The world of pharmacy is rapidly evolving, with new and growing technology continually shaping its future. One striking example is how the same technology capable of writing a college essay in under five minutes can also sift through thousands of codes to identify optimal chemicals for drug development. This technological advancement has opened up exciting avenues for researchers to explore diverse sources of medication.
One significant breakthrough is in pharmacogenomics (PGx) testing. This innovative approach allows individuals to receive a comprehensive analysis of their genetic composition and how it affects their compatibility with various medications. Many people are unaware that genetic differences can significantly influence how well a medication works and the potential side effects it might cause. A PGx test can reveal how well your current medication aligns with your genetic profile and guide future prescriptions. However, like many advanced technologies, PGx testing comes with a hefty price tag, often not covered by insurance. Interestingly, AI has already begun to reduce costs in drug research, raising the question: could it do the same for PGx testing?
In the context of PGx testing, AI utilizes machine learning (ML) to engage in deep learning (DL), which creates artificial neural networks (ANN). These neural networks play a crucial role in advancing PGx testing. Currently, AI applications in pharmacogenomics include pattern recognition, or variant calling, and data analysis, where AI can detect missed patterns or correlations. The ability to use technology in this way is vital for accurately predicting the best medication for a patient’s specific genetic makeup, enhancing the efficacy of treatments, and providing precise predictions for clinical applications.
Furthermore, AI and machine learning are being developed to detect adverse drug events through the analysis of clinical narratives in electronic medical records. This represents a significant advancement, as it allows for more personalized therapy, increasing the safety profile of treatments and boosting patient compliance. Compliance is a persistent issue, particularly due to adverse drug reactions or undesired symptoms. Therefore, using AI to detect these events is crucial for developing PGx testing. Additionally, AI can analyze gene expression and its impact on drug uptake and receptor interactions.
In practical terms, AI could determine the appropriate dosage of medications, such as statins, by analyzing specific genes like SLCO1B1, thereby avoiding adverse events or symptoms. Interestingly, AI systems can even create “digital twins” of patients, replicating their physiological states. This allows healthcare providers to test different treatment strategies and predict how changes in medication might affect a patient’s health without exposing them to the potential adverse effects.
Most importantly, AI can address issues of time and cost by leveraging consumer DNA testing and whole-genome sequencing in PGx. This means genetic testing companies and labs could expand their services into PGx without needing to collect additional samples or develop new tests, saving both time and money.
At Baylor College of Medicine, researchers are exploring a different application of AI. Their technology, similar to ChatGPT, is designed to answer specific questions about PGx test results for patients and physicians. The study focuses on pharmacogenomic testing for statins, aiming to determine whether patients are genetically predisposed to better or worse responses to different statin medications. The AI technology at Baylor uses Retrieval Augmented Generation (RAG), sourcing knowledge from the latest Clinical Pharmacogenetics Implementation Consortium data and publications. With impressive accuracy and relevancy—81% and 62%, respectively—Baylor’s AI chatbot rivals, if not surpasses, ChatGPT. However, there are limitations, as the technology is still in development. Researchers emphasize that it is not yet ready for clinical use, struggling to recognize biomedical terms and unable to communicate in the precise language used by genetic counselors. Additionally, ethical, safety, and regulatory concerns must be addressed before clinical implementation.
While AI holds tremendous promise in the pharmacogenomics community, much work remains before it can be fully integrated to increase efficiency. Nevertheless, it points toward a future where healthcare is more personalized, enabling patients to receive tailored treatment and therapy.
- TriHealth. (n.d.). Pharmacogenomics. Retrieved July 17, 2024, from
https://www.trihealth.com/services/precision-medicine-and-genetic-services/our-services/pharmacogenomics - Halder, A. (n.d.). Pharmacogenomics and the AI-powered future for managed care. Tata Consultancy Services. Retrieved July 17, 2024, from https://www.tcs.com/what-we-do/industries/healthcare/white-paper/pharmacogenomics-ai-healthcare-personalized treatment#:~:text=AI%2Dpowered%20algorithms%20can%2Dbe,of%20the%20body%20was%20targeted
- Bannister, P., & Kohler, A. T. (2024, April). Revolutionizing primary care: The role of pharmacogenomics and AI in personalized medicine. MedCity News. Retrieved July 17, 2024, from https://www.trihealth.com/services/precision-medicine-and-genetic-services/our-services/pharmacogenomics
- Adamson, B., Makady, A., Sarri, G., Mohamed, O., Babar, Z., & Dawoud, D. M. (2024). Editorial: Novel methods and technologies for the evaluation of drug outcomes and policies. Frontiers in Pharmacology, 15. doi:10.3389/fphar.2024.1396034. Retrieved from
https://www.frontiersin.org/journals/pharmacology/articles/10.3389/fphar.2024.1396034/full - Baylor College of Medicine. (2023, March 14). Using generative AI assistant to interpret pharmacogenetic test results. Retrieved July 17, 2024, from
https://www.bcm.edu/news/using-generative-ai-assistant-to-interpret-pharmacogenetic-test-results
