AI has great potential to transform drug development and patient care as we know it – from elevating a clinical trial’s data to speed up research, to supporting clinical decision-making to drive confident care. But, the data mining capabilities that make AI and machine learning so promising are the same ones that could cause an algorithm to do more harm than good. AI is only as strong as the data that it’s fed, and if an algorithm doesn’t have a diverse data foundation, it can perpetuate human biases that could put minority populations at a disadvantage or even a health risk.
As AI and machine learning tools increasingly become a pivotal part of how we research drugs and deliver new treatments, we have a responsibility to prioritize equality in the technology our patients and organizations use. We can help make healthcare a more inclusive industry by rigorously testing algorithms to eradicate biases, and effectively engaging with minority populations to improve their participation in clinical research in the first place.
Alleviating Bias in AI & Machine Learning
Since AiCure’s beginnings, ensuring diversity in our algorithms’ data has been a grounding principle in all that we do. For the sake of the patients we serve and the practice of good science, building algorithms using diverse data is not just the icing on the cake or something that naturally happens over time – it’s a necessary, deliberate framework for developing quality AI. For AiCure, this took trial and error, a lot of time, and concentrated effort – but it was worth it to ensure our tools work with all patients. As an industry, we need to normalize going back to the drawing board when algorithms don’t perform as planned, rather than introducing patients to inadequate and potentially harmful algorithms.
Similar to how new drugs go through years of clinical trial testing with thousands of patients to determine adverse events, a similar checks and balances process for AI is needed to understand if it falls short in real-world scenarios. When this technology is governed sufficiently, it holds significant potential for automating processes across all industries and taking innovation to new heights.
Boosting Diversity in Clinical Trials
While the equitability of the tools our patients use is critical, the diversity of the patient population who use them is equally important. Because everyone reacts differently to drug therapies, a lack of diversity in clinical trials can have serious implications for underrepresented populations once that drug goes to market.
For many reasons, minority populations are often skeptical about clinical trials’ intentions. Working with advocates and role models who are trusted within their communities and equipping them with resources to educate patients about clinical trials can help alleviate these suspicions. Pharmaceutical companies and clinicians can also help build trust in the mission of a trial by engaging with patients in locations where there is trust, whether it’s a local barber shop or place of worship. Once a patient is enrolled in a trial, technology that first, works for them regardless of their appearance, and second, offers open lines of communication with a doctor they trust, can help maintain consistent engagement and build affinity for the research.
Progress Takes Time
There’s a lot of work to be done to make healthcare – an industry historically riddled with disparities – more equitable. But, change is on the horizon. As biases in AI and machine learning receive more attention, scrutiny around how algorithms are built will help create a new standard for developing these tools. And, more and more clinicians are prioritizing working with pharmaceutical companies that run equitable, diverse trials, setting the bar higher for clinical research. Awareness of the problem, coupled with foundational leg work and collaboration, will help advance equality in healthcare.
