Machine learning is transforming the pharma sector

In the U.S., drug companies spend more than $50 billion on R&D, while in Europe spending surpasses €30 billion. To help lower costs and to reduce the time taken for new drugs to hit the market, pharma is turning to machine learning.

The pharmaceutical regulatory environment is becoming more challenging, and drugs must go through extensive testing before they hit the market. As a result, there are major incentives for drug companies to reduce R&D spending in order to free up funds for additional ventures and offer lower prices for their products.

By adopting sophisticated data science, and machine learning, pharmaceutical researchers can save money and time on R&D. On top of that, machine learning technology provides new ways for drug companies to streamline nearly every other aspect of their businesses.

To outline how machine learning can be best applied, data science software company, Dataiku, has recently released a whitepaper titled “How Machine Learning is Transforming Pharmaceuticals”.

The paper assesses how drugs have been discovered in the past and notes that rather than through ‘inspiration’ or years of iterative experimental approach, the next major medical breakthroughs will come as the result of data review and data analysis.

The paper goes on to assess how big data and machine learning are transforming the pharmaceutical industry. The focus is with how big data analytics can improve the patient recruitment process. This involves accessing and searching cloud services, where relevant databases contain terabytes of publicly- and privately-held data to enable clinical trial recruitment with better precision.

Another example is with Natural Language Processing, which speeds up hypothesis testing. When scientists design a drug trial tend to use their own knowledge of medical literature and previous studies. Now algorithms can screen decades of different studies and trials.

A third application is with data science methods, which can assist scientists with identifying patterns in public and private datasets far faster and more meaningfully. Such analytics can also assist with developing personalized medicines, where drugs can be customized to small populations of patients with particular genetic profiles.

Analytics can also aid companies with optimizing the shipping of medications, finding the quickest and most cost-effective routes. A different application is with assessing prescribing rates, to verify that patients genuinely need the medication and to ensure that medics are consistently prescribing drugs based on patient need.

Posted by Dr. Tim Sandle, Pharmaceutical Microbiology