Since 2011, when Watson won Jeopardy, the promises of data, analytics, and AI have grown to a crescendo. If we can’t watch a sporting event without a big tech company telling us how AI improves athlete safety and how to beat the house with a betting app, the question the pharmaceutical industry needs to answer is the following: what is hope and what is hype? when it comes to improving patient outcomes and R&D productivity?
To better understand the challenge, one must understand that, depending on the functionality, there are very different attributes, from practical to theoretical. For example:
- Responsive machine AI can synthesize and analyze large data sets to make an assessment or recommendation. Think about search engines and viewer recommendations from streaming services. However, he has no memory.
- Limited-memory AI can research past events and evaluate the results to make predictions.
- Theoretical AI encompasses advanced concepts such as theory of mind and self-aware AI.
From a capabilities perspective, AI that performs very specific tasks within a subset of cognitive abilities is sometimes called narrow artificial AI. ChatGPT would fall into this category due to its reliance on a single task, namely text chat.
Sponsor companies and others are already beginning to use AI to complement their data quality monitoring approach. Additional potential applications include:
- Reduce the time needed to identify targets during preclinical drug discovery, which otherwise takes several months.
- Analytical tools for site selection for clinical trials.
- ML, AI and augmented intelligence are used to gather insights into the volumes of data collected for marketing and marketing purposes.
Although clinical development lags slightly in the adoption of new technologies, the industry is reaching an inflection point. In healthcare, it is important to evaluate tools that can improve our ability to provide medications to patients who need them. It is therefore not surprising to see the considerable investment and enthusiasm generated by the evolution of this AI. Paradoxically, since the health and well-being of patients are at stake and industrial research is highly regulated for the same reason, it is understandable to observe both confusion and concern about the ability to use AI ethically and appropriately. This concern is driven by the inability to see “under the hood,” so to speak, to understand the accuracy of forecasts and the details related to the data and data quality that support them.
Can humans and machines team up to give us more hope and less hype?
One of the persistent challenges that has made improving cycle times difficult is recruiting. Although AI was successfully used to identify more patients and more sites after protocol design, it did little to reduce screening failure rates, which continue to hover at unacceptable rates of 40 to 90% depending on the published series. Rather than using AI to find more patients for a protocol who might not be representative of the population, companies could guide the design of a protocol that reflects disease areas of interest, thereby reducing screening failure rate and accelerating throughput. More importantly, the results could be more widely applicable to the population of interest.
Some companies use AI tools to automate data aggregation and leverage analytics capabilities. To improve the quality of data used to train AI platforms, improve accuracy and reduce hallucinations, the pharmaceutical industry may need to commit to the adoption of data standards and common data models to make affordable, reliable and scalable. This could enable earlier and more effective detection of operational or clinical risks.
Over the next few years, if companies move away from using AI to deploy trials and find patients to support better trial design, clinical trial protocols will look very different as their eligibility improves – and , therefore, patient recruitment and representativeness (i.e., diversity, equity) will also improve.
Paving the way forward
Although AI has many potential benefits, several things need to happen before it can be fully harnessed for drug development:
- Strengthened data management – In recent years, the industry has recognized the need for good data stewardship and management, both for clinical and operational data. Steps already taken to ensure that the data is in order will need to continue. This will reduce the effort and cost of data acquisition and refocus attention on the information that constitutes true value.
- Balanced benefits and risks – Do the benefits of AI outweigh its risks? Generally speaking, companies are still in the evaluation phase, but a company can examine the benefit/risk balance on a case-by-case basis. Companies may question the risks of adopting specific AI tools in specific drug development processes. A case where the risk is low – like using generative AI to create a basic consumer flyer, for example – could be a good AI “learning opportunity”. The risk may be too high in other cases, for example when making recommendations to patients. This requires partnering with experienced and trained human experts.
- Commitment to sharing and transparency – Processes must be in place to evaluate AI technologies, prove their accuracy, and monitor their performance. Furthermore, the technologies themselves cannot be black boxes. AI technology solution providers need to create transparency about how they work. Likewise, AI solution providers, users, and others must be willing to share what works and what doesn’t. The saying goes: “Success has many fathers; failure is an orphan,” but businesses won’t get very far if they aren’t willing to try new things and sometimes fail.
Biopharmaceutical organizations employ significant safeguards whenever they find new ways of doing things – and AI is just one new tool. Like any other tool, understanding the right problem to apply it to makes the difference between success and failure. Solving a problem is never just about technology; it’s always about people, processes and technology.
Bend over to change
Every individual within the biopharmaceutical ecosystem plays a role in the evolving use of AI to modernize R&D. Each of us can participate by:
- Educate us. Valuable starting points include the U.S. Food and Drug Administration (FDA) discussion papers focused on AI in drug development And medical productsas well as a reflection paper published by the European Medicines Agency (EMA).
- Collaborate with health authorities find a path forward that improves outcomes, success rates and mitigates risk
- Understand the right questions ask.
- Identify the appropriate risk/benefit approach for our areas of product development and expertise.
- Use low-risk opportunities as a “learning laboratory” from which to build.
Businesses are rightly cautious because public welfare is at risk, but they can find ways to balance that risk with the potential benefits of AI. If biotechs, regulators, policymakers, healthcare practitioners, and technology companies can align on improving patient outcomes as their primary goal, we will write about improved development programs through AI in five years.
Photo: metamorworks, Getty Images
Rob DiCicco brings nearly 30 years of pharmaceutical R&D experience to his role as vice president of portfolio management at a nonprofit industry consortium TransCelerate BioPharma Inc.. There, he is responsible for implementing initiatives related to digital transformation, clinical content and reuse, pragmatic testing and real-world data. His current areas of interest include clinical trial design, clinical operations, protocol quality, and research ethics. Rob received his Doctor of Pharmacy degree from the University of the Sciences in Philadelphia.
This message appears via the MedCity Influencers program. Anyone can post their views on healthcare business and innovation on MedCity News through MedCity Influencers. Click here to find out how.
