Lifeboat Foundation

We present the first open-source benchmark to evaluate LLMs in their ability to operate as agents in simulated clinical environments. Diagnosing and managing a patient is a complex, sequential decision making process that requires physicians to obtain information—such as which tests to perform—and to act upon it. Recent advances in artificial intelligence (AI) and large language models (LLMs) promise to profoundly impact clinical care. However, current evaluation schemes overrely on static medical question-answering benchmarks, falling short on interactive decision-making that is required in real-life clinical work. Here, we present AgentClinic: a multimodal benchmark to evaluate LLMs in their ability to operate as agents in simulated clinical environments. In our benchmark, the doctor agent must uncover the patient’s diagnosis through dialogue and active data collection. We present two open benchmarks: a multimodal image and dialogue environment, AgentClinic-NEJM, and a dialogue-only environment, AgentClinic-MedQA. Agents in AgentClinic-MedQA are grounded in cases from the US Medical Licensing Exam~(USMLE) and AgentClinic-NEJM are grounded in multimodal New England Journal of Medicine (NEJM) case challenges. We embed cognitive and implicit biases both in patient and doctor agents to emulate realistic interactions between biased agents. We find that introducing bias leads to large reductions in diagnostic accuracy of the doctor agents, as well as reduced compliance, confidence, and follow-up consultation willingness in patient agents. Evaluating a suite of state-of-the-art LLMs, we find that several models that excel in benchmarks like MedQA are performing poorly in AgentClinic-MedQA. We find that the LLM used in the patient agent is an important factor for performance in the AgentClinic benchmark. We show that both having limited interactions as well as too many interaction reduces diagnostic accuracy in doctor agents.

Psychological stress disrupts communication between the brain and gut, reducing protective mucus production in the intestines. This weakens the immune system and makes the body more vulnerable to infections, but probiotics may help restore balance.

Have we created an operating system for life? How close are we to cloning humans, and what would that even look like?

You’re in for a fascinating episode as the line between science and science fiction gets blurred. My guest is microbiologist and geneticist Andrew Hessel, the CEO and Founder of The Genome Project-Write, and author of \.

How can machine learning help individuals with type 1 diabetes (T1D)? This is what a study presented at this year’s Annual Meeting of the European Association for the Study of Diabetes (EASD) hopes to address as a team of researchers have developed a system using machine learning capable of managing blood sugars levels with such proficiency that those using system were able to lead lives far more active than the average T1D patient.

For the study, the researchers developed the AID system, which uses closed-loop technology that delivers insulin based on readings from the machine learning algorithm, resulting in a 50-year-old man, a 40-year-old man, and a 34-year-old woman with T1D being able to run hours-long marathons in Tokyo, Santiago, and Paris, respectively. This study holds the potential to help develop better technology capable of allowing T1D diabetes patients to stay in shape without constantly fearing for their blood sugar levels, which can lead to long-term health problems, including hyperglycemia, nerve damage, or a heart attack.

“Despite better systems for monitoring blood sugars and delivering insulin, maintaining glucose levels in target range during aerobic training and athletic competition is especially difficult,” said Dr. Maria Onetto, who is in the Department of Nutrition at the Pontifical Catholic University of Chile and lead author of the study. “The use of automated insulin delivery technology is increasing, but exercise continues to be a challenge for individuals with T1D, who can still struggle to reach the recommended blood sugar targets.”

St. Jude’s Liz Kellogg, Ph.D., uses cryo-EM to study programmable transposons for targeted gene therapies, including potential new treatments for cystic fibrosis.

The future of CRISPR involves clinical trials focused on treatment for blood diseases and cancers, cardiovascular disease, T1D, and HIV.

A GROUNDBREAKING cancer vaccine could stop tumours growing in patients with advanced disease, researchers say.

Designed to prime the body to recognise and fight cancer cells, the jab could stimulate the immune system to help treat the disease more effectively, early trial results show.

Researchers described the results as “an important first step” in developing a new treatment for people with advanced cancers.

Cleveland Clinic researchers have made a significant discovery about how the gut microbiome interacts with cells to cause cardiovascular disease. The study published in Nature Communications found that phenylacetylglutamine (PAG), produced by gut bacteria as a waste product, then absorbed and formed in the liver, interacts with previously undiscovered locations on beta-2 adrenergic receptors on heart cells once it enters the circulation.

PAG was shown to interact with beta-2 adrenergic receptors to influence how forcefully the heart muscle cells contract—a process that investigators believe contributes to heart failure. Researchers showed mutating parts of the beta-2 adrenergic receptor that were previously thought to be unrelated to signaling activity in preclinical models prevented PAG from depressing the function of the receptor.

This is the latest in a series of investigations into PAG, led by Stanley Hazen, MD, Ph.D., chair of Cardiovascular and Metabolic Sciences in Cleveland Clinic’s Lerner Research Institute and co-section head of Preventive Cardiology. Dr. Hazen’s lab previously demonstrated that elevated circulating levels of PAG in subjects are associated with heightened risk for developing heart failure, and lead to worse outcomes for patients with heart failure.

A new study has uncovered significant differences in how male and female mice process threats, even as they exhibit similar behavioural responses. The discovery suggests that including both male and female subjects in neuroscience research will lead to more accurate conclusions and ultimately better health outcomes. Understanding the influence of sex on brain function can help explain why males and females develop certain psychiatric disorders at different rates or with different symptoms, the researchers said. ‘Unless we thoughtfully and rigorously integrate sex into biomedical research, a huge amount of the population may be underserved by scientific knowledge,’ said McGill University Associate Professor and Canada Research Chair in Behavioural Neurogenomics Rosemary Bagot, who led the study. ‘Our work shows that sex is an important variable to consider, even if initial observations don’t necessarily show clear sex differences,” said Bagot. “If males and females are using different brain circuits to solve similar problems, they may be differently vulnerable to stress and respond differently to treatments.’ How brain circuits process threats and cues The study focused on two related brain circuits and their roles in processing information about threats and the cues that predict them. The researchers trained mice to recognize a sound that signalled a threat and another sound that meant safety. By observing brain activity, the team saw how communication between different brain areas processed these signals. Then, they temporarily turned off each brain connection to see how it affected the mice’s reactions, helping them understand how the brain handles threats. ‘We found that even though male and female mice respond similarly to threats, the brain circuits underlying these responses are not the same,’ Bagot said. For female mice, a connection between two specific brain areas (the medial prefrontal cortex and the nucleus accumbens) played a key role. The study found that in male mice, a different connection (between the ventral hippocampus and the nucleus accumbens) was more important for handling the same situation. It was previously assumed that similar behavior meant similar brain function. Now, the researchers are exploring how sex impacts brain circuits in processing threats, focusing on the role of sex hormones and different learning strategies. This research is supported by funding from CIHR. About the study Sex-biased neural encoding of threat discrimination in nucleus accumbens afferents drives suppression of reward behavior by Jessie Muir, Eshaan Iyer et al., was published in Nature Neuroscience.