5 Takeaways for the Future of AI in Healthcare

Artificial intelligence is moving at light speed. How is the healthcare industry implementing this game-changing technology without putting patients — and their data — at risk?

Members of the Array Insights team joined the Fatty Liver Foundation (FLF) to tackle this topic. FLF is a leading patient advocacy organization focused on nonalcoholic fatty liver disease (NAFLD) and it’s more advanced form, nonalcoholic steatohepatitis (NASH) — conditions that affect more than 1.66 billion people around the world.

The Array and FLF panelists discussed how artificial intelligence is shaping the world of clinical research and patient advocacy during FLF’s webinar Helpful or Hurtful? The Future of AI In Healthcare.

The webinar featured four experts, each with a unique perspective:

• Anne Kim, Co-Founder and CEO, Array Insights used her graduate work at MIT as inspiration for creating Array Insights;
• Adam Hall, Senior Machine Learning Scientist, Array Insights, holds a Master’s degree in advanced security and digital forensics, and recently submitted his Ph.D. thesis in privacy-preserving machine learning;
• Wayne Eskridge, Co-Founder and CEO, FLF has worked in software and electronics for 50 years and was first diagnosed with liver disease in 2010;
• Neeraj Mistry, Chief Medical Officer, FLF is a global and public health executive and physician with 20+ years of experience in non-profits, business, academia, and multilateral organizations.

 

Let’s review five key points from the panelists:

1.) With proper oversight, AI will be an unambiguous boost to healthcare.

Eskeridge’s own diagnosis of cirrhotic NASH inspired the creation of FLF, which seeks to address an acute need for patient-facing educational resources.

As someone with five decades of experience in the world of tech, Eskridge expressed optimism about the potential of AI in healthcare. He acknowledged that many people feel confusion or apprehension about AI’s ultimate impact, but see more good than harm in AI’s ability to accelerate life-saving research:

“We believe that AI and machine learning will be an unambiguous benefit to healthcare,” Eskeridge said. “There will be issues. There will be problems to solve: no mistaking that. But it’s going to be something that is going to change healthcare — the way we research it, the way we manage it, the way patients interact with doctors — and those are going to be valuable things.”

2.) AI can help us understand disease risk factors and comorbidities.

Over the past several years, the medical community has gained a more robust understanding of fatty liver disease’s causes, pathology and progression. As a veteran physician, Mistry says AI and ML tools can increase this understanding exponentially.

AI and ML technologies can analyze vast swaths of data at lightning speeds, helping us determine data correlations between the presence of NASH or NAFDL and a patient’s genetics, physiological parameters, lifestyle and other factors.

“The traditional way in which a doctor drew a diagram of “this is causing that” doesn’t apply anymore,” Mistry said. “We need a much more complex framework to collect data, to understand it and analyze it to detect trends and patterns. This is where the entire world of machine learning and AI comes in.”

3.) Computational methods can help promote patient-centric privacy practices.

Kim helped found Array Insights on the idea that we need a careful balance between the utilization of patient data to spur discovery and the trust of patients who create this data.

“Patients like Wayne and others are trusting patient advocacy groups, hospital physicians and also researchers with that data,” Kim said. “I think it’s extremely important that we take that very seriously.”

Hall discussed three computational methods that Array Insights is using to keep their technology patient-centric:

• Analytics on federated data, which allows datasets to be analyzed by ML or AI without making it accessible to other parties;
• Encrypted computation, which enables mathematical operations to be performed on datasets even when they’re encrypted;
• Differential privacy, which helps lower the risk that data can be re-constructed or re-identified.

“A bunch of techniques have been popularized which allow us to put an end to this trade-off between the availability of data and the potential privacy risks surrounding availability,” Hall added.

4.) Data privacy regulations are lagging behind — but industry players can help us keep up.

AI is being implemented quickly — and these capabilities are advancing at a rapid pace, as well. This underscores the need for continuous government regulation.

HIPAA is more focused on data portability than data privacy. GDPR is a more detailed form of regulation, but it’s still not keeping pace with AI advancement.

Kim sees a need for collaboration between patient-centric groups, data scientists and medical professionals to find the right balance and promote meaningful regulation.

“I think it’s very important for technologists and also physicians to come together and have some consensus about, what are the tradeoffs between the utility of patient data and also being able to protect patient privacy,” Kim said.

5.) AI is a tool to augment humans, not a replacement for humans.

At the end of the day, machine learning and AI are mathematical and data analysis tools; not a replacement for your doctor. Healthcare is more than data, and there’s immeasurable importance around the human connections that patients and physicians share.

The medical community should use these AI tools wherever they can to accelerate research and make care more accessible. But at every step, AI should be complemented by empathy and human wisdom.

“If you have just the right amount of AI with humans in the loop, then you have this perfect concert of what can be coordinated in order to improve patient care,” Kim added.

Patient-centric AI practices for advocacy organizations

As the leader of a patient-centric non-profit, Eskridge concluded the webinar by putting the emphasis back on the most important stakeholder: the patient.

“I think one of the great promises of this kind of tooling is that it can become individualized,” Eskeridge said. “The thing that I want is support for the things that are important to me and how I live my life day to day.”

AI has the power to help patients take control of their healthcare journey and data usage. FLF’s AI Fatty Liver Risk Stratification Tool, developed in conjunction with Array Insights, is an excellent example of how this technology can spur meaningful change and put the power back with the patient — which should be the goal of any patient advocate.

 

Are you part of a patient-centric organization that’s looking to implement AI to accelerate your research goals? Reach out to the Array Insights team to learn how our AI and ML solutions can help you serve your patients’ needs.