September 2025 Healthcare Roundup: Apple Watches Blood Pressure, Humanoid AI Helpers, and Superintelligence Dreams

September has shown us the extremes of healthcare AI ambition, with Apple's FDA-cleared hypertension monitoring representing measured innovation built on existing hardware, while Soul Machines launched eerily human-like digital workers that blur the line between helpful automation and unsettling simulation for some. Meanwhile, we have ourselves a new unicorn, with Lila Sciences raising an additional  $235 million for "scientific superintelligence" that promises to automate the entire scientific method—a moonshot that either represents the future of drug discovery or biotech's most spectacular bubble yet.

Apple Gets FDA Clearance for Watch Blood Pressure Alerts, Making PPG the New Standard

September 16 - Apple received FDA clearance for its hypertension notification feature on Apple Watch, using photoplethysmography (try saying that three times fast) to analyze blood vessel responses over 30-day periods. The feature will roll out to Apple Watch Series 9, 10, and 11, plus Ultra 2 and 3 models in over 150 countries before month-end, potentially alerting more than one million people to undiagnosed high blood pressure in the first year.

Why it matters: Apple's approach represents a fundamental shift in how we think about chronic disease detection—from episodic clinical encounters to continuous, passive monitoring. The company trained its algorithm on data from over 100,000 participants and validated it in a 2,000-person clinical study, demonstrating that existing sensor technology can be repurposed for medical applications without requiring new hardware.

What's particularly clever about Apple's strategy is that this isn't limited to their newest models. Users with Series 9 watches will get the same hypertension alerts as those with the latest Series 11, making this a software-driven health upgrade rather than a hardware sales play. This democratization of health monitoring within Apple's ecosystem could set a precedent for how other wearable companies approach health features.

The timing is notable given the broader regulatory landscape around wearable blood pressure monitoring. Just months after Whoop's public battle with the FDA over blood pressure insights, Apple's clearance shows how proper clinical validation and FDA engagement can enable similar capabilities. Meanwhile, Aktiia's Hilo Band received OTC clearance in July for 24/7 cuffless monitoring, suggesting the entire category could be moving toward photoplethysmography-based solutions.

However, industry experts emphasize the importance of managing user expectations. The feature is designed for early detection, not medical diagnosis, and false reassurance remains a concern—users who don't receive alerts might incorrectly assume they don't have hypertension. The real test will be whether users follow through on notifications with proper cuff-based measurements and clinical consultations.

This clearance adds to Apple's growing portfolio of FDA-authorized health features, including sleep apnea detection, irregular heart rhythm notifications, and even hearing aid functionality in AirPods Pro. The company is systematically turning consumer electronics into medical screening tools, potentially reaching populations that traditional healthcare delivery struggles to engage with preventive care.

Soul Machines Brings Digital Sims to Healthcare: When AI Gets Uncomfortably Human

September 3 - Soul Machines launched its Digital Workforce platform, deploying lifelike AI-powered digital workers for healthcare, HR, and sales functions across large enterprises. The company's technology creates eerily realistic digital humans that can "see, hear, react, remember and even empathize" through what they call Experiential AI and their patented Digital Brain—essentially creating The Sims characters that have escaped the computer screen and wandered into your doctor's office.

Why it matters: While patients increasingly express concerns about AI in healthcare, particularly around empathy and human connection, Soul Machines represents a fascinating countertrend—doubling down on making AI more human-like rather than less. Despite research showing that 12% of patients don't trust AI and prefer human interaction even when AI demonstrates cost savings and efficiency, the company is betting that the solution isn't less AI, but more convincingly human AI.

The healthcare applications are particularly striking: clinical trial onboarding specialists, appointment schedulers, billing administrators, and prescription educators—all roles requiring significant patient interaction but often limited by staffing constraints. Early implementations like RAVATAR's AI Healthcare Assistant in a virtual hospital in Italy have demonstrated real-time patient triage and empathetic responses, suggesting these digital workers could genuinely fill healthcare gaps rather than simply adding technological complexity.

However, research reveals a persistent "anti-AI bias" in healthcare—advice labeled as involving AI is consistently rated as less reliable and empathetic compared to human-labeled advice, even when content is identical. Patients specifically worry about AI's inability to show emotion, provide context, or conduct physical examinations—limitations that humanized avatars attempt to address through sophisticated emotional modeling and conversational capabilities.

The "Uncanny Valley" theory poses another challenge, describing the unsettling feeling people experience when digital beings closely resemble humans but aren't quite convincingly realistic. Current AI avatars still struggle with complex medical terminology, diverse accents, and contextual awareness compared to human providers, potentially limiting their effectiveness in nuanced healthcare scenarios.

Yet, the market is growing considerably - valued at $5.9 billion in 2023 and is projected to grow at over 30% annually through 2032. Soul Machines' $135 million in funding and partnerships with Microsoft and SoftBank Vision Fund 2 signal serious institutional backing for human-AI hybrid approaches. Patient surveys consistently show preference for AI as a supportive tool rather than replacement for human physicians, which aligns with Soul Machines' positioning of digital workers as augmentation rather than substitution.

Lila Sciences Raises $235M for "Superintelligence": The Ultimate AI Moonshot or Biotech's Biggest Bubble?

September 16 - Lila Sciences secured $235 million in Series A funding to develop what it calls "scientific superintelligence"—autonomous AI labs that can generate hypotheses, design experiments, and learn from results without human intervention. This brings the Cambridge-based startup's total funding to $435 million in just six months, valuing the company at over $1 billion before producing a single FDA-approved drug, leading some observers to question whether we're witnessing biotech's boldest vision or its most spectacular bubble.

Why it matters: Lila represents the largest bet so far on fully autonomous science—claiming machines can replicate the entire scientific method beyond current AI applications like protein folding. CEO Geoffrey von Maltzahn defines scientific superintelligence as "the ability to conduct the scientific method at a level beyond human intelligence at every step," predicting "100% of science is going to change hands in the next decade." The company claims breakthrough results including genetic medicines that outperform commercial therapeutics, though these achievements remain largely unverified. The "superintelligence" framing raises questions about whether this solves real problems or chases Silicon Valley buzzwords, especially given the industry's track record of overpromising on AI capabilities.

The timing is particularly notable given that this follows a $200 million seed round just six months earlier—an unprecedented pace of capital raising for an unproven technology. Founded within Flagship Pioneering's labs and led by CRISPR pioneer George Church as chief scientist, Lila's approach centers on "AI Science Factories" where artificial intelligence supposedly handles experiment design, execution, and analysis with minimal human oversight.

However, the broader AI drug discovery landscape provides sobering context. Multiple well-funded companies have struggled to translate AI promises into clinical success, with the fundamental challenge remaining that biology's complexityoften gets flattened in computational modelling, reducing our ability to predict biologically accurate behaviour. The company's bold claims about revolutionizing centuries-old scientific methods will face the ultimate test in clinical trials, where patient safety and regulatory scrutiny leave no room for Silicon Valley's "move fast and break things" mentality.