Sydney researchers uncover a hidden limit to AI predictability

Study offers pathways to understand instability in AI systems

University of Sydney

Facebook

XLinkedInWeChatBlueskyMessageWhatsAppEmail

image: 

Each coloured point represents a starting setup for an AI model: blue leads to one training outcome, orange to another. Together, they form a striking butterfly-like pattern, with orange points scattered throughout blue regions and blue points throughout orange regions. Moving clockwise, successive close-ups of the areas indicated by the black lines reveal this intricate mixing at ever finer scales. This shows how even tiny differences in the starting setup can lead to different training outcomes, making them difficult to predict and reproduce. 

view more 

Credit: Professor Pulin Gong/The University of Sydney

Researchers at the University of Sydney have identified a fundamental mathematical reason why artificial intelligence systems can learn different behaviours even when trained under almost identical conditions, a discovery that could help improve the reliability and trustworthiness of AI technologies.

The study by Professor Pulin Gong from the School of Physics and his PhD student Andrew Ly, has been published in Nature Machine Intelligence, one of the world’s leading journals in artificial intelligence research.

As AI systems become increasingly embedded in healthcare, business, research and everyday life, ensuring they behave consistently and predictably has become a major challenge. The new research suggests that some forms of AI unpredictability are not due to engineering flaws or inconsistencies computer hardware, but in the underlying mathematics of how neural networks learn.

“Researchers have often attributed differences in AI behaviour to external sources of randomness, such as the order in which training data is presented or tiny numerical variations introduced by hardware,” Professor Gong said.

“Our findings point to something deeper. Even when these factors are removed, extremely small differences in a model’s starting conditions can still lead to different outcomes.”

Neural networks, the technology behind today’s most advanced AI systems, learn by adjusting millions or even billions of internal parameters during training. The Sydney team investigated why networks sometimes arrive at different solutions despite using the same data, training procedure and deterministic computing environment.

Andrew Ly said: “Imagine two AI systems trained to detect cancer using the same medical images and the same training procedure. Tiny differences in their initial parameters could lead them to learn different behaviours. Our research helps explain how such differences arise and might inform methods for making these systems more consistent.”

Understanding the source of these differences could guide the design of more dependable systems and reduce the cost and uncertainty associated with developing large-scale models.

AI TRAINING ‘MAP’ IS A FRACTAL

To explain the phenomenon, the researchers use the analogy of a map. Imagine every possible starting point for training as a location on that map, coloured according to the outcome it eventually produces. Regions leading to the same outcome form what mathematicians call a “basin of attraction”.

The team found evidence that these basins are organised as highly intricate fractals known as “riddled basins”. In such a landscape, every point lies arbitrarily close to points leading somewhere else. No matter how far you zoom in, the different regions remain interwoven.

The implication is striking. Increasing the precision of a network’s starting conditions may do little to improve confidence about where training will end up. In one larger image-recognition network, making the starting settings 10 billion times more precise produced no measurable improvement in reproducibility. Altering that network’s starting settings by the smallest amount a computer can register, a single “bit”, sent training to a different internal configuration in close to half (about 48 percent) of cases.

To test the idea, the researchers trained approximately 500,000 neural networks, analysing systems used for tasks including image recognition and text processing. Evidence consistent with “riddled basins” fractal structure was found across a wide range of network architectures, datasets and training methods.

Importantly, the effect remained even in fully deterministic environments where repeated training runs would otherwise produce identical results.

MODEL HELPS EXPLAIN DIFFERENT AI OUTCOMES

The findings may help explain several longstanding observations in AI research, including why independently trained models often behave differently and why both large and tiny disturbances can sometimes have similar effects on outcomes.

The work also revealed a surprising trade-off between performance and reproducibility. The settings enabling networks to learn most effectively also made their outcomes harder to predict and reproduce. However, reducing the ‘learning rate’ – the size of the step a network takes each time it adjusts itself – made the results reproducible, but the networks learned less effectively.

Rather than viewing unpredictability as a problem that can be completely engineered away, the researchers argue it may represent a fundamental feature of how many neural networks operate. The study introduces new tools for measuring this sensitivity and points towards future approaches, such as redesigning network architectures to reduce the mathematical conditions that give rise to instability.

“AI is becoming increasingly important across society,” Professor Gong said. “To build trustworthy systems, we need not only accurate predictions but also confidence that those predictions can be reproduced. This research provides a new way of understanding the limits of predictability in AI, and how we might overcome them.”

DOWNLOAD images of fractals, photos of researchers and the paper at this link.

INTERVIEWS

Professor Pulin Gong | pulin.gong@sydney.edu.au

MEDIA ENQUIRIES

Marcus Strom | marcus.strom@sydney.edu.au | +61 474 269 459
Outside of work hours: please call +61 2 8627 0246 (directs to a mobile number) or email media.office@sydney.edu.au.

RESEARCH

Ly, A. and Gong, P. ‘Fractal basin geometry and the limits to predictability and reproducibility in deep learning’ (Nature Machine Intelligence 2026). DOI: 10.1038/s42256-026-01336-3

DECLARATION

The authors declare no competing interests. The study was supported by the Australian Research Council.

---

Journal

Nature Machine Intelligence

DOI

10.1038/s42256-026-01336-3

Method of Research

Computational simulation/modeling

Subject of Research

Not applicable

Article Title

Fractal basin geometry and the limits to predictability and reproducibility in deep learning

Article Publication Date

5-Nov-2026

 

A putty-like electrode could improve vocal rehabilitation

McGill University researchers said their device could make wearable health sensors more comfortable, practical and effective for patients with voice disorders, among others

McGill University

Facebook

XLinkedIn
MessageWhatsAppEmail

image: 

A new sensor developed by McGill University researchers could help track breathing, swallowing and speaking during rehabilitation after a stroke or in neurodegenerative diseases such as Parkinson's.

view more 

Credit: Care of Nicole Li-Jessen

A team of McGill University researchers has developed a putty-like electrode that moulds to the skin. It could improve treatment and monitoring of voice disorders and upper airway diseases. In a proof-of-concept study, the sensor matched a standard clinical electrode for noise and stability while recording activity from small neck muscles.

 The sensor could help track breathing, swallowing and speaking during rehabilitation after a stroke or in neurodegenerative diseases such as Parkinson's, the researchers said.

“As a speech pathologist, I have muscle-strengthening exercises I can give to patients, but we can only see them 30 minutes a week, or even monthly,” explained study co-author Nicole Li-Jessen, Professor in the School of Communication Sciences and Disorders and an Associate Investigator in the Translational Research in Respiratory Diseases Program at the Research Institute of the McGill University Health Centre.  “If we have a muscle sensor that can be used as biofeedback for the patient while they practice at home, they can know if they are using the muscle correctly.”

Small muscles, tricky signals

Current electromyography (EMG) sensors are effective for large areas like the legs and arms, but developing practical options for the neck has been challenging because the muscles are small and the surface is highly curved, the researchers said.

Additionally, most commercial sensors require a gel to adhere to the skin and conduct physiological signals. Wet sensors can eventually get messy and dry out, creating more “noise” in the signal, Li-Jessen said.

The new sensor is made of a soft, silicone-based putty that conducts electricity like a metal via tiny silver wires. Because it is dry, it can be applied closely to the skin, meaning there is little air between the skin and the sensor to interfere with the signal.

“Its surprising feature is that stretching it makes it conduct electricity even better,” said senior author Jianyu Li, Associate Professor of Mechanical Engineering. “As the putty stretches, the wires rearrange, helping the electricity keep flowing.”

To test the device, the researchers shaped the putty into neck electrodes and recorded muscle activity as three volunteers completed vocal tasks designed to induce fatigue, so that they could also test whether the device worked at low muscle activity signals.

“The human study is a proof of concept, not a clinical trial,” Li-Jessen noted. “It shows that the electrode records clean signals from small neck muscles, where standard electrodes struggle.”

Larger studies and regulatory approval are required before the device is ready for clinical use, the researchers said.

About this study

“Mechanoadaptive Nanowire Viscoplastic Composites with Strain-Enhanced Conductivity for Reprocessable Bioelectronic Interfaces,” by Ran Huo, Shengyun Huang, Dixiang Gong, Xin Zhou, Heng Zhu, Pengyun Zhang, Hideaki Okuyama, Guangyu Bao, Xuan Li, Aram Bahmani, Lisa Martignetti, Zixin Zhang. Hongyu Hou, Changhong Cao, Yixun Cheng, Wei Gao, Dongling Ma, Nicole Y. K. Li-Jessen and Jianyu Li, was published in Science Advances.

The study was funded by the National Institutes of Health of the United States, Natural Sciences and Engineering Research Council of Canada, the New Frontiers in Research Fund and Canada Research Chairs Program.

---

Journal

Science Advances

DOI

10.1126/sciadv.aef5042

Method of Research

Experimental study

Subject of Research

Not applicable

Article Title

Mechanoadaptive nanowire viscoplastic composites with strain-enhanced conductivity for reprocessable bioelectronic interfaces

 

Multi-society panel endorses vaccine recommendations to protect respiratory health

The panel’s purpose is to provide clinicians focused on respiratory health with a single, up-to-date reference for immunization against influenza, RSV and COVID- during the current transition in national immunization advisory processes

American Thoracic Society

Facebook

XLinkedInWeChatBlueskyMessageWhatsAppEmail

image: 

Four societies endorse recommendations for flu, COVID-19 and RSV vaccines in adults and children.

view more 

Credit: ATS

NEW YORK, NY – Nov. 5, 2026 – Hoping to provide much-needed guidance to clinicians treating patients with lung disease and to promote respiratory health, the Vaccine Guidance Review Panel, a multi-medical society effort, announced its endorsement of influenza, RSV and COVID-19 recommendations for adults and children.

Experts from the American Thoracic Society; the American Academy of Allergy, Asthma & Immunology (AAAAI); the American College of Chest Physicians (CHEST); and the Society of Critical Care Medicine (SCCM) reviewed and unanimously endorsed the vaccination-development process and resulting guidance from the American Academy of Pediatrics (AAP), American Academy of Family Physicians (AAFP), American College of Obstetricians and Gynecologists (ACOG), and Infectious Diseases Society of America (IDSA), and reached consensus guidance aligned with Advisory Committee on Immunization Practices (ACIP) recommendations issued prior to June 2025.

The changes in CDC and ACIP processes in 2025 resulted in potential harmful updates to the vaccine schedule. Seeing this alarming departure from evidence-based vaccine guidance, the multi-society panel was moved to collaborate to protect respiratory health, as well as ensure, among other measures, that COVID-19 vaccine recommendations aligned with the AAP, ACOG, AAFP and the IDSA.

“Clinicians and their patients with serious lung disease should not have to navigate a labyrinth of confusion, particularly as we face the upcoming respiratory virus season,” said ATS President Michelle Ng Gong, MD, ATSF. “The evidence-based vaccine recommendations endorsed by ATS, AAAAI, CHEST and SCCM provide clinicians with confidence and practical vaccine guidance when treating patients at the highest risk of severe illness.”

“AAAAI strongly supports the adoption of evidence-based vaccine recommendations that help health care professionals make informed decisions and provide the best possible care for their patients. These recommendations reflect the importance of scientific evidence, clinical expertise, and a shared commitment to protecting public health. We are grateful to partner with ATS, SCCM and CHEST to promote support of these recommendations,” said AAAAI President Carla Davis, MD.

“The work we are doing at the bedside is infinitely more effective when supported by a foundation of preventative medicine,” said American College of Chest Physicians President Neil Freedman, MD. “By ensuring clarity of unified recommendations from the voice of the community that cares for those with chronic and acute respiratory diseases with the broader physician community, this multi-society partnership aims to provide education and protect equitable access to preventative medicine and care.”

“As critical care professionals, we see firsthand the devastating consequences that respiratory viral illnesses can have on patients and families. Vaccination remains one of our most effective tools to prevent severe disease. We strongly support the influenza, RSV, and COVID-19 vaccination recommendations from the AAP, ACOG, AAFP, and IDSA,” SCCM President Cherylee W. J. Chang, MD.

For more details on vaccine recommendations for different patient populations, visit the ATS vaccine resources page.

###

About the American Thoracic Society

Founded in 1905, the American Thoracic Society is the world's leading medical society dedicated to accelerating the advancement of global respiratory health through multidisciplinary collaboration, education, and advocacy. Core activities of the Society’s more than 22,000 members are focused on leading scientific discoveries, advancing professional development, impacting global health, and transforming patient care. Key areas of member focus include developing clinical practice guidelines, hosting the annual International Conference, publishing four peer-reviewed journals, advocating for improved respiratory health globally, and developing an array of patient education and career development resources.

 

Please visit our website to learn more. Join the conversation and our community on Bluesky, Twitter/X, Facebook, Instagram and LinkedIn.

 

About American Academy of Allergy, Asthma & Immunology (AAAAI)

The American Academy of Allergy, Asthma & Immunology (AAAAI) is the leading membership organization of more than 7,100 allergists, asthma specialists, clinical immunologists and other professionals with a special interest in the research and treatment of allergic and immunologic diseases. Established in 1943, the AAAAI is the go-to resource for patients living with allergies, asthma and immune deficiency disorders.

 

About American College of Chest Physicians (CHEST)

The American College of Chest Physicians® (CHEST) is the global leader in the prevention, diagnosis, and treatment of chest diseases. Its mission is to champion the prevention, diagnosis, and treatment of chest diseases through education, communication, and research. It serves as an essential connection to clinical knowledge and resources for its 18,000+ members from around the world who provide patient care in pulmonary, critical care, and sleep medicine. For information about CHEST and its family of journals, including the flagship journal ®, visit chestnet.org.

 

About Society of Critical Care Medicine (SCCM)

The Society of Critical Care Medicine (SCCM) is the largest nonprofit medical organization dedicated to promoting excellence and consistency in the practice of critical care. With members in more than 90 countries, SCCM is the only organization that represents all professional components of the critical care team. Visit sccm.org for more information. Follow SCCM on LinkedIn, X, or Facebook.

 

Smart wearables bring personalized health monitoring closer to everyday life

Shanghai Jiao Tong University Journal Center

Facebook

XLinkedInWeChatBlueskyMessageWhatsAppEmail

image: 

• Multimodal sensing strategies—electrical, electrochemical, transistor-based, and optical—integrated with AI-enabled interpretation establish a systems-level framework for continuous, individualized precision diagnosis across metabolites, electrophysiological signals, and molecular biomarkers in diverse biofluids.
• Advanced manufacturing and heterogeneous integration strategies, spanning printing, photolithography, 3D fabrication, and laser writing, enable skin-conformal platforms that unify sensing, signal transduction, power, and wireless communication into compact, long-term wearable diagnostic systems.
• Wearable diagnostic platforms demonstrate broad clinical applicability across neurological, cardiovascular, respiratory, ocular, musculoskeletal, oncological, diabetic, and inflammatory conditions; by enabling longitudinal multi-parameter data fusion and personalized physiological baselines, they provide a clinically actionable pathway for early disease detection, risk stratification, and real-time health management.

view more 

Credit: Xia Gong, Ying Zheng, Xuyin Ding, Jie Gao, Bolang Cheng, Xinyi Shao, Jian Li, Lelun Jiang, Hossam Haick*, Jian Yang*.

As chronic diseases continue to pose major global health challenges, conventional medical diagnosis remains largely dependent on intermittent measurements taken in clinical settings, making it difficult to capture dynamic physiological changes and detect early signs of disease. Researchers from the University of South China, Technion–Israel Institute of Technology, and Sun Yat-Sen University have reviewed recent advances in wearable electronics, highlighting how advanced manufacturing, heterogeneous integration, multimodal sensing, and artificial intelligence (AI) are transforming wearable devices into promising platforms for continuous, personalized precision diagnosis.

Why This Technology Matters

Traditional diagnostic approaches often provide only snapshots of an individual’s health, potentially overlooking subtle changes that develop between clinical visits. Wearable diagnostic systems address this limitation by continuously monitoring physiological signals and molecular biomarkers directly on the body. By combining information from different sensing modalities and tracking changes against personalized physiological baselines, these systems can help identify abnormalities earlier, improve disease risk assessment, and support more individualized healthcare decisions.

Innovative Design and Mechanism

The review examines four major sensing strategies: electrical, electrochemical, transistor-based, and optical sensing. Each offers distinct advantages, from recording electrophysiological activity and monitoring physical changes to detecting metabolites, proteins, hormones, and other molecular biomarkers in sweat, saliva, tears, interstitial fluid, and other biological samples. Advanced manufacturing techniques—including printing, coating, photolithography, three-dimensional fabrication, and laser writing—enable the precise construction of flexible sensing components. Heterogeneous integration then combines sensors, electronic circuits, power supplies, and wireless communication into compact, skin-conformal systems. AI-enabled algorithms process these complex data streams, extract meaningful features, identify abnormal patterns, and support real-time interpretation.

Outstanding Performance

The reviewed wearable platforms demonstrate broad potential across neurological, cardiovascular, respiratory, ocular, musculoskeletal, oncological, diabetic, and inflammatory conditions. By integrating multiple signals rather than relying on a single biomarker, these systems can provide a more comprehensive picture of physiological status. Examples discussed in the review include wearable platforms for cardiac monitoring, continuous biochemical analysis, and disease-associated biomarker detection. Some AI-assisted devices can also distinguish clinically relevant patterns from background physiological activity, illustrating how wearable electronics can progress beyond simple data collection toward intelligent health assessment.

Applications and Future Outlook

Despite these advances, several challenges must be addressed before widespread clinical adoption. Long-term sensor stability, motion artifacts, biofouling, signal drift, interference from complex biological samples, and differences between individual users can compromise measurement reliability. Integrating flexible sensing materials with rigid electronic components while maintaining low power consumption and dependable wireless communication also remains difficult. Future progress will require scalable manufacturing, robust biointerfaces, improved multimodal data fusion, and AI models validated across diverse populations and real-world conditions. Ultimately, the convergence of advanced manufacturing and intelligent data analysis could help establish wearable electronics as accessible tools for early disease detection, longitudinal health monitoring, and personalized medical management.

---

Journal

Nano-Micro Letters

DOI

10.1007/s40820-026-02311-8

Method of Research

News article

Article Title

Wearable Electronics for Precision Diagnosis Through Advanced Manufacturing and Integration

 

Psilocybin providers follow safety guidelines, study finds

First look at how centers screen callers reassuring, but cost may impede greater access

Washington University in St. Louis

Facebook

XLinkedInWeChatBlueskyMessageWhatsAppEmail

In Brief

• Researchers evaluated licensed psilocybin centers in Oregon.
• All centers followed the law for patient screening and treatment provision; none allowed unsupervised use; all asked about psychiatric history, medications and risk factors before recommending services.
• Sessions cost around $1,900, and no center took insurance.

A new study has found that Oregon’s licensed psilocybin centers largely follow recommended safety guidelines, offering reassurance about how the first regulated psilocybin programs in the U.S. function in practice — though the cost of treatment may keep it out of reach for many.

The study, led by researchers at Washington University in St. Louis and the University of Missouri, is the first independent look at how these centers vet their potential clients.

Since 2023, when Oregon became the first state to legalize regulated, supervised psilocybin — the psychoactive ingredient in “magic mushrooms” — it has drawn people with mental health conditions ranging from depression and anxiety to post-traumatic stress disorder and substance use disorders.

Some clinical trials have suggested that psilocybin, combined with psychological support or psychotherapy, can provide relief for some of these conditions, though the drug remains federally illegal. Until now, little has been known about what people actually encounter when they contact a psilocybin center — or what safeguards are in place.

To find out, researchers posing as prospective clients reached 15 of Oregon’s 22 active centers by phone and found that they consistently followed safety practices required by state law: no center allowed unsupervised or take-home use, and staff routinely asked about psychiatric history, medications and contraindications before discussing next steps. These findings complement a recent cohort study in which serious adverse reactions among 346 Oregon clients were uncommon.

The new study is officially published Oct. 12 in Health Affairs Scholar.

“Going in, we wondered whether Oregon’s model might work like a dispensary in practice — that you could call a center, pick up psilocybin and take it home with little screening along the way,” said co-senior author Kevin Xu, MD, an assistant professor of psychiatry at WashU Medicine. “Far from a dispensary model, I was surprised by how service centers asked detailed questions about callers’ mental health and risk factors. Services were also very expensive.”

The research team later interviewed center staff and found that high operating and licensing costs make these services hard to sustain for centers and hard to afford for people who might seek them. Indeed, four of the 26 centers Oregon had licensed closed before the study’s calls began, and another seven could not be reached by phone.

What do prospective clients experience when seeking psilocybin services?

The research team, which included experts in medicine, social work and public health, designed the study to follow the path a real client would take: reviewing the center’s website, then calling to ask questions.

“This latest innovative study highlights our unique interdisciplinary approach to psychedelic science.” 

— Ginger Nicol, MD, WashU Medicine

Three co-authors made scripted calls in which they posed as potential clients with depression and a history of alcohol use disorder or a family history of psychosis. They documented whether staff asked about medications, suicidal ideation and psychiatric history, as required by Oregon’s legal framework for psilocybin services. They also documented the price quoted for services.

“These interactions were not just, ‘Hey, give me mushrooms,’ and that’s it,” said co-senior author Leopoldo J. Cabassa, PhD, a professor at WashU’s Brown School who also co-directs WashU’s Center for Mental Health Services Research. “These were thoughtful conversations where the facilitator engaged people, asked good questions, and tried to get as much information as possible. Facilitators pushed past a simple checklist to actually understand a caller’s history before making a recommendation for services.”

Nearly every center flagged psychoactive medications that should not be combined with psilocybin, and no center advised callers taking antidepressants to stop them without first consulting with their primary care physician.

The median cost of a session was $1,900, and no center accepted insurance, which would limit accessibility if the treatment is legalized in other U.S. states.

“Oregon has dosed more people with psilocybin in a regulated model than anywhere else in the world,” said first author Dani Adams, PhD, an assistant professor of social work in the College of Health Sciences at the University of Missouri and Director of the Collaborative for Missouri Psychedelic-Assisted Therapy, who began this line of research while a postdoctoral fellow at WashU working with Cabassa. “But it is still out of reach for most Americans who are struggling with their mental health.”

Bipartisan efforts to expand psilocybin access

Veterans in particular face high rates of PTSD, depression and suicide, which has fueled bipartisan interest in getting psychedelic-assisted therapy approved by the U.S. Food and Drug Administration (FDA). Cabassa noted that in the meantime, some veterans unable to access or afford regulated care in the U.S. have instead sought unregulated treatments abroad, highlighting the importance of getting domestic access right.

WashU has been a leader in psychedelic research on multiple fronts. Co-author Ginger Nicol, MD, a professor of psychiatry at WashU Medicine, co-directs the Center for Holistic Interdisciplinary Research in Psychedelics alongside Cabassa, and was the principal investigator on a landmark 2024 study in Nature that showed how psilocybin temporarily desynchronizes activity across the human brain, among the first studies to map the drug’s effects at that scale.

“This latest innovative study highlights our unique interdisciplinary approach to psychedelic science,” Nicol said. “The team recognized an opportunity and acted swiftly to collect data that fills a major gap in scientific knowledge.”

With the new study, Cabassa and Nicol noted, WashU’s footprint in the field has grown from understanding brain circuitry to real-world implementation.

“We have to figure out what safeguards are needed and make sure those are maintained as the system solves for accessibility,” said Cabassa.

Adams DR, Hartman M, Gong L, Shankar S, Ling R, Steinle JT, Siegel JS, Nicol GE, Grucza RA, Borodovsky JT, Cabassa LJ, Xu KY. “Am I a candidate for magic mushrooms?” A mystery shopper study of Oregon psilocybin service centers. Health Affairs Scholar. October 12, 2026. DOI: 10.1093/haschl/qxag231

The work was supported by NIH grants K08DA061258 and R25 MH112473. Additional funding came from the WashU PJ000033039 Funding Award, the WashU Center for Dissemination and Implementation, and WashU’s Here & Next Transcend Initiative. The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication. The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH.

J.S. Siegel reported former employment by Sumitomo Pharma America; receiving consulting fees from Longitude Capital and Otsuka Pharmaceutical and being coinventor on a provisional patent (No. 020949/US 15060-1787) related to the use of precision functional mapping to assess target engagement of experimental therapeutics. G.E. Nicol reported receiving grants from COMPASS Pathways and Usona; receiving drug supply from Usona; and being a paid consultant for Carelon, Alkermes, Sunovion, and Novartis outside the submitted work as well as serving as codirector of the Washington University Center for Holistic Interdisciplinary Research in Psychedelics (CHIRP), a Transcend Initiative funded by the Washington University Here and Next strategic plan.

About WashU Medicine

WashU Medicine is a global leader in academic medicine, including biomedical research, patient care and educational programs with 3,100 faculty. Its National Institutes of Health (NIH) research funding portfolio is the second largest among U.S. medical schools and has grown 78% since 2016. Together with institutional investment, WashU Medicine commits over $1.6 billion annually to basic and clinical research innovation and training. Its faculty practice is consistently among the top five in the country, with more than 2,550 faculty physicians practicing at 200 locations. WashU Medicine physicians exclusively staff Barnes-Jewish and St. Louis Children’s hospitals — the academic hospitals of BJC HealthCare — and Siteman Cancer Center, a partnership between BJC HealthCare and WashU Medicine and the only National Cancer Institute-designated comprehensive cancer center in Missouri and southern Illinois. WashU Medicine physicians also treat patients at BJC’s community hospitals in our region. With a storied history in MD/PhD training, WashU Medicine recently dedicated $100 million to scholarships and curriculum renewal for its medical students, and is home to top-notch training programs in every medical subspecialty as well as physical therapy, occupational therapy, and audiology and communications sciences.

---

Journal

Health Affairs Scholar

Article Title

“Am I a candidate for magic mushrooms?” A mystery shopper study of Oregon psilocybin service centers

Article Publication Date

12-Oct-2026

COI Statement

The work was supported by NIH grants K08DA061258 and R25 MH112473. Additional funding came from the WashU PJ000033039 Funding Award, the WashU Center for Dissemination and Implementation, and WashU’s Here & Next Transcend Initiative. The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication. The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH. J.S. Siegel reported former employment by Sumitomo Pharma America; receiving consulting fees from Longitude Capital and Otsuka Pharmaceutical and being coinventor on a provisional patent (No. 020949/US 15060-1787) related to the use of precision functional mapping to assess target engagement of experimental therapeutics. G.E. Nicol reported receiving grants from COMPASS Pathways and Usona; receiving drug supply from Usona; and being a paid consultant for Carelon, Alkermes, Sunovion, and Novartis outside the submitted work as well as serving as codirector of the Washington University Center for Holistic Interdisciplinary Research in Psychedelics (CHIRP), a Transcend Initiative funded by the Washington University Here and Next strategic plan.