AI in healthcare: Vast potential, ethical questions

Artificial intelligence (AI) continues to grow in sophistication, largely due to advances in machine learning (ML). However, there are still critical questions that need to be answered.

Machine learning has close ties to predictive analytics. Both can be powerful tools for uncovering insights and identifying patterns in large amounts of data. These capabilities could serve the healthcare sector quite well, particularly when you consider that 30% of all data generated worldwide comes from healthcare alone.

However, AI in the healthcare industry is still in its relative infancy in numerous areas, often relegated to managing medical records or automating repetitive, mundane tasks. Of course, neither of those things lacks value, but moving toward greater industry-wide adoption has the potential to solve the “triple As” of healthcare: accessibility, affordability and accuracy. Explainable AI has even more potential: It can help institutions better find correlations through data and improve diagnostics.

Consider mental disorders. For the past 20 to 30 years, there’s been surprisingly little progress in the field of mental disorders. Healthcare providers often don’t always know what triggers certain mental disorders in different people. Mental disorders are, by their nature, highly personalized. Fortunately, the use of explainable AI presents an opportunity to find a correlation between data points, allowing physicians to offer more personalized diagnostic results.

Explainable AI can move the healthcare industry beyond the “black box” in ML, helping users uncover and understand the correlations presented to them. It offers customization in everything, from treatments to care delivery, and it is the direction healthcare has been headed for some time now. It’s what patients want — and deserve. It also makes healthcare workers much more efficient.

Embracing the opportunity of AI in healthcare

As AI adoption increases across the healthcare industry, repetitive work will obviously be less and less of an issue. Medical coding alone could become much more efficient with the addition of AI capabilities. Cataloging the unique reasons for a patient’s visit takes a lot of time. Advances in AI, however, are helping not only coding systems identify and validate codes, but also coders themselves make better sense of unstructured data.

Medical imaging, too, could experience vast improvements with AI and ML. As it stands, physicians review and label many images each day to arrive at diagnoses. Technology can now analyze medical images to help detect and diagnose certain conditions. As a result, physicians can focus on early intervention and treatment rather than review. They’re also able to see more patients, which improves access to care.

On the pharmaceutical side, you’ll find AlphaFold, an AI system developed by Google’s DeepMind. Using this AI tool helps scientists better predict the structure of protein folding, which means they could move on to the drug development phase much faster. This has the potential of bringing life-saving medications to the market at speeds once thought impossible.

Understanding the ethical considerations around patient data

Turning to the ethical considerations of AI in the context of patient data, many healthcare organizations question where to draw the line — and what the implications are of using patient data to improve care. These organizations are responsible for managing, storing and securing often very sensitive information.

HIPAA has established baseline requirements, but the key is understanding the value of the data and the technology used to track, monitor, capture, analyze and protect patient information. Any policy involved with patient information should include accessibility controls and risk assessments (that is, identifying potential weak points in the system).

When it comes to data privacy, attention should turn to the guardrails around data. When using patient data, you need to enable some sort of alarm. After all, that information could tell the whole story of a patient’s life. It’s important to get controls in place to allow for the isolation of data. Such measures can ensure that an organization is using the technology and patient data in a good cause.

Another key ethical concern is the bias that can arise with data collection and usage. If you have biased data, the algorithm will become biased as well. The information available to the organization won’t likely represent the community as a whole. It’s critical to have diverse coverage. It’s just as crucial to have technology in place that can categorize and use such diverse information.

On the one hand, new technology is enabling the healthcare industry to use AI and data to cure many diseases — an important advancement, no matter how you look at it. At the same time, that same data can potentially improve patients’ well-being.

With the help of technology, healthcare professionals can slice and dice the information to better monitor and prevent serious health conditions. If the healthcare industry can work around the hurdles and enable AI to do more preventable and early intervention work, it’s entirely possible to offer people a higher quality of care and life.

Lu Zhang is founder and managing partner of Fusion Fund. A renowned Silicon Valley investor and serial entrepreneur in healthcare, Zhang was recently selected as one of the best 25 female early-stage investors by Business Insider.