Apple Revives 'AI Doctor' Initiative with Project Mulberry to Launch Health+ in 2026

On March 30, Bloomberg reported that Apple is overhauling its Health app with the aim of developing a new “AI Doctor.”

Reportedly, the initiative, codenamed Project Mulberry, currently comprises four core modules: an AI health coach, food tracking, exercise analysis, and educational content. It aims to leverage the iPhone as a platform to provide users with an algorithm-driven virtual physician.

The Four Health Modules of Project Mulberry

If users agree to the app’s terms, Apple will collect various health data generated in their daily lives and provide it to AI physicians for analysis, thereby creating personalized health management plans for each user.

Meanwhile, Apple will co-create educational videos with physicians specializing in sleep, nutrition, physical therapy, psychology, and cardiology. These materials will serve as a supplement to the AI doctor, intuitively presenting users’ health information or explaining the rationale behind the formulation of health plans.

For example, when Apple’s App collects data indicating abnormal heart rates, the AI physician not only alerts the user but also pushes videos explaining cardiovascular disease risks, urging patients to seek timely intervention.

However, as early as the nascent stage of deep learning, tech giants such as Google, Microsoft, and IBM launched their own distinctive “AI doctors” based on their respective philosophies. Yet the vast majority of these initiatives failed, and few companies now speak of “AI doctors.”

In this era of large language models, Apple’s doctor seems to have arrived somewhat late.

Based solely on the content of Project Mulberry, Apple’s newly unveiled four modules offer little to capture attention. AI health coaches and video-based education are hardly novel concepts, while computer vision-enabled applications such as food tracking and exercise analysis have already been explored by both domestic and international vendors.

But for Apple, now may be its best time.

As is well known, Apple’s generative AI capabilities have consistently lagged behind those of other members of GAFA. While Gemini and GPT were engaged in fierce competition in AI logical reasoning, conversational Siri merely delivered a test success rate of 66%–80%. With one erroneous response occurring in every three interactions, the original launch plan scheduled for 2024 has been pushed directly to 2027.

To reverse its declining momentum, Apple announced a $500 billion domestic investment plan in February, launching a strategic turnaround effort. Under this plan, Apple will prioritize the development of AI server manufacturing and R&D expansion over the next four years, strengthening its capabilities in cloud computing, AI inference, and data analytics. The company is expected to unveil its self-developed AI system at the 2025 WWDC, integrating voice, vision, and multimodal interaction.

Therefore, the emergence of this new plan at the current juncture suggests that either the concept of an “AI doctor” has been incubating within Apple’s health ecosystem vision for many years, or Project Mulberry has capitalized on Apple’s own AI momentum, driven by Tim Cook’s strong commitment to the healthcare industry.

Second, although domestic and international companies have long been strategically positioning themselves in the field of “AI doctors,” many have failed to bring their applications to market due to algorithmic limitations and data deficiencies, resulting in unsatisfactory development outcomes.

For example, many physicians have previously expressed a need for AI vendors to develop intelligent algorithms that can accurately identify the caloric content of various foods in images, thereby providing users with intuitive calculation results to help control caloric intake.

In the era before generative AI, AI vendors could only perform food segmentation using deep learning and estimate caloric content based on average values for processed foods. However, due to the diversity of food types and varying preparation methods, AI systems at that time exhibited significant discrepancies from reality in both food component segmentation and caloric estimation.

Apple has chosen to enter this arena at this juncture, largely because the relevant AI technologies have reached maturity.

Today, the iPhone 16 can already estimate the calorie and carbohydrate content of food with considerable accuracy, assisting patients with conditions such as obesity and diabetes in managing their health. Furthermore, empowered by generative AI, the new generation of “food tracking” is highly likely to achieve greater precision in segmentation and calculation, meeting medical-grade standards.

Similarly, exercise analysis can also achieve breakthroughs with the help of new AI technologies. Leveraging the rear camera of the iPhone, AI doctors can obtain richer data to study users' exercise conditions, provide corresponding movement and posture corrections, and realize the function of a "digital physical therapist."

Overall, with its existing AI infrastructure investment plans combined with years of accumulated AI vision capabilities, Apple is poised to develop a new generation of digital intelligence applications based on these established concepts. Moreover, bolstered by its ecosystem advantage in wearable devices, Apple may indeed revitalize the concept of the “AI doctor,” enabling its large-scale commercial deployment.

Based on the existing content of Project Mulberry, the domain that Apple’s AI Doctor aims to enter could be “AI + Health Management” or “AI + Digital Therapeutics.”

In recent years, amidst a rapidly changing landscape, many startups in both sectors have failed to achieve their goals. The root cause lies in the lack of data, which hinders these companies from delivering precise personalized diagnosis and treatment plans; consequently, the value generated by their services has failed to cover the high costs incurred during research, development, and promotion.

Thanks to the multimodal data integration capabilities enabled by its wearable device ecosystem, Apple is first able to address the challenges posed by “precision” and “personalization.”

In 2014, the first-generation Apple Watch integrated a heart rate sensor, enabling continuous monitoring of users’ heart rate variations. After 2018, the Apple Watch introduced ECG functionality, allowing it to dynamically detect users’ cardiac rhythms in the background.

ECG can be regarded as the first “killer app” feature of wearable devices in terms of healthcare functionality, enabling patients who truly require continuous vital sign monitoring to benefit from technological advancements, thereby unlocking the vast market of individuals with chronic diseases. In 2021, Apple’s single-lead ECG and atrial fibrillation notification features received FDA clearance as “Breakthrough Devices,” making it the world’s first over-the-counter product directly marketed to consumers that detects electrocardiograms from the wrist, thus marking Apple Watch’s transition from a consumer-grade to a medical-grade device.

In September 2020, the newly released Apple Watch Series 6 added blood oxygen monitoring functionality. Combined with heart rate variability (HRV) analysis, Apple can generate in-depth sleep quality reports that reveal the correlation between users' nocturnal respiration and cardiovascular status, providing an early warning window for suboptimal health conditions such as chronic fatigue and sleep disorders.

In the realm of blood glucose monitoring, Apple previously collaborated with Rockley Photonics to develop silicon photonics technology, aiming to achieve non-invasive blood glucose monitoring through laser spectroscopy analysis. Although the two parties parted ways in 2021, Apple’s accumulated expertise in micro-sensors and optical algorithms is accelerating the realization of this critical project. If successful, the Apple Watch will, for the first time, enable closed-loop monitoring of metabolic and cardiovascular indicators such as blood glucose, blood pressure, and electrocardiogram (ECG), thereby disrupting the existing paradigm of chronic disease management.

Through multimodal data fusion, Apple has created a comprehensive health profile for its users. By leveraging continuous, multidimensional data to support decision-making, AI physicians can theoretically formulate precise health plans and dynamically adjust them in response to changes in the user’s physical condition.

The Roles of Various Hardware and Software Devices in Apple's Health Ecosystem

Secondly, many companies in the fields of “AI health management” and “AI digital therapeutics” have long faced high operating expenses, as they must incur substantial costs in research and development, customer acquisition, and physician services.

Within the content modules disclosed for Project Mulberry, most services require only in-house development, with only video-related content necessitating physician support.

Furthermore, Apple already possesses a platform with a massive user base, which allows it to save significant upfront costs and achieve profitability even with a lower average transaction value.

Finally, there is the business model. The vast majority of a startup’s revenue needs to be generated by providing services, whereas Apple has a richer array of business model options.

For instance, AI doctors can be offered as software integrated into the Apple One bundle via Health+, with a 30% commission on medical subscription fees; alternatively, they can serve solely as a cost center, leveraging tight integration with the Apple ecosystem to boost sales of other Apple wearable devices.

With three major advantages, Apple may leverage its AI doctor to penetrate markets that startups previously failed to crack, tapping into the new hundred-billion-dollar out-of-hospital health market.

Although existing AI technologies still have a way to go before fully integrating into serious medical practice, by examining the current application development trends of AI companies, the reform path of review and approval by the National Medical Products Administration (NMPA), and the rapid acceptance of AI technology by the general public, we can clearly foresee an era in which AI serves as a personal physician.

Apple’s typically cautious entry into strategic positioning may signal that this sector has already demonstrated a viable path to success.

Of course, it is not just “AI doctors” that are undergoing fundamental changes.

At this juncture, it is imperative that we follow Apple’s example and re-examine the intelligent applications that fell short during the past decade of AI development. As the technological foundation leaps forward, those previously abandoned medical AI scenarios may now find new possibilities for revitalization.