OR WAIT null SECS
Over the last several years, the pharmaceutical industry’s digital transformation has not only helped improve efficiencies and optimized drug development, but also equipped sponsors with once-unfathomable insights into patient behavior. The AI innovation at sponsors’ fingertips today has the potential to truly revolutionize data accuracy and integrity in every phase of clinical research, helping to more quickly bring life-saving drugs to patients who need them. While the promise of AI is on the horizon, we still have work to do to realize its full potential. As we look forward to the year ahead, we will need to work collaboratively and strategically to improve the quality of bias-free AI and advance the use of novel assessments like digital biomarkers to drive precision medicine both in clinical research and in clinical practice. The following trends will define AI development in the year to come:
With the pandemic shedding light on healthcare disparities, the healthcare industry will be challenged to re-evaluate the equitability of the AI our patient care decisions increasingly rely on. The quality and generalizability of an AI model depend on the diversity of the data sets it is trained on. However, today’s AI developers tend to lack access to highly diverse, large data sets, and often train algorithms on small, homogeneous data samples, which can lead to biases that can skew the outputs. While on a smaller scale, a non-optimized AI algorithm can impact research operations—for example, an algorithm that helps identify optimal investigator sites might lead to a slight skew in the quality of sites chosen to participate. But these biases can also have more harmful consequences and affect patient outcomes, including a sponsor’s ability to contribute accurate, meaningful data around a drug’s safety.
In 2022, we will need to work together to ensure the technology put in the hands of our patients and pharmaceutical companies has the foundation it needs to foster equality. We can do this by focusing on the diversity of training data sets and the generalizability of results, as well as implementing rigorous evaluation processes for AI. When unbiased and reliable, AI tools can reach their full potential in addressing real-world variability and help bridge an increasingly diverse world.
A precise and accurate understanding of a patient’s health and any changes in their response to treatment is imperative to measure the state of a patient’s disease and the efficacy of one’s treatment plan. At the same time, the better we can understand a patient’s individual experience with their disease and treatment, the better we can understand the ideal patient profile for a specific drug. Instead of relying on in-person clinical visits or a patient’s self-reported outcomes, AI-powered tools are increasingly taking on a more important role in tapping into these nuanced patient behaviors and will continue to gain momentum in the year ahead.
For example, video and audio-based digital biomarkers can remotely detect a patient’s subtle responses to treatment that otherwise might be missed. These include behaviors such as their response time, physical movements, and speech patterns, which can be critical indicators of disease progression and how treatment is impacting a patient’s quality of life. Because of their consistency and automation, these tools will increasingly provide the level of sensitivity and objectivity needed to drive precision medicine and guide highly personalized interventions.
These novel assessments represent a paradigm shift in assessing the effectiveness of one’s treatment plan and elevating the integrity of data—and yet, they are largely an uncapitalized resource to date. The proprietary nature of many of these algorithms keeps researchers from exercising them, ultimately stunting their improvement and validation. In 2022 and beyond, we will see a push to build trust in these novel measurements in the public domain through peer review. Open-source platforms that house these algorithms can break down these historical barriers, allow researchers to apply them to their own data sets and jointly contribute to their advancement. By fostering collaboration within the scientific community, digital biomarkers can become a widely recognized, legitimate means to understand disease accurately and objectively.
The past few years have proven that the healthcare and pharmaceutical industries are very capable of modernizing, collaborating, and transforming in ways that we never imagined. More and more, we’re seeing AI’s potential to transform drug development and patient care, and these industries are ripe to embrace AI innovation. By taking a thoughtful approach to AI’s development and deployment with equality at its core, our drug cycles, patients, and genuine understanding of disease will reap the benefits.
Ed Ikeguchi, MD, is the CEO of AiCure