Published on October 25, 2023, 6:00 pm

Generative AI is being used in oncology to identify potential drug targets for cancer treatment. Insitro, a life sciences AI firm, has developed machine learning technology that can analyze images of cancerous tissue and accurately predict genetic changes with high accuracy. To overcome the scarcity of tissue samples for analysis, generative AI techniques are being employed to create synthetic images, multiplying the available samples. This enables scientists to uncover new drug targets and develop more effective treatments. The fusion of generative AI, machine learning, and quantitative biology has implications beyond healthcare, impacting various domains such as environmental conservation and sustainable agriculture. Embracing multi-modality, where generative AI methods are applied to different data types, represents a significant direction for future development in the field.

Generative artificial intelligence (AI) is set to play a crucial role in advancing breakthroughs in the field of oncology, according to Daphne Koller, co-founder and CEO of life sciences AI firm Insitro. Speaking at a workshop hosted by Stanford University’s Human-Centered AI institute, Koller explained how generative AI can be used to identify potential drug targets for cancer treatment.

Insitro’s machine learning AI technology can analyze images of cancerous tissue, providing valuable insights that are often overlooked by human pathologists. By “learning the language of histopathology,” the computer system can accurately predict genetic changes in cancer patients with an impressive 90% to 95% accuracy. This level of precision allows clinicians to identify specific genetic mutations that may affect a patient’s treatment plan.

However, one limitation researchers face is the scarcity of tissue samples needed for analysis. To overcome this challenge, Koller and her team utilized generative AI techniques to create synthetic images of pathology slides, effectively multiplying the available samples from hundreds to thousands. The increased sample size enables scientists to ask more meaningful questions and uncover novel drug targets that were previously unknown.

By using innovative tools such as an “ATAC-seq” assay developed at Stanford, Insitro was able to analyze almost 100,000 deep fake images of cancer tissue samples generated through generative AI. This method revealed previously unidentified genetic changes that could serve as potential drug targets for diseases like triple-negative breast cancer. These discoveries offer promising avenues for developing new and effective treatments.

Koller emphasized the complexity of applying generative AI in biology, stating that it surpasses human comprehension. She highlighted the need for extensive data collection in order to allow machines to interpret intricate patterns and redefine our understanding of human disease. The fusion of data-driven approaches with biological research has given rise to a new field known as digital biology. This emerging discipline combines advancements in data science and AI with techniques such as CRISPR and combinatorial chemistry, enabling scientists to manipulate biological systems in unprecedented ways.

The union of machine learning, AI, and quantitative biology has far-reaching implications not only in healthcare but also in environmental conservation, energy production, bio-materials development, sustainable agriculture, and various other domains. By harnessing the power of generative AI and machine learning, researchers are engineering a future that promises better outcomes for our society.

The workshop hosted by Stanford University showcased the diverse applications of generative AI across different fields. Speakers like Daphne Koller shed light on the importance of embracing multi-modality, where generative AI methods can be applied to various data types such as biological data and even audio recordings of whale songs. This concept represents a significant direction for the field’s future development.

In conclusion, generative AI is revolutionizing the way we approach cancer research and drug discovery. With its ability to unveil hidden patterns in vast amounts of data, this technology offers great potential for identifying novel drug targets and advancing medical treatments. As we continue to explore the possibilities of generative AI across disciplines, we are paving the way for a brighter future characterized by improved health outcomes and scientific advancements.


Comments are closed.