In the nine years since my last visit to the Docklands’ Excel conference center for the 2017 annual Alzheimer Association International Conference, or AAIC, little of London’s infrastructure has changed. I traveled via the same low-ceilinged, clattering trains with signs advising riders to “mind the gap” and “mind the step.” The center has the same weird jack-o’-lantern-esque façade and forgettable restaurants and hotels. I was even here for the same conference.
Same meeting, in the same place, but what transpired was quite different. In just nine years, how we think about the aging brain has transformed.
One moment captured this. It was a speaker’s remark at a breakfast meeting convened by the Davos Alzheimer’s Collaborative. The 5-year-old organization describes itself as “uniting leading organizations worldwide to build an innovation ecosystem that will accelerate breakthroughs, develop and scale promising solutions and equip every healthcare system to end Alzheimer’s disease everywhere.”
Biotechs are spending billions to cure a rare liver disorder most Americans have never heard of. The contentious race features dueling technologies, patent wars, a broken alliance, and a boiling competition between the U.S. and Chinese drug industries.
The disease, known as alpha-1 antitrypsin deficiency (AATD), is a slow-moving disaster for patients. Thanks to a single misspelled letter of DNA, their livers produce a mutant version of a protein that normally travels through the bloodstream and protects the lung from damage.
It was a skit worthy of an Agatha Christie whodunit reveal: Against the perhaps-too-joyous strains of a jazzy “When The Saints Go Marching In,” a gaggle of health IT professionals and Medicare staffers accompanied a Spirit Halloween-style coffin prop onstage at Health and Human Services headquarters on Monday. Some wore black veils, some held white roses.
The coffin read, “RIP CLIPBOARD,” referring to the decades-old information-gathering staple of health care provider waiting rooms.
“So who killed the clipboard?” asked Zac Jiwa, a federal Medicare official. Health IT officials and experts, reading from cue cards, denied that it was standards implementation, health information exchange networks, electronic health records, or apps — all parts of the various health data initiatives that private industry promised Medicare a year ago that it would improve.
Drug discovery is a high-cost, high-risk endeavor that is under growing pressure from a market increasingly defined by first-mover advantage.
Since the 1950s, the cost of developing new pharmaceuticals has roughly doubled every nine years—a phenomenon known as Eroom’s Law. Today, bringing a new drug to market takes an average of 10-15 years and costs anywhere from $1 billion to $2.5 billion, with failure rates upward of 90%.
AI has become the pharmaceutical industry’s biggest bet on bringing success rates up and timelines down. The faster drug companies can identify, test, and optimize new chemical compounds, the lower the risk of costly failures later in development.
“The main cost in drug discovery is still the clinical phase, so trying to reduce risk and increase your success rates there is obviously hugely beneficial,” says Paul Belcher, director of protein research strategy at global life sciences company Cytiva. “AI is one approach that drug companies hope will not only save time and compress timelines, but enable better quality candidates to reach the clinic.”
Early use of AI in drug discovery shows potential, but also highlights the need for robust and authentic data, as well as integration in lab systems.
AI brings efficiency to the lab
One of the most promising early-stage applications of AI in drug discovery is in hit identification. This involves screening libraries of molecular entities against a disease-related target, such as a protein, to find molecules that bind to it. A successful hit gives researchers a starting point for further testing and refinement, with the aim of eventually developing a viable drug.
Belcher has seen a shift from empirical screening to predictive design: Instead of physically screening libraries, drug companies are now using AI to design drug candidates from scratch and predict how they will interact with disease targets before committing anything to research and development (R&D).
This means companies are no longer limited by how much they can physically screen to identify starting points. “AI does away with that,” says Belcher. “And it can help eliminate low-quality candidates before you have to physically test them, saving time and resources.”
What AI can’t do yet is reliably predict kinetics or developability of new compounds, says Belcher. This means every AI-generated candidate still needs to be validated in the lab.
Traditional screening workflows were built to identify hits at scale, not to profile large numbers of complex candidates in detail. This is placing more pressure on lab teams, who now have to test, characterize, and purify a growing volume of more diverse, AI-generated compounds.
“The current techniques used in hit identification can screen hundreds of thousands, sometimes millions of compounds, using binary or threshold-based techniques producing low-fidelity data—yes-or-no responses,” Belcher explains. “AI can increase the number of hits you get and potentially give you better quality hits as well. That increases demand for higher-throughput, information-rich technologies to then validate and characterize those hits.”
Models need complete, quality data
As AI has accelerated demand for data-rich lab systems, it has also highlighted a fundamental need for better, more complete data.
Many earlier AI models were trained on publicly available datasets and are now hitting what Belcher calls a data wall. Because models have access to the same data, they all reach similar conclusions, with diminishing returns over time. Additionally, the datasets weren’t built with AI in mind, meaning they lack the structure, labeling, and diversity needed to keep models accurate and free of bias.
Publication bias reinforces the problem. “Most publicly available datasets and scientific publications focus exclusively on positive results,” says Belcher. “No one wants to share their failures. This bias is almost like having one hand tied behind your back. AI models can identify patterns associated with success, but they lack the comprehensive understanding of failures that would make predictions more reliable.”
The data Belcher believes would markedly improve models—the failed experiments, the compounds that don’t bind—remains frustratingly difficult to come by. “We often joke that there should be a journal of negative data,” he says. “It’s often buried in lab notebooks, and it’s never used to inform or guide future research.”
This lack of negative data creates a fundamental problem: Without access to a broad range of data, models can’t be adequately trained to avoid bias. “In all machine learning applications, the model’s performance relies heavily on the quality and scope of the training data,” notes Belcher.
Fabrication has also become much easier with AI, compounding concerns around data integrity. Take Western blots, for example. These are part of a standard technique for identifying proteins in blood or tissue samples, and they are among the most common targets for manipulation in biomedical research. Belcher cites research by Dutch microbiologist Elisabeth Bik, who found that almost 4% of biomedical papers contained duplicated or manipulated images. This was back in 2016, before generative AI made fabrication trivial.
“Manipulated or faked data has always been a problem in science, but in the AI world, especially when used to train models, it could have potentially disastrous consequences,” says Belcher. “There needs to be more tools to verify that data is not manipulated.”
Some vendors are starting to tackle this challenge. Belcher points to solutions like Cytiva’s Image Integrity Checker, for instance, which uses secure hash algorithms—the same technology used in blockchain—to detect whether scientific images have been tampered with. “We’re starting to see a lot of interest from publishing houses that want to adopt this as standard because it’s a quick way to ensure that what gets published in the literature is genuine,” he adds.
Autonomous labs could accelerate breakthroughs
Belcher describes the future state of drug discovery as fully autonomous labs that run with minimal human intervention. Foundational to this vision is consistency in data and infrastructure.
These AI-driven dark labs, or labs-in-the-loop, operate around the clock. They cycle through prediction, testing, and optimization, and then feed results back into AI models to guide the next round of experiments. This can improve the success rates of drug candidates entering clinical trials, says Belcher. Better starting points, combined with more rounds of optimization, should result in better candidates with fewer liabilities reaching the clinic.
But automating a lab depends heavily on integration. That means interoperable systems, highly structured and comprehensive datasets, and information flowing easily in and out. Most labs aren’t there yet. “Today, a lot of the instruments in labs are standalone,” Belcher notes. “You can have the best technology in the world, but if it’s a closed ecosystem—if the user can’t get the data out—it doesn’t do any good.”
An integrated infrastructure can enable labs to generate FAIR (findable, accessible, interoperable, and reusable) data at scale. This would not only inform individual lab reports, but could also train subsequent generations of AI models, effectively closing the loop between the computational, AI-driven dry lab and the physical wet lab.
“Our goal is to help scientists and researchers accelerate their breakthroughs and make that future state of autonomous labs a real possibility,” says Belcher. “We want to help them generate reliable data, simplify workflows in discovery, and hopefully enable what they’re working on to become tomorrow’s life-changing therapies, faster and with greater confidence.”
On costs and what comes next
AI-driven drug discovery is still in its early days. Notably, no drug discovered primarily through AI-driven design has yet received full FDA approval—although Belcher expects that to change in the next two to three years.
How big of an impact could AI eventually have on drug discovery? “The holy grail would be full in silico prediction of efficacy and toxicity, eliminating the need for the vast majority of physical wet lab work,” says Belcher. But there are many barriers to this beyond the maturity of the models, including regulatory hurdles and cost challenges.
A Stanford study found that the cost of training frontier AI models has more than doubled every year since 2016, adding more financial pressure to a sector already defined by exceptionally high R&D spend.
Belcher acknowledges the tension, but remains optimistic about what’s ahead. “I think we’ll get to a point where there’s a balance between AI and wet work, from a cost perspective and a risk perspective,” he says. “As long as the cost of compute doesn’t ever outweigh the cost of clinical development, I think AI is going to be an advantage.”
This content was produced by Insights, the custom content arm of MIT Technology Review. It was not written by MIT Technology Review’s editorial staff. It was researched, designed, and written by human writers, editors, analysts, and illustrators. This includes the writing of surveys and collection of data for surveys. AI tools that may have been used were limited to secondary production processes that passed thorough human review.
Imagine a healthcare system made up of multiple AI agents: one that manages symptom assessment, another scheduling, a third insurance, and a fourth pharmacy.
Each is an expert in its domain. But they all have their own distinct knowledge and objectives. Today they can exchange data, but they are not yet able to actually coordinate patient care without a human making the decisions.
“The intelligence is already there. What is missing is the connective tissue that turns four strangers into one team,” explains Vijoy Pandey, senior vice president and general manager of Outshift by Cisco.
This “connective tissue” comes from adding a semantic layer—what Outshift calls the “Internet of Cognition”—that enables agents across domains to work together and, critically, “think” together through shared intent, context, and reasoning.
This semantic layer relies on a connectivity layer beneath it called the “Internet of Agents,” which allows autonomous agents to discover one another, prove identity, and exchange messages across domains.
When used together, they enable “the next step on the road to distributed artificial superintelligence,” says Pandey.
From solo silicon savants to the ‘Internet of Cognition’
For years, the AI industry has been focused on growth. Scaling vertically has led to bigger models, trained on more data with more compute. This has produced the reasoning capabilities that can be like a “brain” for AI agents, which can perceive, reason, and act in digital environments.
While vertical scaling can produce more capable agents perpetually, to enable agentic problem solving across different systems, companies, and platforms the next axis of scale must be horizontal, says Pandey.
Multi-agent systems are already being explored in areas like software engineering, drug discovery, and scientific simulations, but their performances so far have been underwhelming. One study finds a failure rate of between 41% and around 87% when evaluating seven open-source multi-agent systems.
“Connected agents handle coordinated action well; taking a task whose shape they have seen, divided and passed around,” Pandey explains. “What they cannot do is hold a goal in common and reason toward something none of them was trained to solve.”
“The gap is architectural, not a prompting problem,” Pandey adds. “Without the right coordination layer, naive multi-agent setups can perform worse than a single agent. The step change is that team of agents converging on its own, on a new problem, with no human stitching the seams.”
To reach this goal, Pandey says Outshift has built a connectivity layer called AGNTCY, an open-source project now under the Linux Foundation. AGNTCY allows agents across different systems, companies, and platforms to find each other, prove identity, and exchange messages through open, standardized protocols.
And, as Pandey explains, this allows the Internet of Cognition thesis to take a step further. It creates a semantic layer that allows agents to align goals (share intent), pool institutional knowledge and compound memory (share context), and make collective trade-offs (share reasoning).
Pandey likens this progression to that of humans: “For hundreds of thousands of years humans got individually smarter, and the gains died with each person who made them,” he explains. “Around 70,000 years ago that changed, when humans learned to share intent, build cumulative knowledge, and reason collectively. That is when scattered individuals became civilization.
“Agents are at the same threshold. We have built the silicon geniuses and given them agency. What they lack is the layer that let humans go collective,” he says.
First steps to distributed superintelligence
Enabling agents to work collectively rests on three pillars in the tech stack:
Shared intent through cognition state protocols: Cognition state protocols are the semantic handshake that allow agents to agree on a goal before they act and then negotiate toward it. Outshift has created an open-source coordination layer called Mycelium, which organizations can clone and use against their own agents.
“We found that unstructured groups reached a decision about a third of the time across 14 scenarios,” says Pandey, speaking about internal testing. “A coordination protocol that makes agents declare a goal, surface missing information, and resolve conflicts before acting raised that to 93%.”
Shared context through cognition fabric: A cognition fabric is a shared institutional memory and communication mesh that allows agent insight to compound over time rather than resetting each session. This policy-governed context layer solves the problem of “organizational amnesia,” says Pandey, by ensuring the baseline intelligence of the systems only ever goes up.
Shared reasoning through cognitive amplifiers and guardrail technologies: Two kinds of cognition engine can be used together to enable shared reasoning. Cognitive amplifiers speed up shared reasoning and modeling, and guardrail technologies (GATs) create security, cost, and compliance frameworks. Humans are active contributors to this layer, making judgment calls the system routes to them (rather than reviewing outputs after the fact).
Cognition sharing in multi-agent systems can create new risks, including unintended delegations, malicious prompt injections or memory poisoning, or over-privileged agents with access to permissions and data far beyond what their tasks require. Environment-specific controls are therefore needed to protect against unintended actions or consequences.
“Agents have human-like attributes but operate at machine speed and scale,” says Pandey. “Everything we built for twenty years—access control, identity, compliance—was built for humans or machines, not both.”
Continuous Agent Semantic Authorization (CASA)—an open-source reference implementation developed by Outshift—is a GAT that works to ensure agent actions remain securely aligned with the user’s original goal through a process of continuous authorization. It does this by reading what the agent is trying to accomplish then checking each tool request against that task.
In the case of a healthcare system, for example, an agent told to summarize a patient record may start by querying a whole database. This could lead to CASA denying the call, because the request no longer matches the task it was authorized for.
“Today’s controls are scoped to a role or a session not to the task so an agent granted a tool can use it for anything,” explains Pandey. “Roughly 90% of the time, an agent has no way to confirm it is even cleared for the job it was handed.”
Experimentation for cross-domain innovation
When horizontally scaling intelligence in the enterprise, businesses should begin by experimenting with one cross-functional workflow that spans three or four teams and currently needs a human authorizing the handoffs, Pandey advises.
“Stand it up as a small multi-agent system on open, interoperable infrastructure, with a measurable baseline,” he says. “Keep building bigger models, add the horizontal axis on top of them, and change what you measure. Track where one agent’s insight made another agent better—that is the signal the horizontal axis is working.”
By starting to experiment now with intent, context, and reasoning layers, organizations can get ahead of the curve. “The problems are open, and the infrastructure is still being written,” says Pandey. “This is the moment to build it.”
For more information on the Internet of Cognition, visit Outshift.com.
This content was produced by Insights, the custom content arm of MIT Technology Review. It was not written by MIT Technology Review’s editorial staff. It was researched, designed, and written by human writers, editors, analysts, and illustrators. This includes the writing of surveys and collection of data for surveys. AI tools that may have been used were limited to secondary production processes that passed thorough human review.
Engineers and neuroscientists at the University of California (UC) San Diego have developed a soft, wearable fingertip patch that continuously tracks a Parkinson’s disease (PD) patient’s levodopa medication levels by measuring chemicals in their sweat, with no batteries required. Tests in healthy volunteers and in Parkinson’s disease patients showed that measurements generated using the device were comparable to those obtained by standard laboratory blood tests.
The wearable device offers a way to continuously track real-time concentrations of levodopa in the body and could enable doctors to precisely customize daily medication schedules for patients at home.
The research was led by Tamoghna Saha, PhD, a postdoctoral researcher in the lab of Joseph Wang, DSc, professor in the Aiiso Yufeng Li Family Department of Chemical and Nano Engineering at the UC San Diego Jacobs School of Engineering. Saha is co-first author of the team’s published paper in PNAS titled “A wearable patch for continuous levodopa monitoring in sweat: Towards exertion and power-free pharmacodynamic assessment in Parkinson’s disease.” In their paper the authors wrote in summary, “Overall, our easy-to-use, energy-efficient wearable supports real-time, stimulation-free monitoring, potentially enabling at-home dosage adjustments and paving the way for future autonomous closed-loop L-dopa therapeutic system development.”
Parkinson’s disease is the second most common and fastest-growing neurodegenerative disorder worldwide, the author wrote. “While no cure for PD exists, levodopa (L-dopa) is the most effective symptomatic treatment, which is typically administered via oral tablets or capsules, and in advanced cases, through inhaled powder or continuous intrajejunal or subcutaneous infusions.” Prescribing the right dose is challenging: reducing levodopa leaves patients unable to move, while too much triggers severe, uncontrollable jerking movements. Initially, the drug’s effects can last several hours.
But as the disease progresses, the therapeutic window narrows down to two hours. Currently, clinicians must rely on subjective patient diaries to adjust treatment. Unfortunately, these methods fail to catch dangerous dosing gaps. “Precision management of Parkinson’s disease (PD) requires frequent levodopa (L-dopa) dose adjustments, yet current monitoring relies on subjective symptom reporting and infrequent blood testing,” the team continued.
Levodopa monitoring patch showing the assembly of the hydrogel and levodopa sensor with the paper fluidic channel on the fingertip. [Tamoghna Saha.]
Saha and the engineering team developed the new finger patch technology in joint collaboration with the lab of Irene Litvan, MD, MPhil, professor in the department of neurosciences at UC San Diego School of Medicine. The project is part of a longstanding collaboration between the Wang and Litvan teams to develop wearable levodopa monitors that can improve personalized care for people living with PD.
Worn on the fingertip, which is packed with a high density of sweat glands, the patch is equipped with a specially engineered absorbent gel that acts like a sweat sponge. The gel contains a highly-concentrated mixture of salts and benign solvents—and that draws sweat out of the pores, since water naturally flows toward areas with higher salt concentrations. Collected sweat is drawn into a serpentine fluidic channel with a self-powered levodopa biosensor connected to a wireless transmitter.
When levodopa in the patient’s sweat comes into contact with enzymes embedded in the patch it triggers a chemical reaction, which in turn generates a small, measurable voltage. This chemical reaction is what powers the patch. The amount of voltage generated also serves as an indicator of the patient’s levodopa level, such that lower voltage signals low levels, while higher voltage signals high levels.
Unassembled integrated levodopa monitoring patch. [David Baillot (University of California, San Diego, San Diego, CA).]
Experimental results from three to five healthy participants and four individuals with PD indicated that levodopa concentrations in sweat measured by the patch are strongly correlated with blood concentrations measured by high-performance liquid chromatography. The patches captured pharmacodynamic responses and patient-specific levodopa clearance trends that could be used to calibrate dosage estimates for individuals.
The data revealed that individuals with Parkinson’s clear levodopa from their systems significantly faster than healthy individuals. This result explains why a patient’s Parkinson’s symptoms can deteriorate so suddenly, the researchers noted.
This technology could lay the groundwork for a closed-loop system, where a levodopa monitoring patch could communicate with a pump to automatically deliver the precise doses of the drug right when the body needs it, the authors suggested. “This approach establishes a foundation for real-time, at-home therapeutic optimization and advances the development of future closed-loop treatment systems for PD.”
The HIV epidemic remains a national priority in the United States, and the Ending the HIV Epidemic initiative has renewed the call for expanded prevention and treatment strategies capable of reducing new HIV infections by 90% by 2030. Achieving this goal requires robust, integrated data for understanding HIV-related needs, barriers to care, and the effectiveness of interventions. However, despite the existence of numerous publicly available datasets, few integrate multiple domains such as HIV outcomes, social determinants of health, and community-level factors. The lack of unified data and difficulty linking datasets hampers efforts for meaningful cross-domain analyses to tailor HIV management and treatment strategies. The resulting fragmentation constitutes a methodological gap: implementation teams lack replicable guidance for constructing unified HIV and contextual databases from public sources. In this viewpoint, we describe our experience building a unified compilation of publicly available HIV and community data to identify factors influencing HIV outcomes and interventions. The completed database comprises 242 variables drawn from 8 public sources mapped across clinic, zip code, county, and state levels of geography. Rather than simply reporting what we built, we position four core decisions as transferable methodological advances: (1) treating source identification as a bounded phase before construction begins, (2) adopting automated data engineering tools from the outset rather than manual entry, (3) establishing a shared data dictionary before the first variable is entered, and (4) integrating quality control throughout the workflow rather than as a final phase. The build required approximately 350 total project hours and revealed an initial spot-check error rate of approximately 33%, which we attribute primarily to manual data entry. By sharing the approach used to develop this database and making the final resource publicly accessible through the Yale Center for Methods in Implementation and Prevention Science, we aim to reduce barriers to data access and encourage similar data integration efforts. The methodological framework described in this paper is intentionally designed to be replicable with modest resources, and we present it as a practical model for research teams operating without specialized infrastructure. Consolidating HIV, social determinants of health, and contextual variables into a unified data source is a critical step toward enabling deeper, more comprehensive analysis and supporting ongoing efforts to end the HIV epidemic in the US.
<img src="https://jmir-production.s3.us-east-2.amazonaws.com/thumbs/5ea9fe72c2273a99e2615ef582c2512a" />
<strong>Background:</strong> Australia’s mental health care system has been characterized by complexity and fragmentation, as highlighted by numerous reports, commissions, and inquiries. In response, digital mental health care navigation tools have emerged as a promising solution to help individuals locate appropriate mental health services. The rapid proliferation of these tools—without a clear understanding of their definitions and characteristics—risks creating confusion rather than clarity for users. Terms such as “navigation” and “navigators” are often used interchangeably, further complicating the landscape. <strong>Objective:</strong> This study addressed the need for a standardized definition and typology of the characteristics of digital mental health care navigation tools. <strong>Methods:</strong> This study was part of the development of a digital mental health care navigation tool for navigators and planners (MChart). It used a co-design approach using expert-based cooperative analysis, which is a nominal group technique to develop a definition and typology of the characteristics of digital mental health care navigation tools. This process was guided by the Technology Readiness Level for Implementation Sciences framework. The co-design process involved two 2-hour sessions with an expert panel comprising 28 participants, including representatives from mental health planning, primary health care, health care financing and delivery, community-managed organizations, clinical settings (psychiatrists, psychologists, and general practitioners), and consumers. <strong>Results:</strong> The expert panel collaboratively developed a consensus definition of digital mental health care navigation tools, outlining their scope and intended targets. Through the co-design process, the panel identified 157 characteristics of digital mental health care navigation tools. These characteristics were organized into 5 primary domains: type, management, content, design, and quality. The definition and typology characteristics provide a structured framework for understanding and evaluating the diverse range of digital mental health care navigation tools currently available. <strong>Conclusions:</strong> The co-designed definition and typology offer a foundational step toward reducing confusion in the digital mental health care navigation space. This study supports the development of quality standards that can be used to assess and compare existing and future tools. This framework has the potential to guide developers, end users, and policymakers in creating more effective, user-centered navigation solutions within Australia’s mental health care system and internationally.
The world risks a resurgence of the HIV epidemic after international financing to combat the infectious disease suffered a “profound shock” last year, according to a new report from UNAIDS, the United Nations agency.
Overall, government funding declined by more than $1.5 billion to $7.3 billion in 2025 — an 18% decline and the lowest level in nearly two decades. Prevention programs, in particular, have historically relied heavily on donor assistance in most regions, with especially high dependency in sub-Saharan Africa, where it reached 83% two years ago.
Other factors contributing to the worrisome outlook are high debt burdens in countries most affected by the disease and backsliding on human rights and gender equality, the report noted.
Researchers headed by a team at the Institute of Metabolic Science, University of Cambridge, have solved the mystery of why both stimulating and blocking a particular receptor, or switch, in the brain can help people lose weight. Their study in mice indicated that the answer lies in where the receptor, called GIPR, is located. The results showed that stimulating this switch in the brainstem suppresses appetite, while the same effect can be achieved by blocking it in the hypothalamus. The researchers say their findings could help in the development of more effectiveness therapeutic strategies.
Jo Lewis, PhD, at the Institute of Metabolic Science at the University of Cambridge, said, “Understanding which brain circuits respond to these medications—and how they do so—could help us design better drugs that produce more weight loss with fewer side effects, and which might work in combination with other obesity medicines to even greater effect.” Lewis is first author of the team’s published paper in Nature Metabolism, titled “Distinct brain regions mediate regulation of food intake in response to GIPR agonism or antagonism.”
More than a billion people worldwide are living with obesity, which increases the risk of diseases such as type 2 diabetes (T2D), cardiovascular disease (CVD) and cancer. Weight loss can help mitigate these complications, but losing weight through diet and exercise alone can prove challenging.
In the past few years, a new generation of weight loss drugs has emerged that target particular receptors in the brain, reducing appetite and leading to weight loss, as well as helping to control blood sugar levels. Several of these drugs, such as Wegovy and Ozempic, work by stimulating the glucagon-like peptide 1 receptor (GLP-1R).
Other weight loss drugs act on both this receptor and on GIPR. “The development of dual agonists for the glucagon-like peptide-1 receptor (GLP-1R) and glucose-dependent insulinotropic polypeptide receptor (GIPR) has been a landmark moment in the treatment of type 2 diabetes and obesity,” the authors wrote.
However, some drugs, such as Mounjaro and Zepbound, stimulate GIPR, while others, such as the Phase III-stage MariTide, block it. Why these opposite actions have the same result has puzzled scientists. “… for reasons that are incompletely understood, in preclinical and clinical studies, adding either a GIPR agonist or GIPR antagonist to GLP-1R agonism causes additional weight loss,” the team continued. “There is emerging evidence that GIPR agonism and antagonism exert their paradoxically similar effects on weight loss via distinct neuronal populations.”
The investigators’ newly reported preclinical research has now shown that the two different types of GIPR drugs act on distinct regions of the brain, but also that they can boost weight loss when combined with certain GLP-1-based weight-loss drugs. For their reported study the team turned to genetically engineered mice and selectively removed GIPR from different parts of the brain to see which regions were responsible for the effects of the obesity drugs.
One group of mice lacked GIPR in the brainstem—the area at the base of the brain, just above the spinal cord, involved in appetite and nausea. A second group lacked GIPR in the hypothalamus, a major center controlling hunger and body weight. A third, control group included normal, unmodified mice. The researchers treated the mice with various combinations of a GIPR agonist (which activates the receptor), a GIPR antagonist (which blocks the receptor) and a GLP-1 drug, and measured food intake, body weight, fat mass, glucose control and brain activity.
“We knock out Gipr in either the area postrema (AP) or hypothalamus of mice (GiprAP-KO and Giprhypo-KO, respectively) and compare body weight and food intake responses to GIPR agonists and antagonists, alone and in combination with the GLP-1R agonist liraglutide,” they wrote in summary.
By comparing the responses of normal mice with mice lacking GIPR in different brain areas the investigators showed that GIPR agonists act on the brainstem to suppress appetite and reduce weight. They then showed that GIPR antagonists help weight loss by acting on this receptor in the hypothalamus, where they release a “brake” that otherwise limits the brainstem’s ability to respond to signals telling us we are full. Blocking GIPR also appeared to boost the effect of emerging new drugs targeting the amylin receptor—such as cagrilintide (Cagri)—suggesting that GIPR antagonists could potentially be used to strengthen several types of anti-obesity medicines.
“Overall, our results suggest that the AP is responsible for the appetite-suppressing effects of GIPR agonism but that GIP receptors in the hypothalamus underlie the ability of GIPR antagonism to enhance the weight loss effects of GLP-1R and amylin receptor agonists,” they stated. “GIPR antagonism and Giprhypo-KO also sensitize to cagrilintide-induced weight loss.
The findings explain why drugs such as MariTide, which combines GIPR antagonism with GLP-1 receptor agonism, are effective, and suggests how to design even better combination therapies. And as the authors noted, “Future work is still, however, required to identify the neuronal networks underlying GIPR interactions in the AP and hypothalamus and their crosstalk with other appetite-regulating circuitry.”
Lewis said the work strengthens the idea that the brain is central to obesity treatment, commenting, “Obesity drugs are not acting simply on the gut or pancreas. Instead, they have important effects on specific, identifiable brain circuits that regulate appetite and food intake.”
![Levodopa monitoring patch showing the assembly of the hydrogel and levodopa sensor with the paper fluidic channel on the fingertip. [Tamoghna Saha.]](https://www.genengnews.com/wp-content/uploads/2026/07/Low-Res_26-10453-2-232x300.jpg)
![Unassembled integrated levodopa monitoring patch. [David Baillot (University of California, San Diego, San Diego, CA).]](https://www.genengnews.com/wp-content/uploads/2026/07/Low-Res_2026-10453-3-300x200.jpeg)
