Experts highlighted the potential of artificial intelligence (AI) and predictive analytics to address key healthcare challenges during a panel discussion at the Council for Affordable Quality Healthcare (CAQH) Connect 2024 in Washington, DC.
The discussion, titled “Future-Proof Patient Care: Innovations in Population Management,” included panelists Deepthi Bathina, MBA, founder and CEO of RhythmX AI; Joe Kimura, MD, MPH, chief medical officer of Somatus; and Kevin Terrell, MBA, CEO and co-founder of Birch.ai.
Addressing healthcare challenges with AI and predictive analytics
Moderator Agnes Buanya, MPP, MA, Population Health Strategy Consulting Lead at CareFirst BlueCross BlueShield, began by asking panelists to discuss healthcare challenges that AI and predictive analytics can help resolve. Terrell pointed to the rapidly aging population, noting that caring for a 65-year-old is different from an 85-year-old. However, there are not enough people to help care for this increasingly aging population.
He explained that Birch.ai, under Sagility Health, leverages generative AI to address this challenge, allowing providers to interpret conversations and documents; this improves their ability to understand the needs and preferences of each patient.
Bathina agrees, emphasizing that the healthcare industry needs to embrace AI technologies to solve these complex problems in the long term because it is now capable of deep reasoning and beating PhD-level experts. In particular, she highlighted the cognitive load on clinicians who must consider years of patient data to provide personalized care. Bathina noted that this is “impossible” without the help of advanced technologies, and that the healthcare sector should therefore view AI as a collaborator rather than just a tool.
“The time has come for us to view AI as a collaborator,” Bathina said. “The role of AI is changing dramatically…Organizations that realize this and adopt it will move forward faster, and those that still view it as a tactical tool will pay the price.”
Kimura then shared his passion for using population health strategies to improve quality of care and eliminate disparities nationwide. Echoing Terrell’s previous points, he emphasized the need for a proactive approach to preventative care, as an aging population with a rising body mass index faces increasing health risks.
Therefore, Kimura emphasized the need to provide effective care directly to patients, address restrictive provider dynamics, and empower patients through technology. If effectively integrated, he noted that aligning finances with value-based care and population health management could lead to transformative improvement.
Buanya continued the discussion by asking what achievements have been made so far in AI and the use of predictive analytics. Bathina discussed specific areas where the industry is seeing progress, using the example of the Medicare population in a pyramid model. She noted that three key accelerators can predict how quickly a person will move into a higher risk cohort, regardless of their condition: social determinants of health, age, mental health and mode factors. of life.
Using a combination of generative AI and predictive analytics, Bathina said healthcare organizations could identify these accelerators and intervene earlier to prevent patients from progressing to more serious and costly conditions; these accelerators amplify the risks associated with chronic diseases such as chronic kidney disease, diabetes and hypertension. She also emphasized her previous point, saying that without these tools, limited interactions with patients make meaningful interventions difficult for clinicians.
Balancing Risk and Reward: How Suppliers Can Prepare for AI Integration
Buanya then asked the panelists what providers could do to better prepare for the integration of AI and predictive analytics. Terrell first emphasized the importance of data and access to it, saying providers are “sitting on a mountain of data” that don’t communicate with each other. Therefore, he called on vendors to create and integrate a strategy where data sources interact to unleash the power of AI and predictive analytics.
Kimura agrees, saying it is “inevitable” that providers will use AI and predictive analytics to achieve the desired results. He suggested they start by providing training and demonstrating the power of the tools. Specific instances where these tools could help providers include using AI to prioritize and focus discussions during limited patient visit periods.
Bathina added to Kimura’s arguments by discussing the fear of the unknown that often prevents health care providers and systems from adopting new technologies. She encouraged the audience not to be “killers in edge cases,” but instead to focus on the value these tools provide, even if they have flaws.
“Aim for low-risk, high-value areas to adopt AI,” she said. “You owe it to the healthcare industry and to patients. »
As a primary care clinician, Kimura recognized the transformative potential of some health services, but criticized the market for being flooded with “entertainment medicine,” solutions with limited impact on meaningful health outcomes. Therefore, he expressed his desire to identify high-value interventions that actually work.
However, he shares Bathina’s sentiment that medical professionals often need too much solid evidence before adopting new approaches. While optimistic about the potential of these tools, Kimura said he and his colleagues face the risk of ineffective solutions, leaving them hesitant to adopt new tools after past disappointments.
Addressing Bias and Health Equity in AI
As AI evolves, Buanya asked Kimura how to ensure it accounts for socioeconomic, racial and ethnic differences without slowing progress. He acknowledged these challenges, emphasizing the need for transparency about how these models work in different populations and to continually monitor potential bias. As the power of these tools increases and leverages increasingly unstructured data sources, Kimura stressed the importance of developing security infrastructure and being vigilant about potential sabotage of the supply chain of data.
Based on this, Bathina explained that her company takes a multi-pronged approach to addressing issues of bias and health equity. It continually integrates the latest guidelines into its software to reflect evolving evidence-based practices. To mitigate bias, she noted that her company builds a team of clinicians and specialists who work closely with the AI team. Finally, Bathina explained that her company is focused on providing “explainable AI” that can articulate the reasoning behind its interventions.
Terrell shared his perspective drawing on his experience with tools such as failure mode and effect analysis to proactively identify and mitigate potential issues. He emphasized that applying this mindset to AI and software development could be “very effective”; it encourages a collaborative approach in which everyone contributes by identifying potential problems and addressing these risks.
Conversely, Terrell questioned Bathina’s concept of “explainable AI,” as human decision-making is often just as opaque, if not more so, than AI models. Instead, he argued that AI capabilities can help illuminate existing processes and workflows, providing more transparency than currently exists.
Preparing Providers for the Future of Care Delivery
Buanya concluded the discussion by asking panelists about concrete steps provider groups could take to advance health care delivery, emphasizing the importance of payer-provider alignment. Terrell reiterated his previous point that they need to develop and implement a robust data architecture.
Bathina gave several suggestions, the first being to implement AI in daily life as it can help combat mental blocks. She also suggested provider groups prioritize their top three challenges and evaluate how technology can help solve them.
Bathina concluded by explaining that, in her experience, technology is “a small part of a bigger puzzle,” with changing human behavior being the most difficult aspect of healthcare. She suggested considering aspects such as incentives, training, processes and other change management strategies as essential elements of successfully implementing technology to solve problems. Terrell agrees, saying 90% of technology failures are due to change management strategies. He then echoed Bathina’s suggestions and gave a few of his own.
“Understand what your processes look like, map them out…and then start combining them with technology in a way that gives you better quality, lower cost, faster,” Terrell said.
Finally, Kimura said he believes the patient will be the focus of care in the future. He said the technologies discussed will enable patients to take on roles traditionally managed by the healthcare system. Therefore, Kimura suggested that health plans begin to decide how they will structure their systems to deal with the impending change.
“You’re going to need a lot of information coming in a very unstructured way, feeding into that system so that you can really be where the puck is skating in the future,” he concluded. “I would say, start looking at patient information and how you’re going to start structuring your system to handle this, because it happens.”