Apple's AI Physician Announcement

What are the possibilities and probabilities?

Apple has always been focused on how humans interact with technology since Steve Jobs returned to the company in 1997. From the introduction of the iPhone in 2007 to the closed source system that links the Apple Watch to the iPhone, to iCloud, to MacBooks, the company has consistently delivered high end experiences rooted in hardware and enhanced by thoughtful software. For that reason, its latest move to develop an AI-driven health coach they are referring to as an "AI physician”, Apple is venturing into an area that is more personal, more regulated, and far more complex than anything it has tackled before.

Healthcare is not only one of the largest sectors globally with over 10 trillion dollars spent (5.2 trillion in the US) but it's also one of the most stubborn for technology to drive down costs. Despite waves of innovation and digitization, many health interactions still rely on paper-based workflows, faxes, human memory, and siloed data. Apple's new initiative, part of what insiders are calling “Project Mulberry” is aiming to provide personalized health guidance powered by AI and driven by user data already being collected via iPhones, Apple Watches, AirPods, and connected health services/devices.

But can Apple deliver something more than a glorified fitness coach? Can it build a trusted, intelligent, proactive health assistant that genuinely improves outcomes and does so without stepping into ethical or regulation of the FDA?

Sponsored by Dave Breininger at Talent Wave
Talent Wave is transforming how healthcare organizations discover and retain top-tier talent. Led by industry specialist David Breininger, Talent Wave connects exceptional individuals with visionary healthcare companies through a personalized, innovative approach. With a deep understanding of the healthcare space and a commitment to lasting partnerships, David and his team help drive growth, spark innovation, and build a brighter future for both organizations and professionals. Learn more at www.talentwaveassociates.com. Contact David at (567) 304-3102 or DBreininger@talentwaveassociates.com

From Health App to Health Advisor

Apple's health ambitions aren't new. Over the past decade, the company has slowly but steadily evolved its Health app from a glorified step counter into a personal health repository. Today, it can track everything from menstrual cycles to hearing health to lab results pulled from their healthcare providers via Health Records. The Apple Watch has grown into a clinically respected device capable of capturing heart rhythms, identifying potential arrhythmias, and even detecting hard falls.

Despite these advances, the Apple Health app has remained largely passive. It collects and stores information. Occasionally, it might nudge you to stand up/breath or get more sleep, but it doesn't offer the kind of proactive, nuanced coaching that users are going to increasingly demand. This is where Apple's AI physician comes in and where many remain skeptical on how successful Apple could be at delivering such a thing.

With the use of large language models and possibly multimodal AI, this assistant is expected to draw from the data already being collected across Apple devices and deliver personalized, actionable insights. Think of it as a personal coach or companion, and a data analyst rolled into one. But for it to truly rise above the noise, it has to offer more than repackaged statistics that we see quite a bit from health apps these days.

The Hardware Foundation

Apple has one crucial advantage in its quest as it completely controls the hardware stack for  155 million Americans and 1.38 billion users globally. Given those stats, the iPhone is omnipresent, the Apple Watch is a market leader in wearables, and the AirPods have quietly evolved into biometric sensing devices. With each new hardware release, Apple introduces more sophisticated sensors from blood oxygen and ECG monitoring to (soon) sleep apnea detection and hearing aid functions.

All of this data is potential fuel in the world of AI, and Apple is arguably the only consumer tech company that can gather it at such scale and quality. The company has already demonstrated its ability to deploy machine learning on-device, using chips like the Neural Engine to preserve privacy while maintaining performance.

The question now is: can Apple marry its hardware prowess with software innovation to deliver real health outcomes?

A Conservative Software Legacy

In the past, Apple has been more iterative than outside the box when it comes to software. iOS and macOS are refined and polished, but not usually on the cutting edge of innovation. Siri has fallen significantly behind Amazon's Alexa and Google Assistant in terms of natural language understanding and utility. Apple Music is struggling as more and more young people are moving towards Spotify as a trendier/easier to use music app.  

Apple’s efforts in cloud-based services have also shown weaknesses. MobileMe was a failure. iCloud, while now more stable, took years to become reliable. Apple Maps launched with significant errors that led to public criticism which is oftentimes surpassed in favor of Waze or other apps. Even development tools like Xcode, though widely used, have often been described by developers as frustrating and buggy.

All of this points to Apple being great at creating beautiful, secure, and user-friendly experiences, but struggles with large and complex software systems especially those that require constant iteration and agility which smaller companies completely dedicated to solving one problem excel at (Spotify, Waze, etc.).

Apple Intelligence: Augmenting Health, Not Replacing It

About a month ago I watched the Dan Ives podcast with Josh Brown where it clicked with me on what Apple Intelligence really is and what the potential value is in the race for AI. What I realized was that it was more of a personalized LLM on-device that could enhance how users interact with their devices and understand their lives.

As someone who focuses on healthcare, I began to realize that Apple could do more than track steps or heart rate in the healthcare space with this type of technology. By blending sensor data, biometric trends, and natural language processing, Apple could do more like:

Proactive insights: Your Apple Watch notices irregular sleep patterns and flags them alongside changes in heart rate variability. Apple Intelligence could suggest a correlation with stress or dietary changes and recommend a personal recovery plan.

Contextual reminders: If you typically go for a morning walk but haven’t moved much today, and it’s now evening, your iPhone might gently suggest a walk before sunset based on your schedule and the weather.

Symptom summarization: If you've logged fatigue, irregular heartbeats, and dizziness in the Health app over several weeks, Apple Intelligence could create a concise summary you can share with your doctor as most of us always seem to forget all of the exact symptoms, when they occurred and levels of discomfort.

Medication adherence: Apple already has basic medication reminders, but with AI, it could detect if a user is skipping doses and offer timely nudges, educational content, or help scheduling a refill which would be extremely helpful for the aging boomers whose kids have peer pressured them into an IPhone.

Privacy-first health analytics: All of this could be processed on-device, using Apple’s hardware advantage to keep sensitive health data private and secure, while still offering meaningful insights.

Apple’s strength in building trusted, user-first experiences may allow it to succeed where others haven’t. If it can stay focused on helping users and providers make better decisions, rather than replacing them outright, its AI health coach could find the right balance of utility and responsibility. I am not sure many in the US are ready for an AI doctor but I feel like they want their doctor to augment with AI. 

What Is Apple Intelligence Really?

What makes Apple Intelligence great is that it isn’t a singular model like GPT-4 or Gemini. t’s a wrapper and an interface layer that uses multiple smaller models developed at Apple and frameworks behind the scenes to deliver highly contextual, personalized, and privacy-focused AI experiences. It leverages Apple’s Neural Engine, integrates with on-device data, and can optionally reach out to larger cloud-based models (OpenAI/ChatGPT Partnership) when necessary information is needed without user privacy concerns.