AI’s Impact on Drug Design, Regenerative Medicine, and Beyond with Stem Cell Expert Dr. Ermelinda Porpiglia

In our conversation for the “Data Driven Biotech” series by AppliedXL, Dr. Porpiglia shares her insights on the applications of AI in biotech research. She acknowledges that while AI helps overcome challenges like data complexity, integration issues, and scalability, human expertise still is and always will be essential, especially for experimental validation of AI-based predictions and ethical considerations. Given the unpredictable nature of AI’s evolving capabilities, there must be ongoing and rigorous research to learn how to harness this technology responsibly.

What do you think are some of the most exciting uses of AI and data in biotech and research, and why? (e.g. new drug discovery, competitive intelligence, identifying new business opportunities, making winning bets etc.)

“Artificial intelligence (AI) is transforming biotech, and I am particularly excited by the opportunities it will bring to drug design and discovery. AI models can integrate multiple layers of scientific data, enabling, for example, the prediction of unique antibody sequences, a leap forward in the development of targeted therapeutics. A recent example is EVQLV’s in-silico platform, which facilitates innovative antibody design by computationally generating novel complementarity-determining regions (CDRs) based on the known structures of two interacting partners. This approach aids in overcoming obstacles such as limited diversity, protein folding constraints, and lead optimization typically associated with antibody development.”

“Additionally, AI’s potential is notable in processing high-dimensional tissue images at the single-cell level, to create maps of local cellular neighborhood within tissues and therefore generate spatial signatures in different biological contexts. Advanced imaging platforms used in biomedical research, like imaging mass cytometry, provide intricate cell-cell interaction data, essential for a deeper understanding of tissue regeneration and for the development of novel regenerative medicine interventions. These data sets enable us to comprehend how cellular components of stem cell niches interact to support stem cell function during effective tissue regeneration, and to monitor alterations in these relationships caused by aging and disease, as well as the response to pharmacological interventions. By leveraging this information, it is possible to quantify different cell types in the stem cell niche, determine their relationships by measuring distances between cell types, and identify cellular neighborhoods associated with effective regeneration. This understanding provides insights into the alterations that occur with diseases.”

“AI algorithms can efficiently handle these high-dimensional analyses, improving accuracy of single-cell identification and uncovering data trends that might elude the human eye, thereby offering deeper insights into stem cell dysfunction.”

What are some of the big challenges the biotech industry faces when working with large amounts of data, and how can the industry address these potential hurdles?

“One of the greatest challenges of big data is the high dimensionality, which cannot be possibly processed by the human brain. AI allows us to extract meaningful trends from these high dimensional data that can uncover novel mechanisms underlying impaired tissue regeneration in aging and disease. For instance, these algorithms can identify stem cell aging signatures and potential therapeutic targets by training on data from different age groups.”

“Moreover, AI helps overcome other big data challenges, such as:

• Data Integration: AI can integrate data from diverse technologies, providing vital insights into disease development, progression, and treatment responses.
• Data Quality: AI can identify inconsistencies within datasets from cutting-edge technologies, aiding the development and testing of normalization tools for accurate data analysis and interpretation.
• Scalability: As the volume of data grows exponentially with technological advancements, AI can enhance data processing, storage, and analysis efficiency. By leveraging cloud computing and other advanced technologies, AI allows scientists to analyze data on an otherwise unachievable scale using traditional methods.”

What impact do you think tools like STAT Trials Pulse by AppliedXL play in supporting decision making?

“STAT Trials Pulse by AppliedXL is an AI-driven platform that provides real-time insights into clinical trial activities, helping monitor multiple ongoing trials and analyze efficacy and toxicity responses. It facilitates quick analysis of large datasets, enabling scientists to gain valuable information about the latest therapies for specific diseases.”

“Tools like STAT Trials Pulse can significantly impact biotech and healthcare strategic decision-making, as it allows stakeholders to make informed decisions about drug development, clinical trial design, and investments. A tool like this can uncover potential areas for drug development, optimize trial design, and improve resource utilization. By offering a centralized data-sharing platform, it enhances collaboration between biotech and healthcare organizations.”

What is the role of human expertise, with the increasing use of AI and data?

“While AI is increasingly employed in biotech and accelerates scientific advancements, the value of human expertise remains unmatched. Human knowledge, experience, and critical thinking are still vital for:

1. Designing intricate experiments that evaluate drug target predictions made by AI, whilst considering the complexities inherent in biological systems.
2. Deciphering the biological implications of data patterns identified by AI.
3. Ensuring that treatments are not only effective but also safe and ethically sound.”

“AI aids researchers in processing vast data volumes with heightened speed and efficiency. However, it lacks the curiosity that pushes the boundaries of scientific knowledge, and the creativity that fuels innovation.”

How do you see AI and data continuing to transform the biotech industry in the next 5-10 years?

“Machine learning can process large data sets beyond human capabilities, finding patterns across millions of data points for superior performance. Historically, its application has been limited to specific tasks. In my opinion, the future of AI involves flexible, reusable models that will be able to perform new tasks, opening the door to novel opportunities in science and medicine. Foundation models, the latest generation of AI models, are trained on diverse datasets, can perform varied tasks, and solve new problems without retraining. Such models will be able to integrate different data types such as OMICs data, and produce outputs that demonstrate advanced reasoning abilities.”

“Recent advances in foundation model research, such as multimodal architectures, self-supervised learning, and in-context learning, will disrupt task-specific paradigms and facilitate the creation of models that can undertake new tasks and be used widely in scientific applications across a variety of scientific fields. Features of these models include dynamic task specification, multimodal inputs and outputs, and the integration of scientific knowledge. These characteristics empower the models to expand the boundaries of scientific discovery and drug development, but the evolving nature of such AI models makes it hard to predict their future applications.”