and in-vivo, and will prepare them for the more rigorous testing conducted in the course of official clinical trials. The system will be at least twice as efficient (as measured by the number of approved drugs) as the industry standard.

Importance: The current process for drug development and approval is snaillike and ludicrously expensive, taking up to twenty years in some cases. It is also incredibly inefficient, as less than 15% of all drug candidates end up being approved. This XPRIZE will ensure that new drugs can be developed and tested much more rapidly, leading to a dramatic improvement in the pace of new drug release to the market.


  • marz62marz62 Seattle, WAPosts: 122 ✭✭✭
    edited September 2019
    This is a promising and potentially invaluable prize idea.

    There currently exists separate/distinct technologies (e.g., automated laboratory 'scientists' [R. King et al], automated continuous flow pharmaceutical formulation machines [Adamo et al) which, if combined or integrated optimally, could accomplish this challenge goal.

    I can also see IP issues arising if a competing team develops an automated system that comes up with a compound (or series of related compounds/derivatives) that a major Pharma company has previously developed but not publicly disclosed (and has a provisional patent pending, or, a formal patent application in the pipeline, or, has filed a 'bloc patent' application that includes unspecified but related compounds).

    There is also the real possibility of a 'negative finding' compound (tested, unsuccessfully, on a specific disease, by a Big Pharma company, and simply 'filed away') being 'rediscovered' by the automated system and 'repurposed' (i.e., applied to a different illness or condition) after in vivo/in vitro testing by a challenge team's automated system. Many pharma companies will patent unsuccessful drugs before (or without) fully vetting/testing them for other purposes (due to high costs associated with testing).

    Further, if the AI/automated system is able to compare the molecular structure of a 'new' compound to existing compounds* used for specific disease treatments, make certain 'tweaks' / alterations, and then produce a derivative compound that has greater target specificity/avidity, etc....this may lead to patent trouble if the alteration is an "obvious" modification to an existing drug.

    So, while I think the general concept of a 'drugbot' is good, there will need to be some function (integrated) that 'flags' discovered compounds that may be similar to, or infringe upon, existing drugs and their respective patents.

    If the pace of automated discovery (via a new 'drug discovery system') is exponentially faster/ more efficient than current methods, there could easily arise a 'patent bottleneck' (or a concomitant patent litigation morass). Licensing/IP issues could arise more frequently if a given automated system is truly more efficient and able to create or derive a larger volume of drugs (and their structural derivatives) in a given time period than a team of human chemists or molecular engineers/pharmacologists (working for a Pharma company, or research lab, public or private).

    * 'Existing compounds' refers to known compounds (whose structures have either been published, or have become generics due to patent expiration).
Sign In or Register to comment.