If you were still thinking that the AI race in drug discovery was mainly about better models, this week’s news will give you a bit of a different perspective.

On Feb 18, Merck announced a strategic collaboration with Mayo Clinic. Merck will integrate Mayo Clinic Platform’s de-identified clinical and genomic datasets with its AI-enabled “virtual cell” systems.

Through Platform Orchestrate, it gains access to multimodal data across EHR, imaging, and molecular layers, initially focused on IBD, atopic dermatitis, and multiple sclerosis.

For Mayo, this is its first partnership of this scale with a global biopharma company.

At the same time, Mayo expanded work with Siemens Healthineers on AI-enabled MRI protocols and digital twins.

To drive the data-hungry trend, earlier this month, Bristol Myers Squibb partnered with Evinova by pooling their operational data to optimize trial design using AI-native platforms.

What we are witnessing is a change in leverage. It is no longer enough to build models.

The real advantage is controlled access to interoperable, privacy-governed clinical data at scale.

NEW SECTION: There’s been an amazing run (and announcements) of AI Drug Discovery companies that have IPO’d. I expect more to come this year. I just never realized how many that are already publically traded! I’ll continue to improve on this section over the next little bit.

For now, did you know there’s 15 public AI Drug companies?

Hypertension affects 1.4 billion adults worldwide and remains the leading global risk factor for death. In China alone, an estimated 245 million adults live with high blood pressure.

Yet control rates remain low, and poor medication adherence is a major reason. Nearly one in four patients never even start prescribed therapy.

A new study using nationally representative data from the China Health and Retirement Longitudinal Study (CHARLS) tackles a practical question: can we predict who is most likely to stop taking their medication?

What the researchers built: The team analyzed 2,773 adults aged 45 and older with hypertension. More than half, 53.2%, showed low adherence across two survey waves.

Instead of relying on a single self-report snapshot, the researchers defined adherence longitudinally, tracking consistency over time.

Seven machine learning models were tested. XGBoost performed best, with an AUC of 0.828, accuracy of 0.726, and F1 score of 0.713 on the test set. In simple terms, the model showed strong ability to distinguish between patients likely to stay on treatment and those at risk of stopping.

The researchers then made the model interpretable using SHAP analysis and deployed it as a web-based tool for real-time risk assessment.

Who is most at risk: Some findings may surprise clinicians. Patients with multiple chronic conditions, cardiometabolic multimorbidity, overweight or obesity, older age, depression, and urban residence were more likely to adhere.

Smoking, living in western regions, and being employed were linked to non-adherence.

Adherence is shaped not just by disease severity but by social, behavioral, and economic context.

An interpretable risk tool allows primary care providers to identify high-risk patients early and direct limited resources toward education, digital reminders, and multidisciplinary support where they are most needed.

Sources: [Research Article]

What is it: No details have been named yet, but from their website, their closest candidate (IND-enabling) is the one most likely moving to Phase 1 (again unconfirmed). If my guess is right, then it’s their CGX-926.

CGX-926 is a pill designed for people with a genetic form of severe early-onset obesity.

In some people, a key “appetite control” protein in the brain called MC4R is made incorrectly because of a gene mutation. When that happens, the protein can’t reach the surface of cells to do its job, which leads to constant hunger and rapid weight gain.

CGX-926 aims to fix how that protein is folded so it can work properly again.

Will keep you all updated when they make a public announcement!

What is it: NCT07359846 and NCT07276724 is a large global Phase 3 trial testing GB-0895 and for people with severe asthma that isn’t controlled by standard inhalers.

For more info on all AI Drugs in Clinical Trials. NOTE: It’s still under development!

Select the right answer! (See explanation below and source)

According to the WHO, cardiovascular diseases account for approximately 32% of all global deaths, making them the leading cause of death worldwide. [Source]