Testing out newly developed drugs is an
extremely time-consuming process, and it can be difficult to get
right. Now, a team of scientists at Carnegie Mellon University (CMU)
is working to streamline the task, creating a robotically-driven
experimentation system that's able to reduce the number of tests
that have to be carried out by as much as 70 percent.
When working on a new drug, scientists have to determine its effects to ensure that it's both an effective treatment and not harmful to patients. This is hugely time-consuming, and it's simply not practical to perform experiments for every possible set of biological conditions.
UPGRADE TO NEW ATLAS PLUS
More than 1,200 New Atlas Plus subscribers directly support our journalism, and get access to our premium ad-free site and email newsletter. Join them for just US$19 a year.UPGRADE
That's where CMU's new robotic system steps in. It uses a machine learning approach to choose which experiments to conduct, using patterns in the data to accurately predict results of experiments without actually carrying them out.
The system is able to conduct selected experiments on its own, using liquid-handling robots and an automated microscope. Its abilities were put to the test in a study to determine the effects of 96 drugs on 96 cultured mammalian cell clones, containing different, fluorescently-tagged proteins. A total of 9,216 experiments were possible, each of which involved testing the effects of a drug by taking a picture of it mixing with the target cell.
The machine began by imaging all 96 cells, pinpointing the location of the protein within it. The effects of each drug were then recorded in the same way, with the machine learning algorithm slowly identifying patterns in the location of the proteins, known as phenotypes.
By grouping together similar images, the machine learner was able to identify potential new phenotypes without help from the researchers. As more data was gathered, it was used to form a predictive model, guessing the outcomes of unmeasured experiments.
A total of 30 rounds of testing were undertaken by the automated system, with 2,697 experiments completed out of the possible 9,216. The rest of the outcomes were predicted by the machine, to an impressive accuracy rate of 92 percent.
The researchers believe that their work proves that machine learning techniques are viable for use in medical testing, and could have a big impact on both the practical and financial issues faced by the field.
"The immediate challenge will be to use these methods to reduce the cost of achieving the goals of major, multi-site projects, such as The Cancer Genome Atlas, which aims to accelerate understanding of molecular basis of cancer with genome analysis technologies," said senior paper author Robert F. Murphy.
The findings of the research were published online in the journal eLife.