Pieces, a healthcare AI company, recently convened an industry roundtable about the current state of AI and what the future will look like as healthcare leaders cite that they will continue to invest in the technology in 2021. The roundtable participants shared their perspectives on what has been achieved, as well as where more strategic innovation is needed for AI to better serve physicians and patients.
The roundtable participants included:
- Ruben Amarasingham, MD, Founder and CEO, Pieces
- Sarah de Ramirez, MD, MPH, MSc, VP, Chief Medical Officer Clinical Innovation, OSF HealthCare
- Sandeep S. Vijan, MD, FACS, CPE, Chief Medical Officer, Parkview Medical Center
- Garrett Vygantas, MD, Managing Director, Venture Group, OSF Innovation
- Mark Ziemianski, Senior Vice President and Chief Data Officer, Children’s Health
“In 2020, AI has continued to evolve in its applications to the front lines of healthcare – perhaps accelerated by the dire clinical circumstances worldwide,” said Amarasingham. “But as these roundtable discussions revealed, AI can achieve more – and help more. We hope the insights from this roundtable discussion will be useful to healthcare systems, clinical providers, and innovators who are developing their own AI strategy.”
Key takeaways from the roundtable included:
- AI helped to democratize access to COVID-19 care : “During the pandemic, natural language processing (NLP), automated patient screening, and digital care navigation really helped to democratize access to care in ways we couldn’t have done in the past,” said de Ramirez. “We had chatbots before but they were clunky and uninformed. During COVID-19, we really saw them take off for patient screening and optimization of care utilization. OSF HealthCare was able to use conversational platforms for specific AI use cases during COVID-19, and then loop clinicians in as appropriate, which is a theme I think we will continue to see.”
- AI was an important accelerant for vaccine and therapeutics developers: “The rapid genetic and phenotypic characterization of COVID-19 early in the pandemic gave a starting point for vaccine and therapeutics developers to use algorithms in their development work,” said Vygantas. “The diagnostics field benefitted from the application of AI and computer vision to improve the performance of the COVID-19 test, including improving the performance of the first at-home, rapid antigen test, which is now under Emergency Use Authorization (EUA) review.”
- COVID-19 was a catalyst for rapid innovation in AI – but we must also recognize areas where AI could have done more to help combat the pandemic: “In 2020, AI and machine learning allowed us to make fine-tuned predictions of patient risks who have COVID-19,” said Amarasingham. “It’s hard to know exactly where AI has fallen short but I have wondered whether AI-infused tools could have been better deployed to combat misinformation and increase patient education and awareness.”
- AI can do more for physicians in 2021 and beyond by taking insights further – and working within existing workflows: “AI can better support physicians by helping us see what is not always obvious,” said Vijan. “For example, can it reveal which patient is most likely to respond to a given intervention? Similarly, can it uncover how we can tailor an intervention based on a patient’s risk profile? This is important because it takes 15 years for peer-reviewed science to make it into everyday clinical practice but AI can accelerate that integration.”
- “If AI is not reducing the stress or complexity of the workflow, it is either not working, not optimized or not useful,” said Amarasingham. “Everything in healthcare delivery touches some sort of workflow, including patient activities. Therefore, it is through workflows that AI can help improve some of the most critical outcomes, reduce costs and enhance clinical accuracy.”
- AI can help transform healthcare but ongoing innovation requires understanding both its promise and its limitations: According to Ziemianski, “We must stay true to the foundational premise of AI: if you cannot accurately measure an outcome then you can’t possibly wrap AI around it.” He stresses that it is critical not to develop AI in a vacuum. “We must work with subject matter experts to ensure they know what question we are asking and why – and we must be upfront about shortcomings of AI before implementing a solution. Most importantly, we must prevent the use of AI to provide a bad result but faster. That is not the role of AI,” he said.
- This year’s innovation has laid the groundwork for AI to grow and thrive – but more action is needed to ensure continued progress: “2020 has really been all about COVID-19, and building public health and medical systems to be more proactive, in the face of communicable diseases,” said de Ramirez. “Now, our job is to translate these efforts into useful lessons across disease types, both from a precision medicine and precision prevention aspect.”
Most importantly, the roundtable participants agree that the time is now for AI to advance and make a difference in the real-world. “Data is great, but creating workflow around the data is equally important,” said Vijan. “AI companies must not only show what their software can do, but also must help hospitals build the human workflow around the data to deliver real results.”
To download the full roundtable report, please visit the Pieces website.