Will AI-Powered Virtual Assistants Transform the Future of Chronic Disease Management Apps?

A few days ago, Sense.ly, a U.S. digital health startup specializing in AI-powered virtual nurse services, secured $8 million in a new round of financing led by the Chinese investment firm Chengwei Capital. Sense.ly delivers healthcare management services, primarily for chronic disease patients, through a conversational interface featuring a female robotic voice named Molly. The chatbot’s user interface is its most distinctive feature, often referred to as a virtual intelligent assistant. Could this herald a different future for chronic disease management apps? The following article analyzes this question from two perspectives: the potential value of virtual intelligent assistants and the associated risks.

We know that, in China, targeting the field of chronic diseasesAPPA large number have emerged in the past two to three years, with cardiovascular diseases being the primary category. Due to the significant differences between China’s healthcare system and those of Europe and the United States, the most impactful factors are the substantial disparities in medical cost payment systems and tiered diagnosis and treatment models, which pose challenges to domestic chronic diseaseAPP...has brought considerable awkwardness to their survival and development. These chronic disease managementAPPHaving undergone various explorations. Starting with a focus on2CTurn2B, they have sought strategic partnerships with various industry stakeholders, including hospitals, pharmaceutical companies, and insurers, to facilitate value monetization. SomeAPPIt has evolved from an initial single-dimensional service model into a multi-dimensional landscape that integrates software and hardware, combines online and offline operations, and encompasses both self-operated services and outsourced solutions.

While external factors in the industrial environment may determine survival, the true influence of chronic disease management apps on users, as the most critical internal factor, can also shape their fate. The chatbot interface format we are analyzing today falls under product planning and design, representing the cultivation of core competencies. Therefore, we must examine whether the chatbot’s product model helps enhance the value that chronic disease management apps deliver to users.

Let’s return to the most fundamental question: Why do patients with chronic diseases need chronic disease management apps? Did patients with high health awareness fail to practice self-management in the pre-app era? Certainly not. The value of apps lies in helping patients manage their conditions more efficiently. So, what has caused the inefficiency of chronic disease management in the past? First, patients lack medical knowledge and are unclear about what measures help improve their condition; this falls within the realm of patient education. Second, due to this lack of knowledge, patients seek help from doctors but struggle to obtain timely assistance, reflecting issues such as inefficient access to medical services, congestion, or excessive costs. Third, even with adequate knowledge and support, maintaining a healthy lifestyle still requires self-discipline; adherence to medical advice remains problematic, and establishing good lifestyle habits continues to be challenging. This highlights the issue of how to motivate patient engagement.

Before delving into the specific analysis, let us revisit the core value of the chatbot model: it further enhances the human-computer interaction (HCI) interface by enabling interactions through communication methods that align more closely with human habits, whereas all previous models required humans to adapt to machines. (This point has been mentioned in my previous articles.) Some refer to chatbots as the new apps, while others predict that chatbots will replace apps in the future. I view them as an upgrade of the interaction interface; regardless of the type of interface, it remains a software application product. Based on this understanding, chatbots offer several potential advantages in chronic disease management.

Historically, nearly all digital health hardware and software solutions have grappled with the challenge of enhancing user stickiness. Although health is a top priority in life, people often struggle to elevate health-related behaviors to the highest level of urgency until a crisis arises. Countless objective and subjective factors hinder our adherence to healthy lifestyles, preventive measures, and medical advice. The immediacy and mobility of mobile apps can, to some extent, help overcome temporal and spatial barriers, while well-designed chatbots can further reduce operational complexity, offering a simpler and more convenient user experience.

For instance, patients can acquire relevant medical knowledge simply by asking questions, eliminating the need to search, filter, and locate matching content. Chatbots can also interpret semantics and understand instructions to perform tasks on behalf of users, such as recording daily health metrics and dietary intake. Users no longer need to fill out forms; instead, they can use just one or two sentences in natural human language. While these features may offer limited utility to IT professionals, they can significantly improve the experience for older adults who are less accustomed to interacting with machines, particularly through intelligent voice interfaces. This is especially relevant given that a larger proportion of patients in the chronic disease sector are elderly. At the recent HIMSS Conference, Lenovo Health announced a partnership with Orbita, a provider of voice-enabled smart solutions, to develop a health virtual assistant that delivers home healthcare services through intelligent voice conversations. In fact, AI-powered voice virtual assistants developed by overseas tech giants, such as Microsoft’s Cortana and Amazon’s Alexa, aim to integrate their technologies into the home health domain, offering services including medication reminders, appointment scheduling, and health education.

In the product design of chatbots, it is common practice to assign a name to the virtual intelligent assistant. Although these assistants are not real people and often make “silly” mistakes, product designers still prefer to use human names to reinforce their “human-like” qualities. In the context of chronic disease management, a virtual assistant with a human name and even a human avatar can have a certain psychological impact on patients. When there is no real doctor present on the other end of the network, having a virtual “companion” is more motivating than managing one’s condition alone in isolation.

Furthermore, this virtual companion can proactively engage in inquiries, reminders, care, and follow-ups using a human-like tone and persona. Chatbot designers across various sectors generally consider this interaction style to be more proactive and more effective for customer engagement; in the context of chronic disease management, it serves to enhance patient engagement. In reality, however, user retention rates for most chronic disease management apps have not reached expected levels, and their actual impact on patient engagement has been less than ideal. In recent years, community-building and gamification have emerged as relatively effective concepts for customer engagement, and chatbots are poised to offer new mechanisms for driving engagement.

Why Focus Specifically on Chronic Disease Management Rather Than the Entire Mobile Health Sector? This is because chronic disease management applications are inherently high-frequency in nature; they should demand higher user stickiness and require more robust incentives to enhance patient engagement.

The fundamental cause of inefficiency in accessing healthcare services is the scarcity and uneven distribution of medical resources. So, what can chatbots do?

In chronic disease management, chronic disease apps serve as a bridge for communication between doctors and patients. Ultimately, they must make healthcare providers’ work easier rather than more burdensome. From the perspective of healthcare providers, chatbots can function as assistants to medical staff by handling certain routine administrative tasks. They can manage administrative affairs such as appointment scheduling and registration, respond to frequently asked questions about common conditions, and perform routine patient monitoring on behalf of healthcare professionals, among other duties.

One-on-one, on-demand medical services are clearly out of reach for the general public. However, with the intervention of chatbots, it becomes possible to deliver low-cost, one-on-one personalized care for certain common, routine health issues that can be managed remotely. In other words, with the aid of artificial intelligence, chronic disease management apps can reduce the number of offline healthcare professionals required for matching without adversely affecting the service experience.

Bubbles inevitably accompany the emergence of every new innovation, and while these bubbles eventually burst, survivors always remain. The same holds true for intelligent virtual assistants. Here, I have no intention of inflating hype; rather, I aim to provide a rational, forward-looking analysis of the possibilities brought about by such innovations. Consequently, we must also anticipate potential pitfalls to enable well-calibrated innovation and prepared risk mitigation.

First, regarding the technical issues of chatbots.

If the technology fails to meet standards, user experience will suffer, and value may not only fail to increase but could even be diminished. The key technological areas involved are: first, natural language processing; second, speech recognition; and third, knowledge graphs. The first two technologies can leverage open-source solutions and collaborative platforms. However, the third area, in addition to its reliance on artificial intelligence, requires domain-specific expertise in medical and health knowledge, as well as insights into healthcare services and management workflows within chronic disease management. Such vertical specialization demands professional knowledge, necessitating collaboration between medical professionals and AI technology experts to address these challenges effectively.

The first two categories of technology are open to collaboration for anyone and thus cannot serve as competitive barriers; only by mastering the third area can a company potentially develop core competencies. However, from another perspective, the maturity of the first two technologies depends on partnerships with tech giants. If technological maturity is insufficient, product experience will inevitably suffer, making the choice of partners critically important.

Although these related technologies have attracted significant attention from many tech giants and startups in the past one to two years, AI experts report that while they have improved compared to previous iterations, they are not yet fully mature. Natural language understanding remains inadequate in handling complex contextual relationships, and the much-touted 97% accuracy rate for speech recognition is achieved only under ideal conditions. In fact, accurate recognition of dialects is certainly difficult to achieve. Of course, voice interaction is not strictly necessary; chatbots can also operate via text-based conversations.

From my perspective, although the technology is not yet fully mature, we can still derive certain value within its limited capabilities. Therefore, if developing a health virtual assistant, one should not expect the technology to handle highly complex contextual conversations. Instead, functionalities should be confined to a specific scope, starting with simple and clear commands. For instance, straightforward tasks already implemented overseas—such as ordering pizza via chatbots, setting alarms, and making restaurant reservations—are ideal use cases that can be handled effectively.

Secondly, data issues will become a bottleneck.

Current artificial intelligence technologies are heavily reliant on data, and it is now time to test the efforts invested in data over the past few years. The concept of “data cultivation” has been proposed, suggesting that realizing the monetary value of data requires significant time and effort to “nurture” it. In many cases, this process cannot be rushed due to inherent temporal constraints, thereby enabling early movers to secure a superior data advantage.

We know that the patient community PatientsLikeMe provides data hygiene services to pharmaceutical companies, but this monetization model took about a decade after its establishment to realize. The chatbot consultation service offered by the overseas telemedicine company HealthTap bases its Q&A on data retained from past real doctor-patient interactions. So, has our chronic disease management app been effectively “nurturing data” over the past few years?

If sufficient and appropriate data has not been accumulated in the past, is it necessary to start preparing now? Even though the maturity of relevant technologies still needs improvement, we may still need to make early preparations, or at least maintain close attention. The user experience of chatbots often requires a period of machine learning to be trained and improved. When the experience is not yet ideal, chatbots in other fields may adopt a combination of human and machine operations, running in parallel to accumulate data and validate results.

Finally, the external environment of the business ecosystem remains unchanged.

As mentioned at the beginning of this article, chronic disease management apps still face an awkward survival environment, and their profit models have not been established smoothly. Chatbot interfaces enhance the value of the product form to users but do not affect the external business environment or even the underlying service operations; one should not expect them to solve all problems. However, I would like to add that while environmental changes may be slow, especially in the healthcare sector, the overall trend is irreversible. Regardless of the healthcare system, the ultimate goal will always be to pursue more efficient operational methods. Three years ago, when I read Eric Topol's "The Patient Will See You Now" (Chinese title: "Disrupting Healthcare"), I wrote a book review titled "Disruption is a Slow Motion," emphasizing that change does not happen overnight. The Chinese translation of the book's title seems somewhat exaggerated. Over the past three years, the various successes and shortcomings of internet healthcare have been inevitable experiences. Entrepreneurs must clarify the fundamental logic and prepare for a long-term struggle.

Author Gu Beini (Personal self-media account:futuretalking）

To read more articles by this author, please viewGu Beini Column