Goldman Sachs analyst calls AI a key tool in the fight against diabetes

Goldman Sachs analyst David Roman, who covers medical technology and healthcare IT systems, said that artificial intelligence can play a "significant" role in helping diabetes patients — and major industry players are already moving from experiments to ready-made products.

Diabetes as a testing ground for AI

Type 2 diabetes is one of the few chronic diseases where patients make dozens of medical micro-decisions daily: what and how much to eat, what insulin dose to inject, when to expect a spike or drop in glucose. Each requires data and calculations. This is where AI can take over the routine and substantially reduce the cognitive load on the patient.

Abbott and Dexcom — the world's two largest manufacturers of continuous glucose monitoring systems — are already actively integrating machine learning algorithms into disease management. The technology allows automatic recognition and counting of carbohydrates in food from a photo of a plate, then calculating the optimal insulin dose without manual data entry by the patient. Traditionally, Type 1 diabetics had to independently assess the composition of each meal and manually adjust dosing. This requires constant concentration and often leads to errors that, at best, result in poor health, at worst, hospitalization.

Goldman Sachs Conference: Analysts are optimistic

Roman spoke immediately after Goldman Sachs' annual medical technology conference — one of the key industry forums for investors and companies. He shared impressions jointly with Bloomberg reporter Madison Mueller, who specializes in healthcare.

"AI can play a significant role in helping patients keep diabetes under control," —

David Roman, Goldman Sachs analyst.

According to Roman, he left the conference "very inspired" by innovative developments in the industry. Participants discussed a wide range of applications: from improved early diagnosis to personalized therapeutic recommendations. However, analysts single out diabetes management as one of the most mature and closest to mass adoption — unlike many other medtech concepts, Abbott and Dexcom products are already sold and used by millions of patients.

What AI can do in diabetes technology

Modern AI-based systems offer diabetes patients several key capabilities:

• Food recognition from photos and automatic carbohydrate counting
• Forecasting glucose spikes and drops based on eating history, physical activity and sleep
• Automatic or advisory correction of bolus insulin dose
• Early warning of hypoglycemia before dangerous blood sugar drop
• Seamless synchronization with insulin pump without manual intervention

Abbott with FreeStyle Libre devices and Dexcom with its CGM G-series sensor line compete precisely in these parameters: algorithm accuracy, response speed, and depth of integration with wearable devices. The competition also includes the market for so-called "artificial pancreases" — systems where sensor and pump work in autonomous connection without patient involvement.

What this means

According to the International Diabetes Federation, more than 530 million people worldwide have diabetes, and by 2045 this number could exceed 780 million. If AI systems prove clinical effectiveness in long-term studies, chronic disease management will shift from daily patient effort to automated monitoring — and the number of complications and hospitalizations will decrease. For medtech companies, this means a huge market with far from exhausted potential.