AI-Powered Drug Discovery: Addressing Labor Shortages in Rare Disease Treatment

Despite significant advances in biotechnology, including gene editing and precision drug design, thousands of rare diseases remain without effective treatment options. Industry leaders from Insilico Medicine and GenEditBio identify a critical bottleneck: insufficient specialized human resources to address these neglected conditions. Artificial intelligence is emerging as a transformative solution, enabling researchers to tackle previously intractable therapeutic challenges.

At Web Summit Qatar, Insilico Medicine's CEO and founder Alex Aliper outlined the company's vision for developing "pharmaceutical superintelligence." The organization recently launched its MMAI Gym initiative, designed to train generalist large language models—comparable to ChatGPT and Gemini—to perform at the level of specialized domain models. The objective is to create a multi-modal, multi-task AI system capable of simultaneously addressing diverse drug discovery challenges with superhuman precision.

"We require this technology to enhance pharmaceutical industry productivity and address the critical shortage of specialized talent, as thousands of diseases still lack curative treatments or any therapeutic interventions. Numerous rare disorders remain neglected, necessitating more intelligent systems to address this gap," Aliper stated.

Insilico's platform integrates biological, chemical, and clinical datasets to generate hypotheses regarding disease targets and candidate molecules. By automating processes that traditionally required large teams of chemists and biologists, the platform can efficiently explore vast molecular design spaces, identify high-quality therapeutic candidates, and repurpose existing pharmaceuticals—significantly reducing both development time and costs. Recent applications include using AI models to identify existing drugs for potential repurposing in treating ALS, a rare neurological disorder.

However, labor constraints extend beyond drug discovery. Even when AI successfully identifies promising targets or therapies, many diseases require interventions at fundamental biological levels, particularly in gene editing applications.

GenEditBio represents the "second wave" of CRISPR gene editing technology, transitioning from ex vivo cellular editing to precise in vivo delivery systems. The company's approach aims to enable gene editing through single-injection administration directly into affected tissues.

"We have developed a proprietary engineered protein delivery vehicle (ePDV)—a virus-like particle. We leverage natural biological systems and apply AI machine learning methodologies to mine natural resources and identify viral structures with affinity for specific tissue types," explained GenEditBio co-founder and CEO Tian Zhu.

GenEditBio maintains an extensive library containing thousands of unique, non-viral, non-lipid polymer nanoparticles—essentially specialized delivery vehicles designed to safely transport gene-editing tools into target cells. The company's NanoGalaxy platform utilizes AI to analyze correlations between chemical structures and specific tissue targets (including ocular, hepatic, and nervous system tissues). The AI predicts optimal chemical modifications to delivery vehicles that enable payload transport without triggering immune responses.

GenEditBio conducts in vivo testing of its ePDVs in wet laboratory environments, with results fed back into the AI system to refine predictive accuracy iteratively. Efficient, tissue-specific delivery represents a fundamental requirement for in vivo gene editing, according to Zhu. She argues this approach reduces manufacturing costs and standardizes processes that have historically been difficult to scale.

"This approach is analogous to developing off-the-shelf therapeutics effective across multiple patients, making treatments more affordable and globally accessible," Zhu noted. The company recently received FDA approval to initiate clinical trials for CRISPR therapy targeting corneal dystrophy.

Addressing the Persistent Data Challenge

As with many AI-driven systems, biotech progress ultimately encounters data limitations. Modeling edge cases in human biology requires substantially more high-quality data than currently available.

"We require additional ground truth data from patient populations. The existing data corpus is heavily biased toward Western populations where it originates. We need greater local data generation efforts to create a more balanced ground truth dataset, enabling our models to handle diverse populations more effectively," Aliper emphasized.

Aliper noted that Insilico's automated laboratory facilities generate multi-layered biological data from disease samples at scale without human intervention, which is subsequently integrated into its AI-driven discovery platform.

Zhu suggests the data AI requires already exists within the human genome, shaped by millennia of evolution. While only a small fraction of DNA directly encodes proteins, the remainder functions as regulatory instructions governing gene expression. This information, historically difficult for human interpretation, is becoming increasingly accessible to AI models, including recent developments like Google DeepMind's AlphaGenome.

GenEditBio applies similar principles in laboratory settings, testing thousands of delivery nanoparticles in parallel rather than sequentially. The resulting datasets, which Zhu describes as "gold for AI systems," are used for model training and increasingly support external collaborative partnerships.

According to Aliper, one of the next major initiatives involves constructing digital twins of humans for virtual clinical trial execution—a process currently in nascent stages.

"We're plateaued at approximately 50 FDA-approved drugs annually, and we need to see growth. There's a rising incidence of chronic disorders as the global population ages. My expectation is that within 10 to 20 years, we will have significantly more therapeutic options for personalized patient treatment," Aliper concluded.

Sources:
Fortune - Insilico Medicine MMAI Gym
DDW Online - GenEditBio FDA Approval
Nature - FDA Drug Approvals