Read time: 5 minutes
- AI is uniquely suited to address specific healthcare challenges: vast amounts of complex data and human variability.
- Potential benefits of healthcare AI include improved diagnostic accuracy, personalized treatment plans, and increased operational efficiency.
- There is no substitute for human connection in healthcare. Artificial intelligence should augment human intelligence, not replace it.
Once reserved for science fiction movies, artificial intelligence (AI) has now entered mainstream industries, including healthcare. The potential benefits of healthcare AI are vast, ranging from improved diagnostic accuracy to personalized treatment plans to increased operational efficiency. However, using AI to improve healthcare is a complex task, and there are significant differences between healthcare AI and its application elsewhere, such as the development of Chat GPT.
AI is Uniquely Suited to Address Specific Healthcare Challenges
Challenge 1: Data Surplus
A growing challenge, literally and figuratively, is the amount of data that healthcare systems, technology, and services generate. The sources are numerous—electronic health records, administrative records, medical imaging, lab tests (e.g., blood, urine), wearables (e.g. activity, sleep), genomics, health surveys, insurance claims, clinical trials, peer-reviewed literature, public health agencies, disease registries, social determinants of health—and produce complex data at a rapid pace. It’s unreasonable to expect patients or providers to make sense of all this data. Instead, we can rely on AI algorithms and machine learning (ML) models for data processing, analyses, and pattern recognition. When healthcare AI can tackle the more complicated, data-focused work, providers and products can proactively respond to patient needs with greater speed and accuracy.
At One Drop, we use artificial intelligence to transform nearly 50 billion data points from millions of members worldwide into health predictions and insights people can use. AI-powered features like eight-hour blood glucose (BG) forecasts simplify healthy decision-making by connecting everyday behaviors with outcomes and offering ongoing guidance to complement the support members receive from their healthcare providers. Research recently published in JMIR Diabetes validated the strength of One Drop’s predictive intelligence, linking BG forecasts to better engagement and diabetes outcomes. One Drop is the first and only glucose forecast provider for people with type 2 diabetes.
Challenge 2: Human Variability
Every human is unique. Our uniqueness, while beautiful, poses a challenge to healthcare. Variations in biology, psychological makeup, personal preferences, lifestyle choices, and past experiences mean what works best for one person won't necessarily work for another. While general guidelines exist for the management of diseases and conditions, a one-size-fits-all approach to health just doesn’t work. AI-powered data analysis and pattern recognition can help providers and healthcare platforms tailor treatment to meet patients' needs. A personalized and dynamic plan can result in better health outcomes and a superior patient experience.
Thanks to real-time health data and predictive intelligence, One Drop Premium delivers what traditional healthcare cannot: a continuous, personalized care experience that helps people better understand their bodies, make informed decisions, and prevent problems before they happen. In addition to AI-powered health predictions and trends, members receive customized learning content (e.g., weight management tools, meal plans) and one-on-one coaching with a certified health professional who can offer feedback and proactive support.
OpenAI, Healthcare AI, and the Human Eye
When prompted, Chat GPT described itself as “a language model developed by OpenAI that uses deep learning algorithms to generate responses to text-based prompts.” While some AI chatbots are capable of responding to questions you may otherwise ask Web MD (e.g., “What are the early symptoms of type 1 diabetes?”), Chat GPT and similar technologies are, in their current form, best suited for relatively straightforward and lower-risk tasks such as customer service, localization (language translation), and optimized search queries.
The prevailing fear is that artificial intelligence will take over everything when the reality is AI should augment human intelligence, not replace it. This is particularly true in healthcare, where human connection and soft skills are paramount, and decisions made by healthcare AI can directly and permanently impact a patient’s health and well-being.
“AI is neither good nor evil. It's a tool. It's a technology for us to use.”
Oren Etzioni, Founding CEO of AI2
Artificial intelligence and the human touch of clinical health coaching are independent, critical components of a data-driven feedback loop for One Drop Premium members. For example, AI-powered insights allow members to view health trends (e.g., glucose will rise in the next two hours) and get actionable advice on what to do next (e.g., take a 10-minute walk). Meanwhile, One Drop coaches—a team of registered nurses, dietitians, and certified diabetes care and education specialists—facilitate behavior change via in-app text chat. With real-time access to member data and their responses to prompts, quizzes, and commitment pledges for healthy behaviors, One Drop coaches can deliver timely content and proactively offer clinical and emotional support.
Building Ethical AI-Based Solutions for Healthcare
There are serious implications for decisions based on healthcare data, making equity a critical component of healthcare AI. While the data on biases brought to the development and design of healthcare AI is limited, emerging research has begun looking into the efficacy and equity of AI-based diabetes interventions and precision medicine—both of which rely on copious amounts of health data. Artificial intelligence affords us speed and convenience, but superior outputs hinge on the quality and diversity of training data. It's essential to prioritize representative data samples and train algorithms, healthcare personnel, and researchers alike to recognize biases and take deliberate steps to be more inclusive.
Another critical component of ethical healthcare AI is transparency. If the everyday person doesn't understand, at a basic level, how their health data is being used, trust can quickly vanish. Therefore, to the best of their ability, healthcare providers, products, and services should educate individuals about how they're using artificial intelligence to improve the healthcare experience without compromising data security.
The application of artificial intelligence in healthcare is complex and nuanced. Still, with proper implementation, human oversight, and suitable safeguards, AI has the potential to revolutionize healthcare, improve outcomes, and transform the lives of millions worldwide.
About the Author
Dan Goldner, Ph.D., MEd, is the executive vice president of advanced technologies research and discovery at One Drop, where he oversees the data science and sensor development teams. Since joining One Drop in 2017, Dr. Goldner has led the development of proprietary AI and machine learning algorithms, including but not limited to the first commercially available blood glucose predictions for people with type 2 diabetes. He holds a Ph.D. from MIT and a BA from Harvard University in physical oceanography.
This article was co-written by Andrea Lagotte.