An AI app that automatically recalibrates insulin doses for Type 2 diabetes patients has cleared the FDA — a regulatory milestone that could reshape how millions of people manage one of the world’s most prevalent chronic conditions. The clearance belongs to DreaMed Diabetes, a Tel Aviv-based digital health startup whose Advisor Pro platform is now authorized to autonomously recommend insulin adjustments without requiring a physician to intervene at every decision point. For the roughly 30 million Americans living with Type 2 diabetes who depend on insulin therapy, that distinction matters enormously.
As Calcalist Tech reported, DreaMed’s clearance covers the use of its algorithm to guide basal and bolus insulin titration — the two core levers clinicians adjust when trying to keep blood glucose in range. The company has been developing its AI dosing engine for years, with earlier iterations focused on Type 1 diabetes and closed-loop insulin pump systems. Expanding into Type 2 represents a significant commercial pivot toward a far larger patient population. This kind of AI-in-the-loop approach to chronic disease management is part of a broader shift in digital health — one our coverage of AI healthcare costs has tracked closely as insurers and providers grapple with its implications.
What the Algorithm Actually Does
Advisor Pro ingests patient glucose data — typically from continuous glucose monitors — and uses a machine learning model trained on large clinical datasets to compute personalized dose recommendations. Rather than applying a static titration protocol, the system adapts to each patient’s individual glucose patterns over time, learning how their body responds to prior dose changes. That adaptive loop is the core technical claim: the algorithm gets more accurate the longer it runs on a given patient’s data.
The FDA clearance is granted under the De Novo pathway, which the agency uses for novel, low-to-moderate-risk devices without a direct predicate — a signal that regulators view the system as genuinely new rather than a modest iteration on existing software. De Novo clearance also establishes a regulatory blueprint that competitors can use in future submissions, which means DreaMed’s approval effectively opens a lane for the broader category of AI insulin titration tools. The company has previously reported clinical study results showing that patients using its algorithm achieved improved time-in-range metrics compared to standard care protocols, though exact benchmark figures from the Type 2 trial were not disclosed in the initial reporting.
Market Reach and What Comes Next
Type 2 diabetes dwarfs Type 1 in sheer scale. The International Diabetes Federation estimates that more than 537 million adults globally live with diabetes, the overwhelming majority with Type 2, and insulin dependency within that group runs into the tens of millions. For DreaMed, cracking this market means moving from a niche clinical tool into a product with mass-market potential — one deployable through endocrinology practices, primary care networks, and eventually telehealth platforms where specialist access is limited.
The clearance also positions DreaMed competitively against a growing field of digital therapeutics companies trying to automate chronic disease management. Insulin titration has historically been one of the most labor-intensive aspects of diabetes care — patients call in glucose logs, clinicians make manual adjustments, and the feedback loop can take days. An FDA-cleared algorithm that handles that loop continuously and autonomously compresses that cycle to near-real-time. The remaining questions are adoption velocity and reimbursement: whether payers will cover AI titration tools at scale, and whether clinicians will trust algorithmic recommendations enough to deploy them broadly. DreaMed’s regulatory win answers the hardest question first. The commercial ones are still live.
