Today we have the pleasure of speaking with Dr Brandon Allgood, a renowned entrepreneur and expert in applying machine learning and large-scale computational methods to advance human health. With a diverse background in physics, computational biology, computational chemistry, and AI, Brandon has made significant contributions to the fields of drug discovery, development, and clinical trials. As the former Chief AI Officer at Valo Health and co-founder and CTO of Numerate, he has led teams in developing cutting-edge ML-driven platforms for optimizing drug design and clinical trials. Let’s delve into his remarkable journey and insights at the intersection of technology and healthcare.
- Thank you for taking time to share your knowledge and experience about Artificial Intelligence in drug discovery and development with our audience. Can you start by telling the audience about yourself?
As you already stated, my name is Brandon Allgood. As for my education, my PhD is actually in cosmology. I had wanted to be a cosmologist since I was 12, which no one around me really understood. So much so, I made it a point to meet Stephen Hawking when I was 18. As a graduate student I wrote software for simulations of galaxy and dark matter halo formation that ran on some of the world’s biggest supercomputers. I also wrote machine learning algorithms to analyze images of galaxies from the Hubble Space Telescope and gamma-ray traces from the Fermi Gamma-Ray Satellite.
Toward the end of my graduate studies however I decided to leave academia for personal reasons. After grad school I landed squarely in Silicon Valley. I was in the interview process for jobs at NASA and Google, when I found an interesting start-up working on AI in chemical design. I took the job at the start-up, because I realized that my passions were at the intersection of machine learning, large-scale computing, and science. The start-up allowed me to do those things. I also found it to be most like academia in terms of the freedom to pursue new ideas.
I had always had an entrepreneurial streak in me, so over the last 18 years in industry I have focused my efforts on building companies with AI at their core. Unfortunately the first start-up (Pharmix) went the way of most start-ups and ran out of money. However after that, I helped found and run two other start-ups (Numerate and Valo Health) and a non-profit (the Alliance for AI in Healthcare). My focus throughout these 18 years has been on pushing the boundaries and building platforms that allow us to better use data and compute to reduce cost and time, and increase probability of success of bringing drugs to market to help patients. Patients are waiting, but disease is not.
- Can you tell our audience about your Obsidian Scientific? What problem is it solving & how?
Over the years, I have worked in most areas of drug discovery and development and have just become more and more excited about the impact AI can have on healthcare. I now run a consulting firm, Obsidian Scientific, LLC, where I help pharma, biotechs, and venture capital firms understand, strategize, and implement AI approaches to improve healthcare.
With the bigger pharma clients, I help them with broad strategy and architecture to deliver short- and long-term ROI. With VCs, I advise them on potential investments, I help their portfolio companies, and I help them build out common infrastructure, incubate and launch new ideas around AI in healthcare. With private and public biotechs, I generally help them with interim technical management, architectural design of their computational support systems, and both AI and technical hiring strategies.
- How can AI be applied to drug discovery and development to improve the efficiency and effectiveness of the process?
Artificial Intelligence (AI) has the potential to revolutionize drug discovery and development by significantly improving efficiency and effectiveness. One key application of AI is in the early stages of the drug discovery process, where it can assist in identifying potential drug candidates, help design proteins, and aid in cell therapies, to name a few things. AI algorithms can analyze vast amounts of data, including chemical structures, protein structures, biological pathways, multi-omics data, and clinical data, to predict the likelihood of a therapy being successful. This enables researchers to focus their efforts on the most promising programs, saving time and resources.
In addition, AI can be used to simulate and predict the drug-target interactions, helping researchers understand the mechanisms of action and optimize the drug design process. Machine learning algorithms can learn from large datasets to predict, not just the affinity between a drug molecule and its target, but also a broad range of properties a drug needs to have in order to be successful. Due to the high-dimensional nature of drug development AI can often produce better outputs at a higher rate than traditional methods.
Furthermore, AI can aid in the optimization of clinical trials, which is a critical step in drug development. By analyzing patient data and clinical trial results, AI algorithms can identify patterns, predict patient responses, and optimize the trial design. This can lead to more efficient and targeted clinical trials, reducing costs and accelerating the overall drug development process.
Overall, AI has the potential to transform drug discovery and development by accelerating the identification of potential drugs, improving our understanding of target-disease relationships, and optimizing clinical trials.
- What are the main challenges and limitations of using AI in drug discovery and development?
While AI holds great promise in drug discovery and development, there are several challenges and limitations that need to be addressed. One challenge is the availability and quality of data. AI algorithms can require large and diverse datasets to learn from, but accessing such datasets can be difficult due to privacy concerns and data sharing limitations. Additionally, the quality and consistency of the data can vary, which can affect the accuracy and reliability of AI predictions.
Another challenge is the interpretability of AI models. Deep learning algorithms, which are commonly used in AI applications, can be highly complex and difficult to interpret. This makes it challenging for researchers to understand the underlying reasons behind AI predictions, which is crucial for gaining insights and making informed decisions in drug discovery and development.
Furthermore, there is a need to ensure the ethical use of AI in drug development. As AI algorithms become more sophisticated, there is a risk of biases and unfairness in decision-making. While this is already a problem, the use of AI will supercharge the problem. It is important to develop robust frameworks and guidelines to address these ethical concerns and ensure that AI is used in a responsible and unbiased manner.
Lastly, the integration of AI into existing drug discovery and development workflows can be a challenge. Incorporating AI tools and technologies into established processes requires careful planning and in many cases the established processes should be completely reworked, not to mention the development of new infrastructure and a need to retrain personnel. While the pharma/biotech industry is not new to innovation, many of those innovations haven’t panned out or have taken longer than expected to pay off. That combined with the highly regulated nature of the work we will continue to see adoption be slower in the pharmaceutical industry.
Addressing these challenges and limitations will be essential to fully harness the potential of AI in drug discovery and development.
- How can AI improve the process of predicting the safety and toxicity of potential drugs?
The prediction of safety and toxicity is a crucial aspect of drug development, and AI can play a significant role in improving this process. AI algorithms can analyze large datasets of chemical and protein structures, biological pathways, genomic, real-world data and toxicity data to help predict the safety and toxicity profiles of potential drug candidates.
One way AI can enhance safety prediction is through the development of predictive models that can identify potential toxicities based on chemical structure, genetics, and other molecular properties. By training these models on large datasets of compounds and their known toxic properties and known induced adverse drug reactions, AI algorithms can learn to recognize patterns and make accurate predictions. This can help researchers identify potentially toxic compounds early in the drug development process and avoid costly and time-consuming experiments.
AI can also aid in the prediction of drug-drug interactions, which can have significant implications for drug safety. By analyzing data on drug metabolism and known interactions, AI algorithms can identify potential interactions between multiple drugs and predict their impact on safety and efficacy. This can help researchers optimize drug combinations and minimize the risk of adverse effects.
Furthermore, AI can contribute to the field of pharmacovigilance, which focuses on monitoring and assessing the safety of drugs in real-world settings. By analyzing large-scale healthcare databases and social media data, AI algorithms can detect potential safety signals and adverse events associated with specific drugs. This can enable early detection of safety issues and facilitate timely interventions.
Overall, AI has the potential to improve the prediction of safety and toxicity in drug development by leveraging large datasets, identifying patterns, and making accurate predictions. This can help researchers prioritize the development of safer and more effective drugs.
- How can AI assist in the repurposing of existing drugs for new therapeutic indications?
Drug repurposing, also known as drug repositioning, is the process of identifying new therapeutic uses for existing drugs. AI can play a crucial role in accelerating this process by leveraging datasets and advanced computational techniques.
One way AI can assist in drug repurposing is through the analysis of existing data on drug-target interaction networks, disease pathways, and clinical outcomes. By integrating and analyzing these diverse datasets, AI algorithms can identify potential connections between drugs and diseases that may have been overlooked through traditional methods. This can uncover new therapeutic indications for existing drugs, potentially saving time and resources compared to developing new drugs from scratch.
Another approach is to utilize AI algorithms to analyze the vast amount of scientific literature and biomedical databases. By extracting and analyzing information from these sources, AI can identify potential associations between drugs and diseases, even if they were not originally intended for that purpose. This can provide valuable insights and guide further experimental validation.
Furthermore, AI can assist in the identification of patient subpopulations that may benefit from existing drugs. By analyzing large-scale real-world patient data, AI algorithms can identify patterns and correlations between patient characteristics and treatment outcomes. This can help identify subgroups of patients who may respond well to a particular drug, even if it was originally developed for a different indication.
Overall, AI has the potential to accelerate the process of drug repurposing by leveraging diverse datasets, analyzing scientific literature, and identifying patient subpopulations. This can lead to the discovery of new therapeutic uses for existing drugs and potentially expedite the delivery of effective treatments to patients.
- Can you explain the concept of “virtual screening” and how AI can enhance this process in drug discovery?
Virtual screening is a computational approach used in drug discovery to identify potential drug candidates that are likely to bind to a specific target of interest as well as have the other properties needed to be a true drug. It involves the screening of large chemical libraries, databases, and generated molecules from generative models using computational models to predict the binding affinity between a drug molecule and its target.
AI can greatly enhance the virtual screening process by leveraging machine learning algorithms and large datasets. Traditional virtual screening methods rely on the knowledge of known ligands and their interaction with the target protein. However, AI can go beyond this by learning from vast amounts of data and identifying novel ligands that may have not been considered before.
Through machine learning, AI algorithms can be trained on data from various sources, including known ligands, compound libraries, and structural information of the target protein. These algorithms can then predict the binding affinity of new compounds to the target protein, prioritizing those with the highest likelihood of binding.
AI can also aid in the optimization of virtual screening by incorporating additional information, such as drug metabolism and toxicity data, to refine the predictions. This can help prioritize compounds that not only have a high binding affinity but also exhibit desirable drug-like properties.
Furthermore, AI algorithms can learn from experimental data on compound-target interactions and other properties to continuously improve their predictions using active learning. By incorporating feedback from high-throughput screening assays or other experimental techniques, AI models can refine their predictions and increase their accuracy over time.
Overall, AI enhances the virtual screening process in drug discovery by leveraging machine learning algorithms, large datasets, and multiple sources of information. This can lead to the identification of novel drugs with a higher likelihood of success.
- How does AI contribute to the optimization of drug design and formulation?
AI plays a crucial role in optimizing drug design and formulation by leveraging computational modeling, machine learning, and big data analytics. It enables researchers to explore a vast chemical and protein space, predict properties, and guide the design of more effective and safer drugs.
One way AI contributes to drug design optimization can be through the generation of novel chemical structures. By training machine learning algorithms on large databases of known compounds, AI can learn patterns and rules that govern chemical reactivity and generate new molecules with desired properties. This can significantly accelerate the process of lead identification and optimization.
Furthermore, AI can assist in predicting and optimizing the physicochemical properties of drug candidates. Through computational modeling, AI algorithms can estimate properties such as solubility, permeability, and stability. This information is crucial in guiding the selection and optimization of drug candidates, as it influences their absorption, distribution, metabolism, and excretion (ADME) properties.
AI can also aid in the formulation of drugs by predicting their behavior in different delivery systems. By analyzing the physicochemical properties of both the drug and the formulation components, AI algorithms can predict factors such as drug release kinetics, stability, and compatibility. This can help researchers optimize the formulation process and develop drug delivery systems with improved efficacy and patient compliance.
Additionally, AI can assist in the optimization of drug dosage and regimen by analyzing patient data and predicting individual responses. By considering factors such as genetics, demographics, and disease characteristics, AI algorithms can personalize treatment regimens and optimize dosing strategies, leading to improved therapeutic outcomes.
In summary, AI contributes to the optimization of drug design and formulation by generating novel chemical structures, predicting physicochemical properties, guiding formulation development, and personalizing treatment regimens. This can lead to the development of more effective and patient-centric drugs.
- How can AI be used to improve the efficiency and success rate of clinical trials?
AI has the potential to significantly improve the efficiency and success rate of clinical trials by optimizing various aspects of the trial design and execution process.
One way AI can enhance clinical trial efficiency is through patient recruitment and selection. By analyzing electronic health records, genomics data, and other patient data sources, AI algorithms can identify eligible patients who meet specific inclusion criteria. This can help accelerate the recruitment process, reduce dropout rates, and ensure the enrollment of a more diverse patient population.
AI can also aid in the design of clinical trials by optimizing sample size calculations and statistical analysis plans. By leveraging machine learning algorithms, AI can analyze historical trial data and simulate different scenarios to identify the optimal sample size, treatment arms, and endpoints. This can help researchers design more efficient and informative trials, reducing costs and time.
Furthermore, AI can assist in the monitoring and analysis of clinical trial data in real-time. By analyzing data from wearable devices, electronic diaries, and other sources, AI algorithms can detect potential safety issues, identify trends, and predict patient outcomes. This can enable early intervention, minimize risks, and improve the overall success rate of the trial.
Additionally, AI can contribute to the analysis and interpretation of clinical trial results. By leveraging advanced analytics and machine learning algorithms, AI can identify subgroups of patients who may benefit the most from the treatment, explore potential biomarkers, and generate novel insights. This can facilitate the identification of patient stratification strategies and inform future trial designs.
Overall, AI has the potential to enhance the efficiency and success rate of clinical trials by optimizing patient recruitment, trial design, data analysis, and interpretation. This can lead to faster and more informative trials, ultimately accelerating the delivery of safe and effective treatments to patients.
- How can AI help in the identification of new drug targets and the discovery of novel mechanisms of action?
AI can play a crucial role in the identification of new drug targets and the discovery of novel mechanisms of action by leveraging large-scale data analysis and advanced computational techniques.
One way AI can assist in identifying new drug targets is through the analysis of multi-omics data, such as genomics, proteomics, and transcriptomics. By integrating and analyzing these diverse datasets, AI algorithms can identify patterns, correlations, and potential therapeutic targets that may have been overlooked through traditional methods. This can lead to the discovery of novel targets for a wide range of diseases.
Furthermore, AI can contribute to the prediction of protein structure and function, which is essential for understanding the mechanisms of action of potential drug targets. By analyzing known protein structures and sequences, AI algorithms can predict the 3D structure of proteins and their interactions with ligands. This can provide valuable insights into the binding sites, molecular interactions, and potential druggability of the target proteins.
AI can also aid in the identification of novel mechanisms of action by analyzing large-scale chemical and biological data. By integrating chemical structure data, biological activity data, and other relevant information, AI algorithms can identify patterns and correlations that can lead to the discovery of new modes of action. This can expand the repertoire of therapeutic approaches and open up new avenues for drug development.
Additionally, AI can contribute to the identification of drug combinations and synergistic effects. By analyzing data on drug-target interactions, gene expression profiles, and known drug combinations, AI algorithms can identify potential combinations that may have synergistic effects or overcome drug resistance. This can improve the efficacy and success rate of drug development efforts.
In summary, AI aids in the identification of new drug targets and the discovery of novel mechanisms of action by analyzing omics data, predicting protein structure and function, identifying patterns in chemical and biological data, and guiding the exploration of drug combinations. This can facilitate the development of innovative and effective treatments for various diseases.
- How can AI help in the process of personalized medicine and individualized treatment selection?
AI has the potential to revolutionize personalized medicine by leveraging patient data, genomics information, and advanced analytics to guide treatment selection and optimize patient outcomes.
One way AI can assist in personalized medicine is through the prediction of disease progression and treatment response. By analyzing large-scale patient data, AI algorithms can identify patterns and correlations between patient characteristics, disease progression, and treatment outcomes. This can help predict individual patient trajectories and guide treatment decisions, enabling proactive and personalized interventions.
Furthermore, AI can aid in the identification of patient subpopulations that may respond differently to a particular treatment. By analyzing genomic data, clinical data, and treatment outcomes, AI algorithms can identify genetic markers, biomarkers, and other factors that influence treatment response. This can enable the identification of subgroups of patients who may benefit the most from a specific treatment, guiding personalized treatment selection.
AI can also contribute to treatment optimization by analyzing data on drug-drug interactions, comorbidities, and patient preferences. By considering multiple factors, AI algorithms can generate treatment recommendations that are tailored to the individual patient’s needs and preferences. This can improve treatment adherence, minimize adverse effects, and enhance patient satisfaction.
Additionally, AI can assist in the monitoring and management of patients through remote monitoring and telemedicine. By analyzing data from wearable devices, electronic health records, and patient-reported outcomes, AI algorithms can detect changes in patient health status, predict exacerbations, and provide timely interventions. This can enable early intervention, reduce hospitalizations, and improve overall patient outcomes.
In summary, AI contributes to personalized medicine by predicting disease progression, identifying patient subpopulations, optimizing treatment selection, and enabling remote monitoring. This can lead to more effective and patient-centric healthcare, ultimately improving individual outcomes.
- What are the future prospects and challenges of AI in drug discovery and development?
The future prospects of AI in drug discovery and development are promising, but there are several challenges that need to be addressed to fully realize its potential.
One future prospect is the integration of AI with other emerging technologies, such as robotics and automation. By combining AI algorithms with robotic platforms, drug discovery processes can be further accelerated and streamlined. Automated synthesis, high-throughput screening, and robotic experimentation can be guided and optimized by AI, leading to faster and more efficient drug development.
Another future prospect is the use of AI in precision medicine and the development of personalized therapies. As genomics and other omics technologies become more accessible and affordable, AI can play a crucial role in analyzing and interpreting these large-scale datasets. By integrating genomic information with clinical data, AI algorithms can guide treatment selection, predict treatment response, and optimize patient outcomes.
Furthermore, the development of AI-driven drug design tools and platforms holds great promise. By leveraging machine learning algorithms and large-scale datasets, AI can assist in the generation of novel chemical and protein structures, prediction of drug-target interactions, and optimization of drug properties. This can significantly accelerate the lead identification and optimization process, leading to the development of more effective and safer drugs.
However, there are challenges that need to be addressed to fully exploit the potential of AI in drug discovery and development. These include the availability and quality of data, the interpretability and explainability of AI models, the ethical use of AI in decision-making, and the integration of AI into existing workflows and infrastructure. Addressing these challenges will require collaboration between academia, industry, and regulatory agencies to develop robust frameworks, guidelines, and standards.
In summary, the future prospects of AI in drug discovery and development are promising, with potential applications in robotic automation, precision medicine, and drug design. However, addressing the challenges related to data, interpretability, ethics, and integration will be crucial to fully harness the potential of AI in the pharmaceutical industry.
Provide Contact Details, so the audience can reach you.
- Business website: https://obsidianscientific.com
- Business email: allgood@obsidianscientific.com
- Business Social Media link: Obsidian Scientific – Linkedin
- Personal Social Media Link: Brandon Allgood – Linkedin
