Recently, scientists at MIT have designed an Artificial Intelligence (AI) model that finds potential drug molecules a thousand times faster. The deep-learning model designed at MIT looks at drug geometry and chemical structure of pharmaceutical actives to identify drugs that have a higher probability of succeeding in a clinical trial. In a paper that will be presented at the ICML – International Conference on Machine Learning, the MIT model, Equibind is purported to be 1,200 times faster than existing drug discovery models in predicting promising pharmaceutical molecules and their optimal protein targets.
Besides MIT, work at Columbia University also is leveraging AI to predict new pharmaceutical medications to help revolutionize cancer treatment. Researchers at the Frederick National Laboratory for Cancer Research (FNL) and Columbia University recently applied their data science knowledge to develop a workflow and machine learning model that accelerated the discovery of a novel drug compound that blocks the action of centromere-associated protein-E (CENPE) that is a cornerstone of cancer therapy. CENP-E research by Columbia University researchers is currently being used to perform large scale simulation studies and pave the ground for future clinical trial for this novel cancer therapy.
At Harvard University’s Wyss Institute, Circa Vent, is a novel Drug Discovery Platform that leverages machine learning, and artificial intelligences to find drugs that can be repurposed to treat complex mental health conditions such as bipolar disorder thus saving valuable research time and cost.
At Cornell University’s Englander Institute for Precision Medicine at the Weill Cornell College of Medicine, researchers have created the PMKB bot with help from Microsoft that helps doctors identify the right drugs to target the right genes based on text and voice interactions with the Precision Medicine Knowledgebase.
The applications of AI at major universities such as MIT, Columbia, Harvard, and Cornell are testament to society’s drive to dramatically reduce the time and cost of preclinical drug discovery. Projects such as Equibind, CENPE, Circa Vent and PMK Bot help deliver on the promise for new drug-like molecules or re-application of older molecules to new diseases.
The process of drug discovery is computationally and financially risky and extremely expensive for most pharmaceutical companies. According to the FDA (Food and Drug Administration), approximately 90% of drugs fail upon being tested on humans. Moreover, the price of failure for drugs according to an article published in the Journal Nature Reviews Drug Discovery, predicts that the price of failure for a drug can be up to $1.3 billion USD. To compensate for the high cost and the high risk associated with drug discovery, pharmaceutical companies often justify charging higher drug prices.
Applications of Artificial Intelligence and Machine Learning to drug discovery and development of new pharmaceutical analytics products are disrupting the healthcare industry. These new technologies are helping eliminate weak drug candidates that have a high probability of clinical trial failure and helping scientists and researchers zero in on drugs and protein targets that have the highest probability of success thus resulting in immense time and cost savings for pharmaceutical companies and research institutes.