Getty Images/iStockphoto

AI-Based Voice Assistants Poised to Support Chronic Disease Management

New data showed that conversational AI and voice assistants helped patient engagement in chronic disease management.

Conversational artificial intelligence (AI) could be the key to unlocking better chronic disease management, with new data in JAMA Network Open showing that a voice assistant tool improved glycemic control and insulin dosing for type 2 diabetes patients.

These findings come as the healthcare industry weighs the pros and cons of consumer AI use and how the technology can streamline the patient experience.

For patients with type 2 diabetes, it can be hard to achieve optimal HbA1c levels in part because it’s difficult to calibrate their insulin dose, the researchers said. Currently, about a quarter of the 33 million patients with type 2 diabetes in the US have poor glycemic control, meaning HbA1c levels above 8 percent, they pointed out.

“Insulin therapy is essential for individuals with poorly controlled diabetes, but effective use requires frequent dose titrations, which can be challenging to achieve in practice because titrations typically occur at outpatient clinic visits every 3 to 6 months,” the researchers wrote in the study’s introduction.

“In addition, many clinicians fail to escalate insulin therapy when indicated due to therapeutic inertia, lack of time, and competing demands in appointments,” they added. “As a result, most patients treated with insulin receive suboptimal doses and do not achieve glycemic control.”

But the insurgence of conversational AI could be poised to change that. Using a voice-based conversational artificial intelligence app, the researchers found that patients could effectively manage self-titration of insulin and achieve better chronic disease management.

The AI system was powered by Amazon Alexa and deployed on a smart speaker, which researchers supplied to patients. Before patients received the system, their primary diabetes provider selected an insulin titration protocol and entered it using an online portal. From there, the system was able to guide patients through self-titration.

After saying, “Alexa, check in with clinical trial,” patients were able to report clinical data, including insulin use and fasting blood glucose values. Once patients were done self-reporting information, the AI system would provide updated insulin dosing instructions for the patient and record those updates in the provider-facing portal.

This feature was key, the researchers stressed, because they prioritized placing clinical decision-making with providers, not with AI. The AI was informed by clinician protocols.

This system was effective, albeit on a smaller patient population. Across the 32 patients included in the trial, those who used the conversational AI achieved optimal insulin dosing more quickly than those who did not—15 days versus 56 days. They also had 32 percent better insulin medication adherence, were more likely to achieve glycemic control, and were more likely to see glycemic improvement over the course of the study.

The researchers said these improvements were possible because the system was user-friendly. The tool used simple language so as not to create language barriers or present health literacy as a hurdle. It also bypassed digital health access and literacy issues by leveraging conversational AI, rather than artificial intelligence housed on a smartphone app.

“Voice-based conversational artificial intelligence has the potential to improve access to technology-enabled care for patients with low digital literacy, while simultaneously enhancing engagement for all patients,” the researchers explained.

“Voice-based conversational artificial intelligence applications that accelerate the time to control can also help counteract the negative effects of clinical engagement attrition that can occur among patients with chronic diseases.”

All of this had a positive impact on the patient, not just clinically but emotionally, too. Self-reported diabetes-related emotional distress was 3.6 points lower for the group using the conversational AI tool than those who did not.

These findings speak to the overall potential for AI to support the patient experience. Conversational AI like that which was studied in this report has numerous applications in other consumer service sectors, but it is only emerging in the healthcare space.

As experts continue to weigh the pros and cons of consumer AI use in healthcare, this data may outline the potential for guiding better patient engagement in chronic disease management.

Next Steps

Dig Deeper on Patient data access

xtelligent Health IT and EHR