Why Pharma's Next Advantage Will Be Predictive Intelligence

The arrival of GLP-1 therapies has changed the conversation around obesity in ways that few innovations in medicine have achieved. In a remarkably short time, they have shifted expectations among clinicians, patients, healthcare systems and investors alike. For the first time, we are no longer discussing whether obesity is a treatable chronic disease, but how to integrate highly effective therapies into everyday clinical practice.

Yet, as often happens in healthcare, breakthrough therapies also expose the limitations of the systems into which they are introduced.

Today, much of the discussion still revolves around the molecule itself: which treatment delivers greater weight loss, which has a better safety profile or which will become the market leader. These are important questions, but they all assume that the drug is the center of the equation.

I think the real opportunity may lie elsewhere. Not in developing another molecule, but in building the intelligence capable of helping every patient achieve the best possible outcome with the therapies we already have.

Obesity is one of the clearest examples of why medicine cannot be reduced to biology alone. Two patients with almost identical clinical profiles may respond very differently to exactly the same treatment. One remains engaged for years, while another stops after a few months. One gradually incorporates healthier habits into daily life, while another regains weight shortly after treatment ends. One loses primarily fat mass, another may lose too much muscle. The medication is the same. The journey is not.

That difference between treatment and trajectory is where I believe the next generation of digital health will create the greatest value.

For years we have focused on digitizing healthcare by collecting more data. More wearables, more connected devices, more patient-reported outcomes, more dashboards. We have become increasingly good at documenting what has already happened. But chronic diseases do not evolve in snapshots taken every three or six months. They evolve continuously, influenced by behaviour, environment, motivation, social determinants, treatment response and hundreds of variables that rarely fit into a structured medical record.

Perhaps the next challenge is not collecting more information, but learning how to anticipate what comes next.

Dynamic Models

This is why I find the concept of digital twins particularly compelling. They fundamentally change the question we are trying to answer.

Projects such as GlucoTwin, a European initiative exploring digital twins for diabetes, are beginning to demonstrate this shift in thinking. Beyond the technology itself, what I find truly interesting is the underlying concept: moving from retrospective records toward dynamic models capable of learning continuously from each patient's evolution and simulating possible future scenarios before they happen.

In many ways, this represents a transition from descriptive medicine to predictive medicine.

Imagine extending that same logic to obesity. Instead of simply measuring weight loss at the next consultation, clinicians could begin identifying which patients are most likely to discontinue treatment, which ones are entering a period of metabolic adaptation, who may be losing excessive muscle mass, or which behavioral intervention has the highest probability of improving long-term outcomes before problems become clinically apparent.

Some of the most valuable information required to build these predictive models may not come from connected devices at all.

Every clinical consultation contains an extraordinary amount of knowledge that healthcare systems routinely lose. Expectations, fears, emotional eating, family dynamics, financial constraints, anxiety, frustration or declining motivation rarely become structured clinical data, despite often determining whether a treatment ultimately succeeds or fails.

For the first time, advances in artificial intelligence make it possible to transform those conversations into meaningful clinical information. Combined with longitudinal physiological data, they provide something that healthcare has traditionally lacked: context.

And context is what makes prediction possible.

For decades, innovation has been measured primarily by the ability to develop better molecules. That will remain essential. But perhaps the next competitive advantage will come from understanding how those molecules perform in the real world, across millions of individual patient journeys.

Patient support programs, for example, could evolve from educational services into intelligent infrastructures capable of adapting to each person's changing needs. Real-world evidence could become continuous rather than retrospective. Treatment optimization could shift from reacting to problems towards anticipating them. Instead of asking why a patient abandoned therapy six months ago, we could identify the signals that suggested this was likely to happen and intervene while there was still time to change the outcome.

Ultimately, this is not about artificial intelligence replacing physicians, nor about monitoring patients more intensively. Quite the opposite. The objective should be to reduce uncertainty, personalize care and help both patients and healthcare professionals make better informed decisions with less effort, not more.

The companies that lead the next decade of obesity care may not necessarily be those with the most effective drug, but those capable of building the intelligence that allows every treatment to deliver its full potential in everyday clinical practice.

We often say that AI will transform healthcare. I suspect that is only partly true. What truly changes healthcare is using intelligence to redesign how we understand disease, how we anticipate risk and how we accompany people throughout their entire journey.

GLP-1 therapies have opened an extraordinary new chapter in metabolic medicine. The next chapter, I believe, will not be written by pharmacology alone. It will be written by our ability to predict, personalize and intervene before disease takes the next step.

Ultimately, the future of obesity care will not be defined only by changing biology. It will be defined by our ability to understand, anticipate and support human behavior over time.