How AI is helping solve the labor issue in treating rare diseases

In the realm of modern biotechnology, despite having advanced tools for gene editing and drug design, a significant number of rare diseases remain without effective treatments. Industry leaders from Insilico Medicine and GenEditBio point to a critical factor that has hindered progress: the lack of skilled professionals to drive these initiatives forward. They assert that artificial intelligence (AI) is emerging as a transformative force, enabling scientists to tackle challenges that have been neglected for too long. During a recent discussion at Web Summit Qatar, Insilico's CEO, Alex Aliper, outlined the company’s vision of creating what he terms 'pharmaceutical superintelligence.' Insilico has introduced the 'MMAI Gym,' designed to train generalist language models like ChatGPT and Gemini to perform at the level of specialized models. This initiative aims to develop a versatile AI capable of handling a variety of drug discovery tasks simultaneously with exceptional accuracy. Aliper emphasized the urgent need for such technology to enhance productivity in the pharmaceutical sector, addressing the ongoing shortage of talent and labor. He pointed out the thousands of diseases that currently lack treatment options, highlighting the necessity of intelligent systems to confront these issues. Insilico’s innovative platform analyzes a plethora of biological, chemical, and clinical data to formulate hypotheses regarding disease targets and potential drug candidates. By automating processes that previously required extensive teams of chemists and biologists, the company claims it can efficiently navigate vast design spaces, identify high-quality therapeutic candidates, and even repurpose existing medications—significantly cutting costs and time. For instance, Insilico recently utilized its AI to determine the feasibility of repurposing existing drugs for the treatment of ALS, a rare neurological condition. However, the challenge extends beyond just drug discovery. Many diseases necessitate interventions that target fundamental biological processes. GenEditBio is at the forefront of the 'second wave' of CRISPR gene editing, shifting focus from editing cells externally to delivering precise treatments directly within the body. Co-founder and CEO Tian Zhu explained that their objective is to create a one-time injection that can effectively edit genes within the target tissue. They have developed a proprietary engineered protein delivery vehicle (ePDV), which mimics virus-like particles. Zhu elaborated on their AI-driven approach, which analyzes a vast library of unique, non-viral, non-lipid polymer nanoparticles to identify delivery vehicles that can transport gene-editing tools to specific tissues without triggering immune responses. The company’s NanoGalaxy platform leverages AI to study how chemical structures correlate with various tissue types, refining its delivery methods based on experimental results. Efficient and targeted delivery is essential for successful in vivo gene editing, according to Zhu. She believes that their method not only reduces production costs but also standardizes a historically complex process, making treatments more affordable and accessible for patients worldwide. Recently, GenEditBio secured FDA approval to initiate trials for a CRISPR therapy aimed at treating corneal dystrophy. Despite the advancements, a significant challenge remains: the data dilemma. Accurately modeling the complexities of human biology requires extensive high-quality data, which is currently limited. Aliper noted the need for more diverse patient data to enhance the robustness of their models, as existing data often reflects a biased demographic. Insilico's automated labs produce multi-layer biological data from disease samples on a large scale, feeding this information into their AI-driven discovery platform. Zhu pointed out that the data AI needs is already present within the human body, shaped by millennia of evolution. While only a small portion of DNA encodes proteins, the remainder provides crucial instructions for gene behavior—data that AI is increasingly able to interpret. Both companies are utilizing AI to test thousands of delivery nanoparticles simultaneously, generating valuable datasets that Zhu refers to as 'gold for AI systems.' These datasets are integral for refining their models and fostering collaborations with external partners. Looking ahead, Aliper envisions the creation of digital twins of humans to conduct virtual clinical trials, a concept still in its infancy. He highlighted the stagnation in drug approvals, with only about 50 new drugs receiving FDA approval annually, and expressed hope that in the next decade or two, there will be a significant increase in therapeutic options available for personalized patient care.