It’s difficult to open a newspaper or flick through LinkedIn these days without coming across something talking about AI.
Artificial intelligence has well and truly become the buzzword of 2023, but while many sectors have embraced AI with rampant enthusiasm, the life sciences industry has adopted a more cautious approach.
The industry is, after all, very highly regulated, and when your work has direct health and safety implications for patients, you’re not in the business of taking unnecessary risks.
The tide does seem to be turning, however, as artificial intelligence, particularly through the lens of machine learning (ML), is emerging as a transformative tool within the domain.
In this article, we’ll be exploring how AI is starting to be used in the life sciences sector, the power it can have to accelerate advancements and the risks associated with doing so.
AI’s Use in Life Sciences
Drug development is notorious for its length, cost and uncertainty. With the world’s 20 largest pharmaceutical companies collectively investing $139 billion in R&D in 2022, the need to cut costs and solidify success rates is clear to see.
AI is proving to be a game-changer, albeit at this stage predominantly in the early stages of a drug’s life cycle. 90% of AI-assisted drugs currently sit in the development or preclinical stages, so it’s clear that AI’s integration across the entire value chain remains in its infancy.
Tools and platforms powered by AI are assisting scientists in identifying novel biological targets, mapping disease pathways and illuminating complex protein interactions, which could lead to the development of new and more effective drugs and treatments, tailored to the individual needs of each patient.
AI’s power to accelerate drug development can be seen in the case of drug discovery company Insilico Medicine. Using GAN and reinforcement learning, the company was able to create a new drug-like molecule in 21 days and validated it in 25 – a process 15 times faster than traditional discovery methodologies!
In personalised medicine, AI is taking centre stage, developing more accurate diagnostics and tailored treatment plans. AI-driven diagnostics can help clinicians identify diseases earlier and more accurately, while predictive models assess patients’ risk of developing certain diseases, enabling early intervention and prevention.
For example, an AI-powered system achieved a remarkable 90% accuracy in predicting which cancer patients were most likely to respond to immunotherapy. This could be used by doctors to not only identify patients most likely to benefit from treatments but also direct those unlikely to respond more rapidly to alternate therapies.
With approximately 80% of clinical trials delayed or closed due to issues with recruitment, it’s evident they could stand to benefit from AI-powered algorithms. Current recruitment methods fail to bring in the best-suited patients to a trial in time, so there is an opportunity for these algorithms to identify promising candidates and design more efficient trials.
In clinical research, AI can optimise trial designs, accelerate data analysis and enable real-world data studies. Additionally, AI-enabled studies utilising real-world data can augment our understanding of treatment outcomes and improve the generalisability of clinical trial results.
Since the post-COVID-19 preference for decentralised clinical trials, AI has facilitated the monitoring of patients in their own homes, expanding the pool of participants while reducing dropouts, leading to higher chances of success. It is, however, crucial to note that while AI improves processes within clinical trials, it cannot compensate for poor planning and strategies at this stage.
The success of machine learning models hinges on the quality of the data they train on. MELLODDY, a shared federated learning model across nine organisations, exemplifies this. Trained on pharmacological and toxicological data for over 21 million small-molecule drug candidates, this shared model outperformed individual organisation models by up to 20%. This draws attention to the importance of training AI systems on high-quality, reliable and accurate data and highlights that collaboration is the key to future success in life sciences.
AI is not confined to drug development – it’s offering unprecedented visibility into supply chains, automating manual tasks, and supporting better decision-making. AI-powered analytics can identify potential risks and opportunities in supply chains, from disruptions in raw material supplies to manufacturing delays. Automation of manual tasks, such as order processing and vendor management, frees up resources for strategic initiatives, while AI-driven insights support better decision-making in supplier selection and pricing strategies.
Despite AI’s significant advances, it still lacks certain essential capabilities for the life sciences industry, such as strategic planning, conceptualisation and soft skills like empathy. As a result, human ingenuity remains crucial for developing and executing strategies in this patient-centric field.
Risks of AI
It’s important to note that AI is not without risks. One of the biggest concerns is the potential for bias in AI systems. If AI systems are trained on biased data, they can produce biased results. This could lead to misdiagnosis, inappropriate treatment recommendations, and other problems. Smart collaborations, robust training processes and governance are essential to mitigate these risks and ensure responsible AI deployment.
Another concern is the potential for AI systems to be hacked or manipulated, potentially leading to the theft of sensitive data or the disruption of critical life sciences operations. When dealing with such large datasets, there is a serious risk of breaking patient confidentiality unless the correct governance, data security safeguards and compliance programmes are in place.
AI’s integration into life sciences necessitates strict regulation to ensure strict patient safety and ethical practice is met and maintained. Anticipating this, regulators are likely to update policies, requiring the industry to stay compliant.
AI introduces various organisational, operational, and digital challenges to the life sciences industry. Cohesive integration demands a well-defined vision, strategic assessments into the areas it could deliver the highest value and substantial investments in digital infrastructure to ensure continuous improvement. Thorough training for both systems and the workforce is therefore essential to pave the way for a smooth transition to AI operations.
In developing and deploying AI systems in the life sciences, it’s crucial to carefully consider risks. Taking the appropriate steps to mitigate these challenges ensures that AI enhances human health and well-being while avoiding unintended consequences.
Conclusion
Artificial intelligence holds immense promise for enhancing efficiency across many facets of the life sciences sector, from drug discovery to personalised medicine, clinical research and supply chain management. Full-scale adoption of these technologies has, however, been hampered by valid apprehensions regarding uncertainty and risks.
To solidify the industry’s trust in AI systems, then, it’s crucial that governments, private enterprises and healthcare institutions unite in collaborative efforts to ensure systems are interoperable, transparent and unbiased.
Additionally, it’s important to acknowledge that AI’s role should be that of an enhancer of human capabilities, not a replacement for them. A healthy, symbiotic relationship between human expertise and AI innovation is therefore imperative to paving the way for a thriving life sciences industry in which the two exist in harmony.
GET IN TOUCH WITH OUR TEAM TODAY
Looking for comprehensive procurement and supply chain solutions for your life sciences business? Get in touch with 4C Associates to discover how we can help today.
