Maintenance window scheduled to begin at February 14th 2200 est. until 0400 est. February 15th


Forgot Password?

    Or login with Facebook
    Defense Visual Information Distribution Service Logo

    The Sum is Greater than Its Parts

    The Sum is Greater than Its Parts

    Courtesy Photo | 2D structure of reserpine. (National Center for Biotechnology Information...... read more read more



    Courtesy Story

    Defense Threat Reduction Agency's Chemical and Biological Technologies Department

    New artificial intelligence (AI) and deep learning-assisted projects aim to accelerate the multiple stages of the drug research and development (R&D) process by using drug repurposing to enable treatments to be approved faster. The Defense Threat Reduction Agency’s (DTRA) Chemical and Biological Technologies Department in its role as the Joint Science and Technology Office (JSTO) for the Chemical and Biological Defense Program is investing in research at the Massachusetts Institute of Technology (MIT) to develop new AI and deep-learning technologies capable of automating scientific data analysis and interpretations in the early stages of drug discovery.

    Recently, the U.S. Government supported accelerated COVID-19 therapeutic drug trials, and investigators began a rigorous selection process of the vast number of candidate drugs with potential benefits. Scientists have made significant progress in treating presymptomatic and early stage COVID-19 cases using repurposed drugs, such as molnupiravir that DTRA-JSTO had invested in for development as a Venezuelan equine encephalitis virus treatment; however, more advanced tools are needed to prioritize drugs and predict repurposed drug combinations that can be beneficial across all disease stages.

    Advanced AI combined with deep learning will automate repurposed drug prioritization with increasing sophistication and reduce the need for hands-on, time-consuming experimentation. The drug-development process normally lasts years, and sometimes decades, beginning with a discovery phase, progressing to R&D, testing in human clinical trials, and then achieving FDA approval if the drug is shown to be safe and effective. The ongoing COVID-19 pandemic illustrates how critical it is to reduce timelines at each stage of drug-development from decades or years to months or weeks.

    For medical personnel, an indication—a sign, symptom, or medical condition—leads to their recommendation for a treatment, test or procedure. Researchers explore drug repurposing to discover new disease indications for pharmaceuticals that have already been tested in clinical trials, which enables them to skip most of the safety and toxicity testing already completed. The MIT research team has gone a step further than others in this field by developing a new AI-assisted computerized approach for discovering drugs that work together with increased effectiveness, or synergism, to alleviate COVID-19 symptoms. They described their approach in a recent article, “Deep learning identifies synergistic drug combinations for treating COVID-19.”

    Synergistic drug combinations are frequently used to treat many diseases, including cancers, asthma, diabetes, hypertension, cardiovascular disease, and some infectious diseases like HIV, viral hepatitis, and tuberculosis. The “one bug, one drug” development approach has a low historical success rate for treating infectious diseases. By contrast, discovering synergistic drugs that act on the infected human, rather than just on the bug, has potential for treating more than one infectious disease.

    Synergistic therapies have many advantages and can:

    • Usually be given at lower doses than single-drug treatments

    • Have more robust therapeutic effects

    • Have a lower incidence of undesirable side effects

    • Reduce the likelihood of developing single-drug resistance in viral and bacterial infections

    Current machine-learning approaches have limited capacity for predicting drug combinations for emerging diseases because they require large quantities of data, like those for cancer and diabetes. This amount of data is not readily available for most infectious diseases and to a lesser extent for emerging diseases such as COVID-19. Even less data is available from drug-combination screening studies, making it difficult to discover synergistic treatments.

    To address the lack of available data for infectious and emerging diseases, computer scientists at MIT developed new deep-learning methods and created a drug-prediction tool named ComboNet that improves drug-target and drug-drug synergy predictions. ComboNet can perform the following actions with better accuracy than preexisting tools and can predict:

    • If a drug binds to a specific biological target

    • Each drug’s antiviral activity

    • How two drugs interact

    The MIT computer scientists used ComboNet to identify two new drug combinations that could be used to combat COVID-19: (1) remdesivir plus reserpine, and (2) remdesivir plus IQ-1S.
    Recently, the FDA expanded approval for using remdesivir to treat patients 28 days of age and older, including those both hospitalized and not hospitalized, for mild-to-moderate COVID-19.
    Remdesivir has FDA approval for treating high blood pressure. IQ-1S is an experimental drug. In cell viability experiments, both combination medicines demonstrated potent SARS-CoV-2 antiviral activity.

    The new machine-learning method and ComboNet drug prediction tool enables scientists to rapidly identify new combination treatments for emerging infectious diseases with greater accuracy than previous tools, which will help minimize the health impacts of infectious diseases and biological warfare threats to the Joint Force.

    Proceedings of National Academy of Sciences “Deep learning identifies synergistic drug combinations for treating COVID-19”

    POC: Dale Taylor,



    Date Taken: 05.13.2022
    Date Posted: 05.13.2022 16:48
    Story ID: 420715
    Location: FT. BELVOIR, VA, US

    Web Views: 183
    Downloads: 0