Artificial Intelligence to Run Optimal Clinical Trials: Are We There Yet?
Have you recently asked Siri or Cortana for help? Did you use Uber to take you to the bar last weekend? Are you waiting for your next order from Amazon to arrive at your doorstep?
It's the era of automation, Silicon Valley, and predictive analytics. Technology is evolving at breakneck speed and data is king. For anyone who's ever noticed the eerily on-point nature of internet advertising, it can seem like your computer knows what you need even before you even realize it yourself. Development of sophisticated technology like artificial intelligence (AI) and machine learning have now replaced neurons with neural networks.
Not a day goes by when you don't see articles about AI and machine learning. For some, these phrases bring visions of new and uncharted opportunities, while for others, they seem scary and complex. Creating AI software is, no doubt, a complicated adventure, but that doesn't mean the average person can't understand the simple concepts underlying these high-level terms. After all, most AI and machine learning mechanisms, at their basic level, are infantile attempts to re-create the workings of the human mind.
Industries such as retail and travel have already started to harness the power of AI through personalized shopping and augmented reality techniques. Can this disruptive technology change the world of life sciences and e-clinical solutions as well? Can AI can help us improve processes, products, and performance? Will a clinical trial run more efficiently in the near future? Let's find out!
According to Vladimir Pyagay, Product Manager for TransPerfect Trial Interactive, "When I first began learning about the different methodologies and concepts available in the tech world for creating AI programs, I quickly started to notice many similarities between them and human cognition. It didn't take long to realize that many of these techniques had concepts from psychology as their basis."
Pyagay, who will be presenting a panel on the topic at DIA 2017, taking place June 18-22 in Chicago, extrapolates how a technique called Incremental Concept Learning could be used in TMF Management. The core idea behind it is to write an algorithm that can continuously create and adjust model frames in its memory based on new positive and negative examples of a concept. The training for this AI agent would be to feed a set of examples into it and specify whether each example is positive or negative. The algorithm would compare the set of attributes between the good and the bad examples and construct a general model in its mind of what set of attributes would positively identify an object matching the concept correctly. Once the agent is trained, it can then attempt to identify any new objects as a match or mismatch for the concept.
For TMF management, this type of agent would be able to answer questions such as "Is it or is it not a 1572 document?" or "Is it or is it not a medical license?" and so on.
A child often observes behaviors of people and objects around them, and imbibes behavioral and cognitive patterns solely based on observation and senses. As the reactions to those behaviors change over time, his/her mind then draws conclusions about what is good and what is bad, what is a fork versus what is a spoon, and so on. The mind creates models for each of these concepts and continues to evolve them over time as the child acquires new experiences. Incremental concept learning is one of many examples of how AI methods are being derived from human cognition.
Attend "The Impact of Artificial Intelligence on Clinical Trials" on June 19, 11 AM at DIA 2017 to hear Pyagay and his fellow panelists dive into more detail on the use of this exciting technology in running efficient and intuitive clinical trials.
Looking for a demo? Reach out to us at email@example.com or click here to schedule an in-person meeting at DIA.