AI has been used to identify potential new drugs for Parkinson's Disease.
A new artificial intelligence (AI) based strategy has significantly sped up the identification of potential new drugs to treat Parkinson's disease. The work, published in the journal Nature Chemical Biology, could mean that new treatments for Parkinson's reach clinical trials and patients more quickly.
Drug discovery for serious diseases is often a slow, laborious and expensive process. Developing a drug from early laboratory testing through to full approval for use in patients typically takes 10-15 years.
“This is an extremely time-consuming process – just identifying a lead candidate for further testing can take months or even years," said Michele Vendruscolo leader of the research and professor in the Yusuf Hamied Department of Chemistry at the University of Cambridge in the U.K.
AI and machine learning techniques have shown promise in speeding up the initial stage of this process, by discovering potential drugs for cancers and several other diseases, leading dozens of biomedical startup companies to bet on the potential of AI for drug discovery.
"One route to search for potential treatments for Parkinson’s requires the identification of small molecules that can inhibit the aggregation of alpha-synuclein, which is a protein closely associated with the disease," said Vendruscolo in a press release.
The new study showed how an AI-based strategy sped up this process significantly and was a thousand times cheaper than traditional methods, identifying a small number of potentially useful compounds which were taken forward for laboratory testing. The results from these experiments were then fed back into the machine learning model to further optimize the predictions.
“The use of AI to develop machine learning approaches to drug discovery for protein aggregation diseases like Parkinson’s, has definitely arrived," said Michael S. Okun, M.D., National Medical Advisor for the Parkinson’s Foundation and Director of the Fixel Institute for Neurological Diseases at the University of Florida. "The over 20-fold improvement over typical high-throughput drug screening hit rates was impressive in this study and will add to the list of potential drugs to consider for clinical trials," added Okun, who was not involved in the research.
Nearly 90,000 Americans are diagnosed with Parkinson's disease annually, according to the Parkinson's Foundation, with a million people in the U.S. currently living with the disease. Despite this, there are currently no curative treatments for the disease, only drugs to manage symptoms which include tremors, balance and mobility issues and muscle stiffness.
"Machine learning is having a real impact on drug discovery – it’s speeding up the whole process of identifying the most promising candidates," said Vendruscolo. “For us, this means we can start work on multiple drug discovery programmes – instead of just one. So much is possible due to the massive reduction in both time and cost – it’s an exciting time."
However, discovering promising new compounds is only one, very early step in actually getting tried and tested drugs to patients.
“Whether this innovation however, will speed discovery of new Parkinson’s therapeutics is complicated, as introducing more compounds could actually slow the pipeline," said Okun. "Thus, a parallel and large investment will be needed in basic science research to better understand the pathogenesis of Parkinson’s disease and to more precisely apply this, and other novel AI derived drug discovery methodologies.”
