Summary: Artificial tool has identified medicine applicants for over 17, 000 unique and neglected conditions, offering hope for thousands of affected individuals. The first type created especially to recycle previously used drugs for ill-health, TxGNN, stands out.
It predicts which medication may work and potential side effects, providing innovative strategies for care. The Artificial concept is completely free to use and aims to make medical discoveries easier for the sick.
Important Information:
- TxGNN identified medicine prospects for over 17, 000 conditions, including neglected ones.
- The unit uses existing drugs to propose new treatments, reducing expenses and time.
- It was 50 % more powerful than previous versions at identifying drug prospects.
Origin: Harvard
There are more than 7, 000 unique and untreated diseases worldwide.
Although each state affects just a small number of people, some 300 million people worldwide are affected by them, which puts an enormous human and financial burden on society.
However, with a simple 5 to 7 percent of these problems having an FDA-approved medication, they remain generally neglected or undertreated.
A new artificial intelligence resource can help people with rare and mistreated conditions and the practitioners who treat them by discovering new treatments from existing ones, providing hope for individuals with rare and mistreated conditions.
The TxGNN AI type is the first to be created specifically to detect drug individuals for rare diseases and non-treating conditions.
It identified medication applicants from existing medications for more than 17, 000 conditions, many of them without any existing therapies. This is the highest number of illnesses that any one AI design has been able to treat to date. The design, according to the researchers, could be applied to even more conditions besides the 17 000 it was tested on in the first experiments.
The job, described Sept. 25 in , Nature Medicine, was led by experts at Harvard Medical School.  , The scientists have made the tool , available for free , and want to promote clinician-scientists to use it in their search for new treatment, particularly for circumstances with no or with minimal treatment options.  ,
” With this tool we aim to identify new treatment across the disease spectrum but when it comes to rare, ultrarare, and neglected circumstances, we foresee this concept could help near, or at least narrow, a space that creates serious health discrepancies”, said guide researcher , Marinka Zitnik, assistant professor of biomedical computing in the Blavatnik Institute at HMS.
This is where we can see the potential of AI in lowering the burden of global diseases, finding new uses for existing medications, and making new treatments more affordable than creating brand-new drugs from scratch, Zitnik, who is an associate professor at the Harvard University’s Kempner Institute for the Study of Natural and Artificial Intelligence, said Zitnik.
The new tool has two key features: one that lists treatment candidates and potential side effects, and the other that explains the justification for the choice.  ,
In total, the tool identified drug candidates from nearly 8, 000 medicines ( both FDA-approved medicines and experimental ones now in clinical trials ) for 17, 080 diseases, including conditions with no available treatments.
Additionally, it identified which medications would have contraindications and side effects for particular diseases, something that the current drug discovery strategy mostly finds through trial-and-error in early clinical trials that were centered on safety.
Compared against the leading AI models for drug repurposing, the new tool was nearly 50 percent better, on average, at identifying drug candidates. It was also 35 percent more accurate in predicting what drugs would have contraindications.
advantages of utilizing medications that have already been approved
Repurposing old medications is a promising method for creating new treatments because it relies on them because they have been studied, have been thoroughly studied, and have gone through the regulatory approval process.
Most medications have multiple effects beyond the specific targets for which they were developed and tested initially. However, many of these effects are still undiscovered and understudied during initial testing, clinical trials, and review, only emerging after years of use by millions of people.
Nearly 30 % of FDA-approved medications have, in fact, acquired at least one additional indication for treatment after receiving initial approval, and many have developed tens of additional treatment indications over the years.
This method of drug recycling is at best haphazard. It depends on patient reports of unanticipated positive side effects or on doctors ‘ judgment when to use a drug for a condition that was n’t intended for, a practice known as off-label use.
We’ve tended to rely on luck and serendipity rather than strategy, which restricts drug discovery to conditions for which drugs are already available, Zitnik said.
The benefits of drug repurposing extend beyond diseases without treatments, Zitnik noted.
New drugs could offer alternatives with fewer side effects or replace medications that are ineffective for some patients, she said, even for more common diseases with approved treatments.
What distinguishes the new AI tool from the more recent models?
Most modern AI algorithms focus on just one disease or a few conditions, as is the case with most drug discovery algorithms. The new tool was trained in a way that enables it to use existing data to make new predictions rather than focusing on particular diseases. It does so by identifying shared features across multiple diseases, such as shared genomic aberrations.
For instance, the AI model can extrapolate from well-understood diseases with known treatments to poorly understood diseases with no known treatments based on common genomic underpinnings.
According to the research team, this capacity brings the AI tool closer to the kind of reasoning a human clinician might employ to generate novel ideas if they had access to all the preexisting knowledge and raw data that the AI model processes but the human brain is unable to possibly access or store.
The tool was trained on vast amounts of data, including DNA information, cell signaling, levels of gene activity, clinical notes, and more. The researchers tested and refined the model by asking it to perform a variety of tasks. Finally, the tool’s performance was examined against 1.2 million patient records to determine potential drug candidates for various illnesses.
The researchers also questioned the tool’s ability to predict which patient characteristics would make the drug candidates identified as contraindicated for some patient populations.
Another task involved examining the tool to discover any small molecules that might have the ability to effectively halt the activity of specific proteins involved in disease-causing processes and pathways.
The researchers prompted the model to find drugs for three uncommon conditions that it had not seen as a part of its training, including a neurodevelopmental disorder, a connective-tissue disease, and a rare genetic condition that causes water imbalance, using a test designed to assess its ability to reason as a human clinician might.
The researchers then compared the drug therapy recommendations in the model to current medical knowledge regarding the mechanism of action. In every example, the tool’s recommendations aligned with current medical knowledge.
Additionally, the model provided the justification for its choice as well as the names of medicines for all three diseases. This explainer feature increases physician confidence and promotes transparency.
The researchers warn that any therapies that the model has suggested would require additional testing for dosing and delivery times. However, they add that the new AI model would enable drug repurposing in a way that has n’t been previously possible due to this unprecedented capacity. The team is already collaborating with a number of foundations for rare diseases to find potential treatments.
Authorship, funding, disclosures
Co-authors included Kexin Huang, Payal Chandak, Qianwen Wang, Shreyas Havaldar, Akhil Vaid, Jure Leskovec, Girish N. Nadkarni, Benjamin S. Glicksberg, and Nils Gehlenborg.
Funding: This work was supported by National Science Foundation CAREER award ( grant 2339524 ), National Institutes of Health ( grant R01-HD108794 ), U. S. Department of Defense ( grant FA8702-15-D-0001 ), Amazon Faculty Research, Google Research Scholar Program, AstraZeneca Research, Roche Alliance with Distinguished Scientists, Sanofi iDEA-TEC H Award, Pfizer Research, Chan Zuckerberg Initiative, John and Virginia Kaneb Fellowship at HMS, Biswas Family Foundation Transformative Computational Biology Grant in partnership with the Milken Institute, HMS Dean’s Innovation Awards for the Use of Artificial Intelligence, Kempner Institute for the Study of Natural and Artificial Intelligence at Harvard University, and Dr. Susanne E. Churchill Summer Institute in Biomedical Informatics at HMS.
About this news about pharmacology and AI research
Author: Ekaterina Pesheva
Source: Harvard
Contact: Ekaterina Pesheva – Harvard
Image: The image is credited to Neuroscience News
Original Research: The findings will appear in Nature Medicine
