Experts Pinpoint AI, Predictive Analytics as Key to Addressing Health Care Challenges

Experts highlighted the potential of artificial intelligence (AI) and predictive analytics to address key health care challenges during a panel discussion at the Council for Affordable Quality Healthcare (CAQH) Connect 2024 in Washington, DC.

The discussion, titled “Future-Proofing 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 cofounder of Birch.ai.

Experts at CAQH Connect 2024 emphasized the potential of artificial intelligence (AI) and predictive analytics in addressing health care challenges, advancing equity, and reducing clinician burden. | Image Credit: Tierney - stock.adobe.com

Addressing Health Care Challenges With AI, Predictive Analytics

Moderator Agnes Buanya, MPP, MA, manager of population health strategy consulting at CareFirst BlueCross BlueShield, began by asking the panelists to discuss what health care challenges AI and predictive analytics can help solve. Terrell highlighted the rapidly aging population, noting that it is different caring for someone aged 65 vs someone aged 85. However, there are not enough people to help care for this increasingly aging population.

He explained that Birch.ai, under Sagility Health, is leveraging generative AI to address this challenge, enabling providers to interpret conversations and documents; this enhances their ability to understand each patient’s needs and preferences.

Bathina agreed, emphasizing that the health care industry must embrace AI technologies to solve these long-term, complex problems since it is now capable of deep reasoning and beating PhD-level experts. In particular, she highlighted the cognitive burden of clinicians who they need to consider years of patient data to provide personalized care. Bathina noted that this is “impossible” without help from advanced technologies, so the health care industry should view AI as a collaborator rather than just a tool.

“The time has come for us to think of AI as a collaborator,” Bathina said. “The role of AI is drastically changing.... The organizations who see that and embrace it are going to move forward faster, and the ones who still think of it as a tactical tool will pay for it.”

Kimura followed up by sharing his passion for using population health strategies to improve care quality and eliminate disparities nationwide. Echoing Terrell’s earlier points, he emphasized the need for a proactive preventive care approach as an aging population with rising body mass index levels faces increasing health risks.

Therefore, Kimura stressed the need to deliver 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 drive transformative improvement.

Buanya continued the discussion by asking what achievements there have been so far in AI and predictive analytic utilization. Bathina discussed specific areas where the industry is seeing progress, using the example of the Medicare population within a pyramid model. She noted that 3 key accelerators can predict how quickly someone will move to a higher-risk cohort, regardless of their condition: social determinants of health, age, and mental health and lifestyle factors.

By using a combination of generative AI and predictive analytics, Bathina said that health care organizations could identify these accelerators and intervene earlier to prevent patients from progressing to more severe, costly conditions; these accelerators amplify the risks associated with chronic conditions like chronic kidney disease, diabetes, and hypertension. She also emphasized her earlier point, saying that, without these tools, limited patient interactions make meaningful interventions challenging for clinicians.

Balancing Risks and Rewards: How Providers Can Prepare for AI Integration

Next, Buanya asked the panelists what providers could be doing to better prepare for the integration of AI and predictive analytics. Terrell first emphasized the importance of, and access to, data, saying that providers are “sitting on a mountain of data” that does not talk to each other. Consequently, he instructed providers to create and integrate a strategy where the data sources interact to unlock the power of AI and predictive analytics.

Kimura agreed, saying that it is “inevitable” that providers use AI and predictive analytics to get desired outcomes. 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 times.

Bathina added to Kimura’s points by discussing the fear of the unknown that often holds providers and health care systems back from embracing new technologies. She encouraged the audience to not be “edge case assassinators,” instead focusing on the value these tools provide, even if they have flaws.

“Go after low-risk, high-value areas to embrace AI,” she said. “You owe it to the health care industry and to the patients.”

As a primary care clinician, Kimura acknowledged the transformative potential of some health care services but criticized the market for being flooded with “entertainment medicine,” solutions with limited impact on meaningful health outcomes. Consequently, he expressed his desire to identify high-value interventions that truly work.

However, he echoed Bathina’s sentiment that medical professionals often need too much strong evidence before adopting new approaches. While optimistic about the potential of these tools, Kimura said he and his colleagues struggle with 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 the progression. He acknowledged these challenges, emphasizing the need for transparency around how these models perform across different populations and continuously monitoring for potential biases. As the power of these tools grows and taps into more unstructured data sources, Kimura emphasized the importance of developing security infrastructure and being vigilant about potential data supply chain sabotage.

Building upon this, Bathina explained that her company is taking a multi-pronged approach to address bias and health equity concerns. It is continuously incorporating the latest guidelines into its software to account for evolving evidence-based practices. To mitigate bias, she noted that her company is building a team of clinicians and specialists who work closely with the AI team. Lastly, Bathina explained that her company focuses on delivering “explainable AI” that can articulate the reasoning behind its interventions.

Terrell shared his perspective by drawing on his experience with tools like Failure Mode and Effects Analysis to proactively identify and mitigate potential issues. He emphasized that applying this mindset to AI and software development could be “highly effective”; it encourages a collaborative approach where everyone contributes by identifying potential issues and addressing those risks.

Conversely, Terrell challenged Bathina’s concept of “explainable AI” since human decision-making is often just as nontransparent, if not more so, than AI models. He instead argued that AI’s capabilities can help illuminate existing processes and workflows, providing more transparency than what currently exists.

Preparing Providers for the Future of Care Delivery

Buanya concluded the discussion by asking the panelists about actionable steps provider groups could take to drive progress in health care delivery, emphasizing the importance of payer-provider alignment. Terrell reiterated his earlier point that they must develop and implement robust data architecture.

Bathina gave multiple suggestions, the first being to implement AI in everyday life as it can help with mental blocks. She also suggested provider groups prioritize their top 3 challenges and assess 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 challenging aspect of health care. She suggested considering aspects like incentives, training, processes, and other change management strategies as essential components of successfully implementing technology to solve pain points. Terrell agreed, claiming that 90% of technology failures are due to change management strategies. He then echoed Bathina’s suggestions and gave some of his own.

“Understand what your processes look like, map those out...and then start to marry that with the technology in a way that gives you better quality, cheaper, and faster,” Terrell said.

Lastly, Kimura said he believes that the patient will be the focus of care delivery in the future. He claimed that these technologies discussed will allow patients to take on roles traditionally managed by the health care system. Consequently, Kimura suggested that health plans start deciding how they will structure their systems to handle the imminent shift.

“You’re going to be needing a lot of information coming in a very unstructured way, feeding into that system in order for you to be able to really be where the puck is skating to in the future,” he concluded. “I’d say, start looking at patient information and how you’re going to start structuring your system to handle that because it’s coming.”