Category: 8. Health

Skip to main content

Discover more from Physics World

Copyright © 2025 by IOP Publishing Ltd and individual contributors

Secure your spot now for exclusive AM insights at our Additive Manufacturing Advantage online conference covering aerospace, space, and defense!

McGill University’s spinout TissueTinker is exploring a new bioprinting approach that could improve the way cancer drugs are tested in preclinical settings.

Co-founded by Benjamin Ringler, Madison Santos, and Isabelle Dummer, the startup recently received a Develop award from the McGill Innovation Fund (MIF) to advance its miniature tumor model platform. 

Designed as a human-relevant alternative to 2D cultures and animal testing, the miniature models aim to reduce the 90% failure rate of cancer drugs after preclinical testing by better capturing tumor complexity and improving predictability early on.

“Because the testing environment more readily simulates the human body, researchers can better assess and understand whether or not their drug works before reaching clinical trial stages,” Ringler detailed. “This is key for drug progression and curbing financial waste in the industry.”

Miniature models offer customization edge

TissueTinker’s platform centers on bioprinting tumor models at a scale of around 300 µm, a size the team considers optimal for balancing biological relevance with resource efficiency. 

Using bioink made from living cells, the models are constructed to include both healthy and cancerous tissue types, positioned with spatial precision. This structure enables the replication of key physiological features, such as hypoxic cores, that influence how tumors grow and respond to treatment.

The platform’s design allows researchers to adjust both the structure and cell composition of each tumor model, depending on the specific biological question being studied. This adaptability makes it possible to replicate a wide range of tumor conditions, offering more targeted insights into how treatments behave under different physiological scenarios.

This approach gains added relevance under updated  US Food and Drug Administration (FDA) guidelines, which now allow drug developers to use human-based models in place of animal testing during preclinical research. By offering a method that reflects the complexity of human tumours more accurately, TissueTinker provides a practical option within this shifting regulatory landscape.

Backed by support from the MIF, the team has refined both the technical and strategic dimensions of the platform. In addition to funding, the program provided mentorship that helped the founders focus on long-term development. They are now working to expand their tumor model library and plan to license the platform to pharmaceutical companies and research institutions.

Rethinking drug testing with bioprinted tumors

With cancer responsible for 10 million deaths in 2020 and cases expected to surpass 28 million by 2040 as referenced by McGill, many are seeking more efficient approaches to drug development.

Previously, Edinburgh-based tumor 3D printing specialist Carcinotech and bioprinting firm CELLINK partnered to advance cancer drug development by creating standardized protocols for bioprinted tumor models built from cancer cell lines. These models were designed to replicate the physiological makeup of specific cancer types, incorporating five key cell types in accurate ratios to improve testing relevance. 

Developed for use with CELLINK’s BIO CELLX system, the protocols were expected to enable automated and reproducible 3D cell culture workflows, streamlining drug screening processes. The partnership built on earlier work combining Carcinotech’s expertise in tumor modeling with CELLINK’s bioinks and bioprinting technology to enhance precision in preclinical research.

In 2021, researchers at the University of Stuttgart and Robert Bosch Hospital developed a 3D printed tissue platform designed to improve cancer drug testing while reducing the need for animal experiments. 

As part of a €3.8 million initiative funded by the state of Baden-Württemberg, the team used bioprinting and simulation data to create skin-like microfluidic structures that more closely mimic tumor behavior in the human body. Their approach combined ex-vivo, de-novo, and in-silico strategies, producing modular, nutrient-loaded cell structures that can be assembled like “lego bricks” to simulate realistic tumors and better predict drug distribution outcomes.

What 3D printing trends should you watch out for in 2025?

How is the future of 3D printing shaping up?

To stay up to date with the latest 3D printing news, don’t forget to subscribe to the 3D Printing Industry newsletter or follow us on Twitter, or like our page on Facebook.

While you’re here, why not subscribe to our Youtube channel? Featuring discussion, debriefs, video shorts, and webinar replays.

Featured image shows McGill Innovation Fund team TissueTinker is reimagining how we test cancer therapies with customizable, human-relevant bioprinted tumor models that replicate human tissue. Photo via McGill University.

Drug discovery efforts based on DNA-encoded chemical libraries are inadvertently overlooking numerous potential drug candidates, new research shows. 

Each molecule in a DNA-encoded chemical library is tagged with a unique DNA sequence that acts like a barcode. Such libraries have revolutionised early drug discovery by allowing researchers to screen millions, if not billions, of compounds simultaneously. And the resulting datasets are often used to train machine learning models that seek out promising drug candidates.

 Keen to understand how reliable data linked to DNA-encoded chemical libraries actually is, Raphael Franzini, from the University of Utah in the US, and colleagues investigated a library with over 58,000 compounds designed to target enzymes involved in DNA repair and cancer. When they synthesised and tested 33 molecules that screens had dismissed, they discovered that these compounds were often just as effective as those flagged as promising. In particular, various screens nearly missed compounds that were structurally similar to olaparib, an approved cancer drug. 

‘We found that DNA-encoded library data often labels good molecules as bad molecules,’ explains Franzini. 

The problem appears to lie with the DNA barcodes themselves. When the team compared molecules with and without these tags, they found that the DNA reduced molecules’ activity. The effect was even more pronounced when molecules were tested against targets they were not originally designed for. 

Laura Guasch, a computational chemist at pharmaceutical company Roche, Switzerland, describes the findings as ‘a highly relevant contribution’. She says the study ‘raises crucial awareness regarding how these numerous false negatives can impair the increasingly popular machine learning algorithms used in this domain.’

 ‘False negatives introduce substantial noise and bias into training datasets, causing machine learning models to learn misleading patterns or ignore valid chemotypes,’ comments Srinivas Chamakuri, an assistant professor at Baylor College of Medicine’s Center for Drug Discovery in the US. 

Franzini and colleagues demonstrated that even when machine learning models appeared to perform well, they were actually just recognising recurring structural fragments rather than developing genuine predictive capabilities. 

‘A primary implication of this study is the significant risk that current drug discovery programs might be overlooking potential drug candidates due to high rates of false negatives,’ notes Guasch. 

The researchers found that removing unreliable data from the training sets and focusing only on confirmed active compounds dramatically improved models’ ability to identify promising drugs. This suggests that current machine learning approaches in drug discovery may need fundamental changes to account for the inherent biases in screening data.

While the use of radiation bridging therapy (BT) in chimeric antigen receptor (CAR) T-cell therapy for blood cancer is expanding, plenty of unanswered questions remain on topics such as ideal timing and doses, a radiologist cautioned hematologist colleagues.

The lack of guidelines has immediate clinical implications, said John P. Plastaras, MD, PhD, professor of radiation oncology at the Hospital of the University of Pennsylvania, Philadelphia, in a presentation at 18th International Conference on Malignant Lymphoma (ICML) 2025 in Lugano, Switzerland.

“This actually just came up the other day when one of our medical colleagues said, ‘I’m really worried about this patient. They’re ready for CAR T cell, but I think you need to radiate this area. Can you do it a week after [therapy]?’ The answer is, ‘We don’t know.’”

On the other hand, clinicians now have clarity about safety and interaction with CAR T-cell therapy, he noted, and data is coming in rapidly.

Here are some questions and answers about radiation BT:

What is BT in CAR T-cell therapy?

BT refers to treatment that provides a “bridge” for patients between the components of CAR T-cell therapy.

As a 2024 report about BT in hematologic cancer explained, the treatment “is delivered after leukapheresis for CAR T-cells” — the process in which white cells are removed from a patient’s blood, which is then returned to the body — “has been completed and before lymphodepleting chemotherapy and CAR T-cell infusion.”

The report said “patients who receive BT are predominantly those with a higher disease burden and rapidly progressive disease. These patients tend to have worse overall outcomes, likely related to their aggressive underlying disease.”

Where does radiation fit into BT?

According to the 2024 report, “combination chemoimmunotherapy has typically been the form of BT that is used most often.” Targeted therapy is another option, the report said, although data is from “very small sample sizes.”

And then there’s radiation, which the report said is useful “particularly in patients with limited sites of disease or patients who are at risk for structural complications such as airway compromise or renal dysfunction.”

What do we know about radiation’s efficacy?

The first oral report on bridging radiation in CAR T-cell therapy only appeared in 2018, Plastaras said, followed by the first published report in 2019. Despite this fairly short time period, “we are certainly seeing a lot of new data,” Plastaras said.

He highlighted the newly released International Lymphoma Radiation Oncology Group (ILROG) study of radiation BT in conjunction with CAR T-cell therapy for relapsed/refractory B-cell lymphomas. The retrospective study of 172 patients at 10 institutions treated from 2018 to 2020 showed that 1- and 2-year progression-free survival (PFS) rate was 43% (95% CI, 36-51) and overall survival rate was 38% (95% CI, 30-45).

In a multivariable model, comprehensive radiation BT was linked to superior PFS than focal therapy (hazard ratio, 0.38; 95% CI, 0.22-0.63; P < .001).

“Comprehensive radiation was a very strong predictor for improved PFS, but we did not see was a huge dose effect,” said Plastaras, who coauthored the study.

What about toxicity?

Questions about other clinical matters were resolved prior to 2022, he said, when CAR T-cell therapy was used primarily in third line and later settings.

“Does radiation cause excess toxicities?” he asked. “A lot of the single-institution studies answered that, and I think most medical oncologists and hematologists are okay with this idea that radiation isn’t causing a lot of excess toxicities.”

As for whether radiation interferes with the effectiveness of CAR T-cell therapy, “the data to this point have demonstrated that probably not,” he said. “We’ve probably put that one to bed.”

What do we know about treatment timing?

“The timing question is still quite open,” Plastaras said. “How much time should there be between radiation and lympho-depleting chemotherapy? Is it better to put the radiation very close to the CAR T-cell [therapy] so this priming effect might happen, or can that happen weeks in advance? We don’t know the answers to those.”

According to Plastaras, researchers are still trying to understand the role radiation the consolidation period after CAR T-cell therapy. “If we wait for day-30 PET [scan], is that OK? Do we need to wait longer? Are we going to mess up the lymph nodes that have CAR T-cells floating around in them?”

What about doses and imaging?

There’s also a lack of insight into technical questions about radiation dose and fractionation. “The [radiation] volume question is one of key importance. Do we just do gross disease? Do we treat all the other small spots out there, and importantly, do we treat regional nodes or not? We get these questions all the time.”

The role of imaging is also unclear, he said, in terms of timing during and after bridging radiation therapy and after CAR T-cell therapy.

What do we need to learn about now?

Looking forward, Plastaras outlined what he called “version 2.0” questions for the evolving field: Can radiation rebulking decrease CAR-T cell toxicities? Will very low dose “priming” radiation affect outcomes? 

He highlighted other questions: Can radiation be part of a combined modality approach in limited stage relapsed/refractory disease? Should central nervous system lymphoma be treated differently? 

When will we get new guidelines?

According to Plastaras, Memorial Sloan Kettering Cancer Center Radiology Oncologist Brandon Imber,MD, MA, in New York City, is leading a new ILROG guideline project with the intention of publishing details in the journal Blood. “This is a work in progress,” Plastaras said. “Our target is 2025 to at least get something submitted.”

Plastaras had no disclosures.

Specialty

Please chooseI’m not a medical professional.Allergy and ImmunologyAnatomyAnesthesiologyBiostatisticsCardiac/Thoracic/Vascular SurgeryCardiologyCritical CareDentistryDermatologyDiabetes and EndocrinologyEmergency MedicineEpidemiology and Public HealthFamily MedicineForensic MedicineGastroenterologyGeneral PracticeGeneticsGeriatricsHealth PolicyHematologyHIV/AIDSHospital-based MedicineI’m not a medical professional.Infectious DiseaseIntegrative/Complementary MedicineInternal MedicineInternal Medicine-PediatricsMedical Education and SimulationMedical PhysicsMedical StudentNephrologyNeurological SurgeryNeurologyNuclear MedicineNutritionObstetrics and GynecologyOccupational HealthOncologyOphthalmologyOptometryOral MedicineOrthopaedicsOsteopathic MedicineOtolaryngologyPain ManagementPalliative CarePathologyPediatricsPediatric SurgeryPharmacologyPhysical Medicine and RehabilitationPlastic SurgeryPodiatryPreventive MedicinePsychiatryPsychologyPulmonologyRadiation OncologyRadiologyRheumatologySubstance Use and AddictionSurgeryTherapeuticsTraumaUrologyMiscellaneous

A rare cell in the lining of lungs is fundamental to the organwide response necessary to repair damage from toxins like those in wildfire smoke or respiratory viruses, Stanford Medicine researchers and their colleagues have found. A similar process occurs in the pancreas, where the cells, called neuroendocrine cells, initiate a biological cascade that protects insulin-producing pancreatic islet cells from damage.

Treating the airways of mice with an experimental drug that activates the repair pathway protected their airways from damage after infection with influenza or the virus that causes COVID-19. Conversely, animals in which the pathway was blocked experienced much more severe damage to their airways.

Activating the signaling pathway initiated by airway or pancreatic neuroendocrine cells in a similar way in humans might enhance the ability of firefighters and those with respiratory illnesses to avoid permanent lung damage, the researchers believe. They also suspect it could help prevent people with metabolic syndrome from progressing to diabetes.

“This whole signaling cascade both protects and regenerates vulnerable cells in the airway and the pancreas,” said Philip Beachy, PhD, professor of urology and of developmental biology. “If this circuit is disrupted, the damage is much worse — specialized airway cells are lost, and the stem cells can’t divide to repair the damage. We think it’s likely to be important in many other tissues in the body.”

Although the study was conducted in mice, there are tantalizing clues of a similar pathway in humans: People treated with a cancer drug that blocks the pathway are twice as likely as their peers to develop diabetes after their treatment.

“The association is highly significant and gives us early hints that activating this pathway might be protective for people with metabolic syndrome who are beginning to lose beta cell function,” Beachy said.

Beachy, who is the Ernest and Amelia Gallo Professor and a member of the Stanford Stem Cell Institute, is the senior author of the research, which was published online June 9 in Cell. Research scientist William Kong, PhD, is the study’s first author.

Neuroendocrine cells are less than 1% of the total number of cells in the cells that line the airway, which is made up of a type of tissue called epithelium. Some of them cluster together in what are called neuroepithelial bodies and play an important role in sensing oxygen levels and modulating immune responses in the lungs. But others, especially those in the tracheal airway, are solitary, nestled alone among other types of epithelial cells. It’s not been clear until now exactly what function these solitary neuroendocrine cells perform.

The hedgehog family

Beachy’s laboratory has focused on the function of a protein family called Hedgehog proteins since Beachy identified the first member in fruit flies in 1992. Members of the family are best known for their critical function in embryo patterning in early development, but they also aid in the rejuvenation of many types of tissue. Desert hedgehog is one of the least studied of the three family members (the others are Sonic hedgehog and Indian hedgehog).

Previous work in Beachy’s lab showed that stem cells in the epithelial lining of the bladder respond to a signal cascade initiated by Sonic hedgehog to regenerate the bladder lining after bacterial infection. They wondered if hedgehog proteins were involved in the repair of damage in other epithelial tissues like the airway.

When Kong used a technique called bulk RNA sequencing to search for genetic messages encoding any of the hedgehog family members in the cells of the trachea, they detected a faint signal for the Desert hedgehog protein, but not for the other two family members. When they engineered mice in a way that caused cells expressing the Desert hedgehog protein to become fluorescent, they saw that the solitary neuroendocrine cells were making the Desert hedgehog protein.

Further research showed that the Desert hedgehog protein leaves the epithelium and travels into the layer of tissue beneath the epithelium, called the mesenchyme. There, it triggers cells to begin producing another protein called Gli1. When the airway cells sense damage, Gli1 induces the expression of a protein called IL-6 that triggers stem cells in the epithelium called basal cells to begin dividing and specializing to repair the damage.

This crosstalk between tissue layers, which the researchers call epithelial-mesenchymal feedback, protects and regenerates specialized cells in the lung epithelium, including multi-ciliated cells that use their feathery arms to sweep particles and viruses out of the lungs and secretory cells that make mucus to trap unwanted invaders. In the absence of these cells, viruses and toxins can penetrate much more deeply into the lungs

Protective process

The entire process happens within hours of toxin exposure in a coordinated cascade that eventually includes even non-Gli1-expressing cells of the airway.

“At each stage, the signal is amplified until the entire trachea is impacted,” Beachy said. “This rapid response not only protects the epithelial cells from dying but it also activates a regenerative response.”

The consequences of impeding this protective message are severe.

“If this signal cascade is disrupted, the damage is much worse. Ciliated and secretory cells are lost, and the basal cells don’t divide. In fact, it’s all they can do to stretch out and try to cover the injured area,” Kong said.

Both Desert hedgehog and Gli1 are critical to the repair process. Mice unable to produce Desert hedgehog or Gli1 were much more sensitive to exposure to sulfur dioxide gas, which is an environmental pollutant and mimics the damage inflicted by other inhaled toxins. While control mice lost 85% of ciliated cells and 41% of secretory cells within 24 hours, mice lacking the protein lost 96% of ciliated cells and 88% of secretory cells during the same time.

Activating the hedgehog signaling pathway with a small molecule dramatically increased cell survival after sulfur dioxide exposure: 66% of ciliated cells and 82% of secretory cells survived in treated animals, versus 9.7% of ciliated cells and 43% of secretory cells in control animals.

Kong next tested the effect of the Desert hedgehog pathway activation on mice infected with influenza and the virus that causes COVID-19. Although no mice unable to make Gli survived more than five days after infection with influenza, all mice treated with the small molecule activator survived at least eight days after infection. Mice infected with the virus that causes COVID-19 that were unable to activate the Desert hedgehog pathway suffered extensive loss of ciliated cells in the airway.

Finally, the researchers turned their attention to the pancreas, which has a similar tissue organization as the airway. They found that the insulin-producing beta cells, which are a type of neuroendocrine cell, also make Desert hedgehog and that the organ exhibits the same epithelial-mesenchymal feedback loop with IL-6 to protect the vulnerable cells.

The researchers are now exploring whether and how the hedgehog pathway could be activated in humans to prevent lung damage in people exposed to airborne toxins or who are at risk for diabetes.

“We have reasons to think it might not be a good idea to activate the hedgehog pathway long term,” Beachy said. “We are considering how to stimulate the pathway in a targeted way, either delivering it to the airway with an aerosol or targeting it to the pancreas. And we have early hints it might be possible.”

Reference: Kong W, Lu WJ, Dubey M, et al. Neuroendocrine cells orchestrate regeneration through Desert hedgehog signaling. Cell. 2025;0(0). doi:10.1016/j.cell.2025.05.012

This article has been republished from the following materials. Note: material may have been edited for length and content. For further information, please contact the cited source. Our press release publishing policy can be accessed here.

In our study, we evaluated the alignment of three LLMs– Gemini, Copilot, and ChatGPT-4.0– with consensus answers provided by a panel of experts. The findings indicate that Gemini consistently demonstrated a higher degree of agreement with the expert consensus compared to the other LLMs examined. Specifically, Gemini excelled in compatibility with both the majority of the consensus members’ answers and with the responses of any of the nine reviewers who participated in the consensus, regardless of majority opinion. Additionally, among the three commonly utilized LLMs, Gemini was the most aligned in queries where a perfect or excellent consensus was achieved.

The extent to which LLMs can be utilized in clinical decision-making processes remains a topic of growing interest. To shed light on this issue, several studies have been conducted—and continue to be conducted—comparing the performance of various LLMs [9,10,11,12]. In a study involving 134 clinical cases, the diagnostic, therapeutic, and management-related decision-making accuracy of three different LLMs was evaluated, and Gemini was found to have the lowest overall performance [9]. In another study focused on surgical planning in glaucoma patients, Gemini demonstrated 32% lower agreement compared to ChatGPT-4, indicating inferior performance [10]. Similarly, when assessed as an intraoperative decision-support tool in plastic surgery, Gemini again exhibited suboptimal performance relative to ChatGPT-4 [11]. Conversely, another study evaluating the ability of ChatGPT-4 and Gemini to assess diagnosis and treatment plans for patients with acute cholecystitis found comparable performances between the two models [12]. In contrast to these previous studies, our study found that, in the context of selecting the appropriate imaging modality for patients with renal colic, Gemini produced responses that were more aligned with those of the expert consensus panel.

One potential factor contributing to variability in LLM responses is the influence of different guideline sources used during model training. While our study utilized the 2019 consensus report [7] as the reference standard, it is possible that the UK NICE guidelines, which have recommended low-dose non-contrast CT as the first-line imaging modality since January 2019, were included in the training datasets of the evaluated LLMs. This difference in guideline exposure may have contributed to discrepancies in LLM recommendations.

However, this does not undermine the validity of our findings, as our study was specifically designed to assess how well LLMs align with an established expert consensus rather than to evaluate the absolute correctness of their responses based on multiple guideline sources. In clinical practice, variations in recommendations across different guidelines are well-recognized and do not indicate an inherent flaw in an individual guideline or its interpretation. Future research could further investigate how different LLMs integrate and prioritize diverse clinical guidelines, providing additional insights into their decision-making processes.

Gemini demonstrated a significantly higher level of concurrence with the consensus, providing responses that were largely similar to those of the majority of consensus participants. This suggests that Gemini may have better performance in understanding and appropriately interpreting clinical case examples consistent with expert opinions. In contrast, only 41.4% of the questions were answered by ChatGPT-4.0 and Copilot in agreement with the consensus, indicating a less robust alignment with expert guidelines. Notably, Gemini achieved an agreement rate of 82.7% when comparing its overall responses with those of the nine reviewers. These findings indicate that Gemini could be a more credible tool for applications requiring strong conformity with expert guidelines [7]. In addition, Gemini’s high rate of agreement with expert evaluations suggests that it may assist clinicians in making imaging decisions in patients with renal colic.The high level of agreement of the Gemini scale with the highest scoring responses suggests that it may have the potential for widespread adoption in professional and academic contexts if supported by future studies. In addition, Gemini and other LLMs can have more sensitive evaluation capabilities with each new update. With the use of more data sets and the development of more fine analysis capabilities with each new update, more accurate results can be achieved in the clinical context. Future studies can contribute to the increase of our knowledge and experience in this subject by focusing on how the reliability and accuracy rates of these LLMs, especially Gemini, can be improved with new updates.

Although the use of AI-enabled LLMs may attract attention with their performance in clinical case assessments, the integration of AI technologies into healthcare systems raises significant ethical and legal concerns that warrant careful consideration [13, 14]. As these complex models become increasingly embedded in critical clinical decision-making processes, it is crucial to meticulously assess the multifaceted risks and responsibilities associated with their potential impact on patient outcomes. A primary and pressing ethical issue pertains to the transparency and explainability of the decision-making mechanisms within AI systems. Healthcare providers must be able to comprehend and trust the recommendations and rationale generated by these AI systems in order to maintain patient confidence and ensure appropriate treatment. The lack of explainability can lead to profound challenges in establishing clear lines of accountability, making it profoundly difficult to determine whether the healthcare professional or the AI system bears responsibility for an erroneous or suboptimal decision that could significantly impact a patient’s well-being [15]. This adaptation is essential for safeguarding patient trust and ensuring that the integration of AI into healthcare supports, rather than undermines, the quality and reliability of patient care.

Another concern regarding the usability of LLMs in clinical case evaluations is the legal processes related to the compliance of the use of AI in healthcare with standards regarding patient privacy and data protection [16]. In Turkiye, the Law on the Protection of Personal Data (KVKK) imposes strict regulations on the management and disclosure of patient data [17]. Similarly, in the United States, the Health Insurance Portability and Accountability Act (HIPAA) sets stringent rules on patient data handling [18]. AI systems must be designed to comply with these requirements in both jurisdictions, ensuring the safeguarding of patient information against unauthorized access and breaches. This highlights the importance of ensuring that AI systems in healthcare settings are designed to comply with stringent data protection regulations—such as KVKK in Türkiye and HIPAA in the U.S.—to safeguard patient information from unauthorized access and breaches. Legally, the utilization of AI in healthcare must adhere to standards pertaining to patient confidentiality and data protection [16]. The Health Insurance Portability and Accountability Act in the United States imposes strict rules on the management and disclosure of patient data [17]. AI systems must be engineered to adhere to these requirements, guaranteeing the safeguarding of patient information against unauthorized access and breaches.

Limitations

This study has several limitations that should be considered when interpreting its findings. Firstly, the variability in the types and phrasing of questions posed to the LLMs can influence their responses, introducing variability that may affect the conclusions. However, the use of a standardized set of 29 clinical scenarios helps mitigate this variability, ensuring a consistent basis for comparison.

Secondly, the study’s generalizability is limited by the specific scenarios and questions presented. While the results may not apply to all medical inquiries, the selected scenarios are representative of common clinical situations in emergency departments. This relevance supports the applicability of the findings within the intended context. Thirdly, each vignette was presented only once in our study. Repetitive testing might increase the quality and robustness of the results of the study. Another limitation is that no power analysis was applied in our study. Instead, the study was designed using all vignettes in the consensus report. Finally, Copilot uses the infrastructure of ChatGPT-4.0. However, it also integrates the Microsoft database into its infrastructure, focusing on producing more balanced, creative and precise answers. Therefore, although they use similar infrastructures, they use different approaches to reach results on the given data, which distinguishes these two LLMs models from each other.

Despite these limitations, the study’s design and scenario selection provide a strong foundation for its conclusions. Future research can further address these limitations to enhance our understanding of LLMs in healthcare settings.

Specialty

Please chooseI’m not a medical professional.Allergy and ImmunologyAnatomyAnesthesiologyBiostatisticsCardiac/Thoracic/Vascular SurgeryCardiologyCritical CareDentistryDermatologyDiabetes and EndocrinologyEmergency MedicineEpidemiology and Public HealthFamily MedicineForensic MedicineGastroenterologyGeneral PracticeGeneticsGeriatricsHealth PolicyHematologyHIV/AIDSHospital-based MedicineI’m not a medical professional.Infectious DiseaseIntegrative/Complementary MedicineInternal MedicineInternal Medicine-PediatricsMedical Education and SimulationMedical PhysicsMedical StudentNephrologyNeurological SurgeryNeurologyNuclear MedicineNutritionObstetrics and GynecologyOccupational HealthOncologyOphthalmologyOptometryOral MedicineOrthopaedicsOsteopathic MedicineOtolaryngologyPain ManagementPalliative CarePathologyPediatricsPediatric SurgeryPharmacologyPhysical Medicine and RehabilitationPlastic SurgeryPodiatryPreventive MedicinePsychiatryPsychologyPulmonologyRadiation OncologyRadiologyRheumatologySubstance Use and AddictionSurgeryTherapeuticsTraumaUrologyMiscellaneous

Beth Parsons/BBC

Beth Parsons/BBC

Beth Parsons/BBC