A lack of proper data is hurting the use of machine learning to develop drugs, which could put U.S. drugmakers at a competitive disadvantage compared to other countries, according to a report from the U.S. Government Accountability Office and the National Academy of Medicine.
Machine learning is a type of artificial intelligence that involves using data to train computers to make decisions and learn from experiences, according to Pharmaphorum. It has the potential to cut costs of research and development for drugmakers by helping researchers to predict what will and won't work in clinical trials.
However, the report says a lot of the data being used in drug development is not suitable for machine learning purposes.
There is a phenomenon known as "garbage in, garbage out," where a machine learning system can't produce credible results because of poor data, according to Pharmaphorum. Biases in data, such as under-representation of certain populations, can limit machine learning's effectiveness.
There is also the issue that a lot of drugmakers don't share their data due to costs, legal issues and lack of economic incentives. There's also a shortage of skilled workers in the machine learning field.
Drugmakers also tend to be confused about what the regulations are surrounding machine learning, which limits investments in it.
However, other countries have invested more into machine learning, which could put U.S. drugmakers at a competitive disadvantage in regards to developing new drugs, Pharmaphorum reported.
The report made recommendations for policymakers to help overcome the challenges facing machine learning, such as enacting legislation to prevent the improper sharing or use of data and creating incentives for drugmakers to share high-quality data.
Read the full article here.