MIT researchers are finding ways to use data analytics in improving the effectiveness of small molecule drugs.

MIT researchers have come up with a solution for the pharmacokinetic limitations of small molecule drugs. These drugs are often used for the treatment of a number of diseases, but their effectiveness is often deprived by the reaction of the body to those drugs. However, it has been studied that nanoparticles, made from lipid or polymers, can dissipate this problem when they are made by a combination of inactive ingredients of drug with small molecule drugs. But these juxtapositions are very difficult to produce and all combinations do not have the capacity to hold enough drugs. So, this kind of situation demands a technology that would automatically screen and identify suitable nanoparticles. MIT researchers are experimenting to use aspects of machine learning to make this identification process successful. Daniel Reker, the lead of this study has stated, "By working at the interface of data science, machine learning, and drug delivery, our hope is to rapidly expand our toolset for making sure a drug gets to the place it needs to be and can actually treat and help a human being."

 The details of the experiment

The researchers have reportedly experimented with 2.1 million pairings and have identified 100 new models of nanoparticles that can be used for treatments of malaria, viral infections, fungal diseases, cancer, and asthma. They are focusing on the positive effects of the inactive ingredients in the process of nano formulation.

For developing a suitable machine learning algorithm for this experiment, the researchers have built a big data set. They have selected 16 types of self-aggregating small-molecule drugs and combined them with a diverse set of 90 widely available compounds. They have made sure that the drug and the ingredient are both FDA-approved so that the resulting nanoparticle would be more likely to pass the FDA-approved test. After the experiment, all of the nanoparticles were tested in Swanson Biotechnology Center at Koch Institute. They have loaded 384 samples at a time to the nano well plates with the help of robotics. The machine learning platform is currently trained on 1440 points. Screening 788 numbers of small-molecule drugs with 2,600 inactive drug ingredients, the platform has been able to identify 38,464 potential self-assembling nanoparticles from 2.1 million numbers of possible combinations.

A new edge in cancer treatment

One of the selected nanoparticles is sorafenib, which is useful for advanced liver cancer and has been combined with glycyrrhizin, commonly used for licorice flavoring. Despite being a drug for cancer treatment, sorafenib had limited effectiveness, but the combined nanoparticle worked twice better than the primary drug as more quantity of the drug was able to enter the cells. The new invention was tested upon model mice suffering from liver cancer and they seemed to have better longevity when compared to the treatment of sorafenib alone.

Customized service

Apart from improving the effectiveness of the drugs, this particular machine learning platform would also customize the inactive compound that would be suitable for the requirement of the individual patients. Some of the inactive ingredients often lead to allergic reactions for the patients. This can be easily avoided with the help of this advanced machine learning toolbox. Reker has further approved this matter by saying, "We now have an opportunity to think about matching the delivery system to the patient." He is confident that matters like drug absorption, genetics as well as allergies can be reduced.

This innovative experiment would be ground-breaking when it comes to accelerating the pace of drug design. The research team is further willing to make medicines for those who are not well suited to the standard formulation. Using data analytics would make them successful in having a microscopic approach to the medication process.