Saturday, April 11, 2026

ARACHNOLOGY

Tarantulas may use learning and memory to search for food and locate their retreats



University of Turku
Cave-dwelling tarantula 

image: 

A blind cave-dwelling tarantula observed in Mexico.

view more 

Credit: Rick C. West






Researchers have documented several cases of spatial orientation in tarantulas living both in trees and in underground burrows. Spatial orientation refers to the ability of an animal to understand where it is in three-dimensional space and how to navigate purposefully within its environment. The observations suggest that tarantulas may remember and reuse information to improve their chances of catching prey or to locate their retreats, for example.

A new study by researcher Alireza Zamani from the University of Turku, Finland, and independent researcher Rick C. West reports on spatial orientation in tarantulas in their natural habitats across North and South America. The observations pertain to both tree-dwelling species as well as burrowing species. All the observed tarantula species showed behaviour that may indicate learning.

Some tree-dwelling individuals were observed leaving their retreats every night and travelling to prey-rich locations about one to two meters away, such as areas near artificial lights that attracted flying insects. After hunting, the spiders returned to the same retreats.

The researchers also reported unusual climbing behaviour in species that normally live in burrows. These tarantulas appeared to adapt to foraging in tree canopies rather than on the ground during the dry season.

Together, these observations suggest that tarantulas may remember and reuse information from previous experiences to improve their chances of catching prey. In addition, in lowland floodplain areas, ground-dwelling tarantulas were seen temporarily moving into shrubs or trees during the rainy season, likely to avoid flooding.

The observed behaviours differ from ontogenetic shifts, which are changes in an animal’s habitat, diet, or behaviour that occur at specific stages of development, typically when it grows from a juvenile to an adult. According to this research, a blind cave-dwelling tarantula from Mexico possibly shows such a shift in foraging behaviour: while juveniles appear to stay closer to fixed retreats, adults move more irregularly and seem less dependent on a permanent retreat. This might be because, as they grow, their energetic demands increase and they begin to hunt larger prey.

Behaviour combines internal signals with external cues

Tarantulas were also observed responding to disturbance by returning quickly and directly to their burrows without hesitation or signs of disorientation. What makes this one of the most important observations of the study is the fact that the blind cave-dwelling tarantula behaved similarly to the tarantulas with vision.

The researchers suggest that this behaviour is likely supported by the spiders’ ability to combine internal body signals related to movement, body position, and direction with environmental information, such as light, vibrations, and chemical cues.

“Previous studies have shown that tarantulas can learn to avoid unpleasant stimuli, navigate complex mazes, and remember spatial locations over time. These abilities suggest that their nervous systems support more flexible behaviour than traditionally assumed,” says the lead author of the study, researcher Alireza Zamani from the University of Turku.

However, the researchers note that the cognitive interpretation remains preliminary. Tarantulas are known to rely heavily on sensory information, particularly chemical and silk-based cues, which may also explain how they recognise retreats and choose foraging sites. Further experimental research will be needed to evaluate the relative roles of learning and sensory mechanisms.

“Overall, studies on spider learning are still relatively recent, especially for tarantulas. Observations from the field, combined with controlled experiments, will be important for understanding how sensory cues, memory, and experience interact to help these spiders navigate and search for prey,” notes Zamani.

The research article was published in the journal Ecology and Evolution.


Goliath birdeater 

A female Goliath birdeater, a species that normally lives in burrows, foraging in tree canopies.

Credit

Rick C. West

 

Study finds no increased risk of respiratory cancers from asbestos-free talc exposure




International Association for the Study of Lung Cancer

Paolo Boffetta, Stony Brook Cancer Center and Department of Family, Population and Preventive Medicine, Renaissance School of Medicine at Stony Brook University 

image: 

Paolo Boffetta, Stony Brook Cancer Center and Department of Family, Population and Preventive Medicine, Renaissance School of Medicine at Stony Brook University

view more 

Credit: Stony Brook Cancer Center and Department of Family, Population and Preventive Medicine, Renaissance School of Medicine at Stony Brook University





(Denver, Colo. April 10, 2026) -- In a systematic review and meta-analysis, researchers found that occupational exposure to talc that is not contaminated with asbestos is not associated with an increase in the risk of lung cancer, mesothelioma, or laryngeal cancer.  The findings are published in the Journal of Thoracic Oncology, the official journal of the International Association for the Study of Lung Cancer.  Access the complete study here: https://www.jto.org/article/S1556-0864(26)00163-2/.

Evidence suggests a potential link between occupational talc exposure and increased risk of lung cancer and mesothelioma, when talc is contaminated with asbestos, a known carcinogen. However, the findings for non-contaminated talc remained inconclusive.

To resolve this issue, researchers led by Paolo Boffetta, Stony Brook Cancer Center and Department of Family, Population and Preventive Medicine, Renaissance School of Medicine at Stony Brook University, NY, identified 13, 8, and 7 publications reporting on lung cancer, mesothelioma, and laryngeal cancer, respectively. Five studies on lung cancer in talc miners and millers and three studies in other industries were included in the meta-analysis.

The meta-analysis showed:

Lung cancer

Relative risk (RR) of 1.13 (95% CI: 0.97–1.33) among miners and millers

RR of 1.12 (95% CI: 0.79–1.57) among workers in other industries

Mesothelioma

No cases reported among talc miners and millers in the primary analyses

Laryngeal cancer

No association (RR = 0.98; 95% CI: 0.58–1.57)

Talc is a naturally occurring mineral that is mined from the earth and then processed into the soft, powdery substance used in products like cosmetics, ceramics, paper, and plastics.  Major talc-producing regions include China, India, Brazil, the United States, France, and Italy.

Lung cancer is the second most common cancer in both men (after prostate cancer) and women (after breast cancer). It accounted in 2022 for an estimated 1,572,000 new cases and 1,233,000 deaths each year among men and 908,000 cases and 587,000 deaths among women.

Mesothelioma is a cancer of the mesothelium which is most frequently diagnosed in the pleura (known as pleural mesothelioma) but also can occur in the abdominopelvic cavity (peritoneal mesothelioma), the heart (pericardial mesothelioma), or the testes (testicular mesothelioma) (1). Mesothelioma was considered a very rare tumor until a large series of cases were reported in the 1960s among workers employed in asbestos mining and manufacturing (2,3).

Laryngeal cancer is one of the most prevalent types of head and neck cancer. According to GLOBOCAN 2022 (4), its Age Standardized Rate (ASR) is only 1.9 per 100,000 and Age-Standardized Mortality Rate (ASMR) is 1 per 100,000 globally.

“In conclusion, current epidemiological evidence does not provide support for an increased risk of lung cancer, mesothelioma, or laryngeal cancer among workers who are primarily exposed to talc that is free from asbestos contamination,” Boffetta and co-authors wrote.

However, according to the study, it is important to continue monitoring occupational groups, enhance the mineralogical characterization of talc deposits, and conduct future studies that include detailed exposure assessments and control for key confounding factors. These steps are essential to better understand any potential low-level risks and to inform strategies for occupational health prevention.

About the IASLC

The International Association for the Study of Lung Cancer (IASLC) is the only global organization dedicated solely to the study of lung cancer and other thoracic malignancies. Founded in 1974, the association's membership includes more than 11,000 lung cancer specialists across all disciplines in over 100 countries, forming a global network working together to conquer lung and thoracic cancers worldwide. The association publishes the Journal of Thoracic Oncology, the primary educational and informational publication for topics relevant to the prevention, detection, diagnosis and treatment of all thoracic malignancies. Visit www.iaslc.org for more information.

About the JTO

Journal of Thoracic Oncology (JTO), the official journal of the International Association for the Study of Lung Cancer, is the primary educational and informational publication for topics relevant to the prevention, detection, diagnosis, and treatment of all thoracic malignancies. JTO emphasizes a multidisciplinary approach and includes original research reviews and opinion pieces. The audience includes epidemiologists, medical oncologists, radiation oncologists, thoracic surgeons, pulmonologists, radiologists, pathologists, nuclear medicine physicians, and research scientists with a special interest in thoracic oncology.

 

AI outperforms doctors at summarizing complex cancer pathology reports


Many open-source AI models generated more complete summaries, especially for molecular findings


Northwestern University

Study senior author Dr. Mohamed Abazeed demonstrates a prototype AI tool that summarizes cancer pathology reports, shown here in a radiation oncology setting. 

image: 

Study senior author Dr. Mohamed Abazeed demonstrates a prototype AI tool that summarizes cancer pathology reports, shown here in a radiation oncology setting. The tool, developed at Northwestern Medicine, is not yet in clinical use and is undergoing further testing.

view more 

Credit: Northwestern University






  •      Study analyzed real-world lung cancer pathology reports from de-identified patients
  •        Six open-source AI models were tested, including systems developed by Meta, Google
  •        Author: ‘This could help physicians focus more on patient care’

CHICAGO --- AI models can generate more complete summaries of complex cancer pathology reports than physicians, according to a new Northwestern Medicine study that tested six models developed by Meta, Google, DeepSeek and Mistral AI.

The study was published on April 8 in JCO Clinical Cancer Informatics, a journal from the American Society of Clinical Oncology.

The findings offer a potential fix to a growing challenge in oncology. As biomarker testing expands, and patients live longer, pathology reports have become increasingly detailed and longitudinal, often spanning multiple institutions and requiring clinicians to synthesize large volumes of information under significant time pressure.

In this study, several open-source AI models consistently produced summaries that were more comprehensive than physician-written versions, particularly in capturing molecular and genetic findings that are critical for treatment decisions.

“As cancer care becomes increasingly complex, the burden of synthesizing complex reports is growing rapidly,” said senior study author Dr. Mohamed Abazeed, chair and professor of radiation oncology at Northwestern University Feinberg School of Medicine. “What we’re seeing is that AI can help ensure critical pathological and genomic details are consistently captured — not as a replacement for physicians, but as a tool to augment clinical decision-making.”

How the study was conducted 

The Northwestern investigators analyzed 94 de-identified pathology reports from lung cancer patients. These reports included detailed text describing:

  • Histopathological findings (microscopic tumor characteristics)
  • Immunohistochemical results (protein expression testing)
  • Molecular and genetic data relevant to treatment

The AI models analyzed the text content of these reports and generated structured summaries.

The AI-generated summaries were compared to clinical summaries previously written by physicians. A panel of oncologists assessed each summary for accuracy, completeness, conciseness and potential clinical risk. Across models, AI-generated summaries were consistently rated as more complete, with the largest differences observed in the inclusion of molecular and genomic findings.

“If AI can reliably synthesize these reports, clinicians can review key findings more efficiently, important genetic details are less likely to be overlooked and documentation becomes more standardized,” said study co-author Troy Teo, instructor of radiation oncology at Feinberg. “This could help physicians focus more on patient care.”

Llama 3.1 and DeepSeek performed best

The scientists evaluated six open-source language models: Meta’s Llama 3.0, 3.1 and 3.2 models, Google’s Gemma 9B, Mistral 7.2B and DeepSeek-R1. These are not chatbots like ChatGPT, but systems that researchers can download and run locally. According to the study, the strongest performers were DeepSeek and Llama 3.1.

The Northwestern team is now developing an app using Llama 3.1 to eventually allow physicians to upload pathology reports and receive AI-generated summaries for their review. But the study authors emphasize that before deploying the app, they need more testing and validation studies.

AI as a second-opinion tool

The authors said they envision AI as a support layer that enhances, rather than replaces, clinical expertise. It could help highlight key findings, identify missing information and improve consistency in documentation.

“Patients with complex cancers might benefit the most,” said study first author Dr. Yirong Liu, a fifth-year resident in radiation oncology at McGaw Medical Center of Northwestern. “In cases where missing a key pathological finding or an actionable genetic marker could change treatment decisions, ensuring that information is consistently captured is critical.”

“Patients are living longer and undergoing repeated biopsies and genetic sequencing,” Liu added. “Their reports can span dozens of pages. Even a single missed detail can impact care, and this is where AI may provide meaningful support.”

The study is titled “Toward Automating the Summarization of Cancer Pathology Reports Using Large Language Models to Improve Clinical Usability.” Troy Teo received funding from the Canadian Institute of Health Research (grant CIHR-472392) and from Amazon Web Services’ Social Impact funding.


A prototype AI tool that summarizes cancer pathology reports. The tool, developed at Northwestern Medicine, is not yet in clinical use and is undergoing further testing.

Study senior author Dr. Mohamed Abazeed demonstrates a prototype AI tool that summarizes cancer pathology reports, shown here in a radiation oncology setting. [VIDEO] 


Study authors Drs. Mohamed Abazeed (right), Yirong Liu and Troy Teo (left) demonstrates a prototype AI tool that summarizes cancer pathology reports, shown here in a radiation oncology setting.

Credit

Northwestern University

Usage Restrictions

 

Scientists develop spatiotemporal correlation-based deep learning framework for bias correction of atmospheric and oceanic variables




Institute of Atmospheric Physics, Chinese Academy of Sciences
Overview of the proposed bias correction architecture 

image: 

Overview of the proposed bias correction architecture

view more 

Credit: Yuze Sun





Daily travel plans and early warnings for extreme weather all rely on traditional numerical weather prediction. However, both traditional numerical weather prediction and AI forecasting large models have long suffered from systematic biases, which compromise forecast accuracy.

To address this challenge, the research group led by Prof. Xiaomeng Huang from Tsinghua University, China, in collaboration with the National Climate Centre, China, has developed an AI bias correction framework based on spatiotemporal correlation deep learning. This framework accurately corrects forecast biases, achieving a maximum 20% reduction in the root-mean-square error of 7-day 2-meter air temperature forecasts. In addition, it can support the bias correction of oceanic variables, making forecasts in meteorological and oceanic scenarios more accurate. The findings were recently published in Atmospheric and Oceanic Science Letters.

The research team systematically integrated three key innovations into their model design: dynamic climatological normalization, ConvLSTM with temporal causality constraints, and residual self-attention mechanisms, enabling systematic bias correction of European Centre for Medium-Range Weather Forecasts (ECMWF) numerical forecasts.

The model was trained and validated using 41 years (1981–2021) of global atmospheric data, with ERA5 (fifth generation ECMWF atmospheric reanalysis) data serving as the ground truth. A decadal stratified sampling strategy, i.e., five non-consecutive years (1981, 1991, 2001, 2011, 2021) selected at 10-year intervals as a testing set, was employed to ensure the model’s generalization capability across distinct climate phases.

Results show that the model boasts outstanding generalization capability. After being trained on the air temperature variable, it only takes 20 minutes to perform cross-variable correction for wind fields and air pressure, cutting the retraining time by 85%. Integrated as a plug-in into existing AI forecasting models, it further improves the forecast skill by 10%. Moreover, the corrected atmospheric data can significantly enhance the prediction performance of ocean models, enabling cross-domain empowerment from meteorology to oceanography.

This research was supported by the National Natural Science Foundation of China. The model code has been made publicly available, providing a reproducible technical solution for meteorological AI research.

 

New AI technology to speed drug development



University of Virginia Health System





University of Virginia School of Medicine scientists have developed a bold new approach to drug development and discovery that could dramatically accelerate the creation of new medicines.

UVA’s Nikolay V. Dokholyan, PhD, and colleagues have developed a suite of artificial intelligence-powered tools, called YuelDesign, YuelPocket and YuelBond, that work together to transform how new drugs are created. The centerpiece, YuelDesign, uses a cutting-edge form of AI called diffusion models to design new drug molecules tailored to fit their protein targets exactly, even accounting for the way proteins flex and shift shape during binding.

A companion tool, YuelPocket, identifies exactly where on a protein a drug can attach, while YuelBond ensures the chemical bonds in designed molecules are accurate. Together, the approach is poised to improve both how new drugs are designed and how quickly and efficiently existing drugs can be evaluated for new purposes.

“Think of it this way: Other methods try to design a key for a lock that's sitting perfectly still, but in your body, that lock is constantly jiggling and changing shape. Our AI designs the key while the lock is moving, so the fit is much more realistic,” said Dokholyan, of UVA’s Department of Neurology. “This could make a real difference for patients with cancer, neurological disorders and many other conditions where we desperately need better drugs targeting these wiggly proteins but keep hitting dead ends.”

The Pitfalls of Drug Development

The average cost of developing a new drug has been estimated to reach or exceed $2.6 billion, and almost 90% of new drugs fail when they reach human testing. That is due, in no small part, to the difficulty of predicting how molecules in a drug will interact, or bind, with their targets in the body. If a molecule doesn’t bind exactly as intended, at exactly the right spot, the drug won’t work, or could have unwanted, harmful side effects.

Artificial intelligence has helped address this problem, greatly accelerating drug design, but Dokholyan’s work takes it to the next level. His YuelDesign overcomes limitations of the existing options by designing drug molecules while treating proteins as flexible, dynamic structures, not the rigid and frozen snapshots used by other methods. This is critical because proteins often change shape when a drug binds to them, a phenomenon known as "induced fit." Ignoring this flexibility can lead to drugs that look promising on a computer screen but fail in reality.

Dokholyan and his team designed YuelDesign specifically to overcome this problem. Using advanced AI “diffusion models,” the technology simultaneously generates both the protein pocket structure and the small molecule that can slot into it – the key that will turn the lock, allowing both to adapt to each other during the design process.

A companion tool, YuelPocket, uses graph neural networks to identify precisely where on a protein a drug should bind, even on predicted protein structures from existing tools such as AlphaFold. “Most existing AI tools treat the protein as a frozen statue, but that's not how biology works. Our approach lets the protein and the drug candidate evolve together during the design process, just as they would in the body,” said researcher Dr. Jian Wang. “We showed, for example, that when designing molecules for a well-known cancer-related protein called CDK2, only YuelDesign could capture the critical structural changes that happen when a drug binds.”

Mapping out protein pockets is critical to “virtually every aspect of modern development,” the researchers note in a new scientific paper outlining their YuelPocket testing. The promising results have Dokholyan hopeful that the technology can reduce drug development costs, improve the success rate of new drug candidates and accelerate how quickly new treatments and cures can reach patients. (Accelerating how quickly lab discoveries can be turned into medicines to benefit patients is the primary mission of UVA’s new Paul and Diane Manning Institute of Biotechnology.)

“Our ultimate goal is to make drug discovery faster, cheaper and more likely to succeed, so that promising treatments can reach patients sooner,” Dokholyan said, adding that he wants to “democratize” drug discovery by putting new tools at scientists’ fingertips. “We've made all of our tools freely available to the scientific community. We want researchers anywhere in the world to be able to use them to tackle the diseases that matter most to their patients.”

Findings Published

Dokholyan and his team have described the development and results of these tools in papers in the scientific journals PNASJCIM and Science Advances. The research team includes Wang, Dong Yan Zhang, Shreshty Budakoti and Dokholyan. The scientists have no financial interest in the work.

The research has been supported by the National Institutes of Health, grant 1R35 GM134864; the National Science Foundation, grant 2210963; the Huck Institutes of the Life Sciences; and the Passan Foundation.

To keep up with the latest medical discoveries from the UVA School of Medicine and the Manning Institute, bookmark the Making of Medicine blog at https://makingofmedicine.virginia.edu.