DR. MATTHEW GORDON CROWSON

Plain English
This study looked at a new computer program designed to help doctors measure bilateral vestibular schwannomas (tumors) in patients with neurofibromatosis type 2 (NF2) using MRI scans. The program showed strong performance, achieving an accuracy score of 0.94 for sporadic tumors and 0.87 for bilateral tumors, making it useful for quickly assessing tumor sizes to inform treatment decisions. This is important because it can help streamline the care for patients with NF2, who often have multiple tumors and complex medical needs. Who this helps: Patients with neurofibromatosis type 2 (NF2).

Plain English
This paper examines how using health data for profit can create ethical problems, especially with the growing use of electronic records and artificial intelligence in healthcare. It argues that current laws don't do enough to protect patient privacy or ensure transparency about how health data is used, particularly with new regulations coming in January 2025. The authors emphasize the need for better rules that put patients first to ensure their data is handled responsibly and ethically. Who this helps: Patients and their rights to privacy in healthcare.

Plain English
This research paper examines how Epic Systems has become the leading provider of electronic health records in the U.S., servicing 42.3% of hospitals and controlling over half of the country's hospital beds. Since 2018, the market for electronic health records has greatly concentrated, with Epic capturing nearly 70% of new hospital contracts in 2024, raising concerns about competition and potential abusive practices. The paper argues that this dominance could hinder innovation and patient care, suggesting that government action is needed to promote fairness and efficiency in healthcare IT. Who this helps: Patients and healthcare providers who rely on effective and accessible electronic health records.

Plain English
Researchers developed a blood test that detects HPV cancer DNA years before oropharyngeal cancer (a common throat cancer) develops, finding the virus DNA in 79% of patient blood samples taken up to 7.8 years before diagnosis, while showing no false positives in healthy people. Using advanced computer analysis, they improved the test to detect the cancer signal in 96% of cases up to 10 years before symptoms appear. This discovery could enable doctors to catch this aggressive cancer much earlier, when treatment is more likely to succeed and causes less harm.

Plain English
This study developed a computer program that automatically analyzes the movements of surgical drills used in ear surgeries, specifically mastoidectomy. The program helps assess a surgeon's technical skills in real time, which could improve surgical training and education. This is important because better training can lead to safer and more effective surgeries. Who this helps: This helps surgeons and medical students learning ear surgery techniques.

Plain English
This study looked at the use of 3D-printed models of temporal bones for surgical practice, focusing on how much it costs and how long it takes to make them compared to using real human bones. Researchers found that they could produce 60 resin models at a cost of just $6.40 each, totaling about $384, which is much cheaper than the price of one real temporal bone, which can range from $400 to $700. This matters because it shows that 3D-printed bones can be a more affordable option for medical training, helping more students get hands-on practice without the high costs of using cadavers. Who this helps: This benefits medical students and training programs looking for cost-effective ways to teach surgical skills.

Plain English
This study looked at ways to automatically assess the skills of doctors performing ear surgeries, specifically focusing on methods that are partially or fully automated. Researchers reviewed 1,141 studies but found only 21 that met their criteria, most of which centered on a specific ear surgery called mastoidectomy and used virtual reality (VR) simulations to evaluate performance. The findings highlight that while there are advancements in using AI for skill assessment, most existing research is limited and does not effectively apply to a variety of real-world surgical situations. Who this helps: This benefits doctors and surgical trainees aiming to improve their skills in otology and neurotology procedures.

Plain English
This study looked at how artificial intelligence (AI) can improve the training of surgeons, particularly in the field of ear, nose, and throat (ENT) medicine. It reviewed 34 studies and found that AI tools can accurately assess surgical skills, with classification accuracy ranging from 66% to 100% for various tasks. This is important because it allows for more objective evaluations, which can help train surgeons more effectively and lead to better patient outcomes. Who this helps: This benefits surgical trainees and educators in otolaryngology.

Plain English
Researchers developed a new blood test using advanced DNA sequencing to detect HPV-associated head and neck cancers early, before symptoms appear, and compared it to three other blood-based detection methods. The new sequencing test detected cancer 98.7% of the time it was present and correctly identified healthy people 98.7% of the time—significantly better than existing blood tests. This breakthrough matters because HPV-related head and neck cancers are becoming more common, yet doctors currently have no way to catch them early when treatment is most effective.

Plain English
Researchers developed a highly sensitive blood test that detects tiny amounts of HPV cancer DNA left in patients after surgery for HPV-related head and neck cancer. They found that patients whose blood still contained this cancer DNA had much worse outcomes—only 60% stayed cancer-free for 2 years compared to 100% of patients with no detectable DNA—and this blood test was better at predicting who would relapse than current methods doctors use. The test could also spot cancer coming back up to 17 months before it shows up on scans or causes symptoms, giving doctors an early warning system to decide whether patients need additional treatment like chemotherapy or radiation.

Plain English
This study looked at how voice data is currently collected and stored by voice centers across North America, focusing on how these practices affect research using artificial intelligence. Out of 72 practitioners surveyed, only 28% used standardized methods for collecting voice data, which is crucial for effective research. The findings show that many practitioners face challenges like a lack of uniform protocols and insufficient resources for data management, making it harder to conduct larger collaborative studies in this field. Who this helps: This helps voice disorder patients and researchers looking to improve treatments through AI advancements.

Plain English
This study focused on using computer vision technology to automatically recognize instruments and anatomy during ear surgery. The researchers developed a model that successfully identified these elements in high-definition videos of the surgery. This is important because it could lead to more accurate and efficient surgeries, ultimately improving patient safety and outcomes. Who this helps: This helps surgeons and patients undergoing ear surgery.

Plain English
This study focused on evaluating the size of ear drum holes, known as tympanic membrane perforations (TMPs), using a computer vision model instead of traditional visual inspection, which can be unreliable. The results showed that the computer model had a very small error of only 0.8% in estimating the size of TMPs, while doctors had a much larger average error of 11%. This matters because using a computer model could improve the accuracy of diagnosing ear conditions and help ensure patients receive better treatment. Who this helps: This helps patients with ear problems by providing more accurate assessments of their condition.

Plain English
This study looked at whether an automated tool for identifying vestibular schwannomas (VS) in MRI scans, developed from a specific group of patients, also works well for patients from a different group with varied imaging results. They found that while the tool was able to accurately identify 7 out of 10 tumors with an average accuracy of 87%, it missed detecting 3 small tumors altogether. This is important because it shows that while automated segmentation can be promising, it still needs improvement to work effectively across different patient backgrounds and imaging systems. Who this helps: This benefits doctors and patients with vestibular schwannomas by potentially improving tumor detection methods.

Plain English
This study looked at how vegetation has changed over time in the Greater Côa Valley in Portugal, where much land has been abandoned. Researchers found that some areas naturally recovered well, creating diverse habitats for wildlife, while others did not recover quickly. These findings help determine when to use different methods for restoring nature to ensure the best outcomes for biodiversity. Who this helps: This helps conservationists and environmental managers working on rewilding and restoration projects.

Plain English
This research paper looks at how artificial intelligence (AI) can be effectively used in ear, nose, and throat (ENT) surgery. It emphasizes the need for careful management of data, security, and ethical standards to make AI successful in this field. By working together across different specialties and following regulations, AI can enhance patient care, make surgeries safer, and improve how hospitals operate. Who this helps: This benefits patients and doctors in the field of otolaryngology.

Plain English
This study looked at how to track the spread of artificial turf in cities using satellite images. Researchers found that current methods are not very effective: large areas of artificial turf were spotted no better than small areas, making it hard to get accurate information about where artificial lawns are growing. This is important because as cities change, we need to understand how these changes affect urban nature and the benefits it provides to people. Who this helps: This helps city planners and environmental advocates.

Plain English
This study looked at how agricultural land use affects the levels of nitrogen and phosphorus in rivers across England, especially in relation to population density. It found that in areas with low population density, agricultural sources contributed significantly to water pollution, while in densely populated areas, pollution mainly came from waste water treatment plants. The research highlights the need to focus pollution management efforts on rural areas, where farming is a major contributor to nutrient runoff. Who this helps: This helps environmental managers and policymakers targeting water quality improvements.

Plain English
Researchers developed a blood test that can detect a common cancer caused by HPV (the virus linked to cervical cancer) years before symptoms appear—up to 10 years early in some cases. The test works by finding tiny pieces of HPV DNA floating in the bloodstream, which the cancer releases long before the disease becomes noticeable. When they tested blood samples taken years before patients were actually diagnosed with cancer, the test caught 79% of future cancer cases while giving zero false alarms in healthy people. Using artificial intelligence to analyze the blood samples improved detection even further, catching 96% of cases. This matters because unlike cervical cancer, there's currently no screening test for this type of HPV-related throat cancer—the most common HPV cancer in America. A blood test that finds it a decade early could save lives by catching cancer when it's easiest to treat.

Plain English
This study looked at how machine learning can better predict the length of surgeries in ear, nose, and throat (ENT) specialty, analyzing data from 50,888 cases. The researchers found that using machine learning models, particularly CatBoost, improved the accuracy of predicted surgery times, reducing errors by nearly 10 minutes compared to current methods. This is important because better predictions can lead to more efficient use of operating rooms and lower costs for hospitals. Who this helps: This helps hospitals and surgical teams in optimizing scheduling and resources.

Plain English
This study compared the ability of human doctors to diagnose middle ear infections in children with a computer program trained to recognize these conditions from images. The computer model correctly identified the issue 95.5% of the time, while the average accuracy for human clinicians was only 65.0%. This is important because using this technology could reduce misdiagnoses, leading to fewer unnecessary treatments and better care for children with ear infections. Who this helps: This helps pediatricians and children suffering from ear infections.

Plain English
This study focused on creating a model that predicts how much benefit a person with hearing loss will get from using hearing aids. Researchers analyzed data from 1,286 patients and found that the best model could predict listening improvement with an accuracy of 0.80 on a scale of 1 to 5. Key factors that affected hearing aid success included younger age, better hearing in one ear, and shorter time spent using hearing aids. Who this helps: This helps patients by providing a clearer understanding of their potential benefit from hearing aids.

Plain English
This study focused on developing a computer program that can predict sleep apnea events in children by analyzing nasal air pressure during sleep. Researchers collected data from 28 pediatric patients and found that their model accurately identified sleep events 77.5% of the time, which is better than the 53.8% accuracy of trained sleep doctors. This improvement in diagnosis can lead to better treatment for children with sleep apnea, making it easier for healthcare providers to identify and address these issues. Who this helps: This helps pediatric patients and their doctors by improving the diagnosis of sleep apnea.

Plain English
Researchers developed a new technology that can create artificial images of the tympanic membrane (eardrum) to help improve training and diagnostic tools in ear surgery. They trained a computer model on 202 high-definition images from real surgeries, achieving a quality measure showing it learned the important features well. The model was validated with 65 experts who reviewed nearly 1,000 images, successfully distinguishing between real and synthetic images to some extent, with a 66% accuracy in identifying these images as fake. Who this helps: This technology benefits doctors and medical trainees by providing them with more training material and improving diagnostic tools.

Plain English
This study looked at whether machine learning can help identify postural dysfunction in middle-aged and older adults using data from a week-long physical activity tracker worn on the waist. Researchers analyzed data from 1,625 participants, finding that about 47% failed a balance test, indicating postural dysfunction. The machine learning models used, particularly two types called SVM and XGBoost, correctly identified dysfunction with an accuracy of about 73%, showing that this technology could effectively assess balance issues in everyday settings. Who this helps: This benefits older adults and healthcare providers by improving the identification of balance problems.

Plain English
This study focused on creating a tool to predict which patients might experience long-lasting dizziness after surgery for vestibular schwannoma, a type of tumor affecting the inner ear. Out of over 1,100 patients analyzed, 63% reported no dizziness post-surgery, while 17% did report dizziness a year later. The findings matter because they can help doctors better inform patients about their risks and manage their recovery more effectively. Who this helps: This helps patients undergoing vestibular schwannoma surgery.

Plain English
This study looked at how well a computer program could read and classify doctors' notes to determine the appropriate billing codes for outpatient visits. The researchers found that the program correctly identified these codes 74.8% of the time when using full notes, but this dropped to only 48.2% accuracy when the notes were stripped of personal patient information. This matters because accurate billing is essential for healthcare funding and patient care, and understanding how de-identification affects accuracy is important for protecting patient privacy while ensuring efficient billing. Who this helps: This helps doctors and healthcare providers streamline the billing process.

Plain English
This study looks at how to effectively manage and oversee the use of artificial intelligence (AI) in the field of ear, nose, and throat surgery. The researchers found that having clear guidelines and policies for using AI can improve patient care and safety. This is important because it helps ensure that new technologies enhance medical practices rather than create confusion or risks. Who this helps: This helps doctors and healthcare providers in otolaryngology.

Plain English
Researchers used a large database of patient health records to create machine learning tools that predict whether oral lesions would turn cancerous. They found that 34% of patients with oral lesions and 54% of those with dysplasia (a precancerous condition) developed cancer. The best model accurately predicted cancer risk at 86% for all lesions tested. Who this helps: This research benefits doctors by providing tools to identify patients at higher risk for oral cancer.

Plain English
This study looked at various tools and methods used to diagnose problems related to balance and dizziness, known as vestibular disorders. The researchers reviewed 18 different tools, including questionnaires that patients fill out and tests given by healthcare professionals. They found that most tools were effective at accurately identifying different types of vestibular issues, though many studies had weak evidence supporting their use. Who this helps: Patients experiencing dizziness and balance problems can benefit from better diagnostic tools in their evaluations.

Plain English
This study looked at the rates of fatal falls among older adults in the U.S. from 2001 to 2018, finding that there were over 465,000 deaths due to falls during this time. Men experienced a higher death rate (9.89 per 100,000) compared to women (6.17 per 100,000), and the death rate increased with age, especially for those over 85 years, who had a staggering rate of 201.1 deaths per 100,000. The findings highlight a critical need for better health resources and policies to address rising fall-related deaths, particularly for the oldest adults. Who this helps: This helps patients, especially older adults and their caregivers, as well as healthcare providers and policymakers.

Plain English
This study reviewed new medical devices and drugs related to ear, nose, and throat (ENT) surgery that were approved by the FDA in 2021. They found 23 devices and 1 drug specifically useful for ENT practices, including features like over-the-counter hearing aids and advanced treatments for sleep apnea. This is important because these innovations can improve treatment options and enhance patient care in various ENT-related issues. Who this helps: Patients seeking better treatment for hearing loss, allergies, and sleep apnea.

Plain English
This study looked at how federated learning, a way for hospitals to collaborate on developing healthcare algorithms without sharing patient data, is currently being used in medicine. Researchers reviewed 13 studies, mainly focused on cancer and radiology, and found that while the majority followed important reporting guidelines, 46% were at high risk of bias. This means improvements are needed to ensure that federated learning can be more effective and reliable in healthcare settings. Who this helps: This benefits patients and doctors by potentially leading to better and more accurate medical treatments and decisions based on combined data insights.

Plain English
This study looked at how well a computer program could identify voices as male or female, which is important for helping transgender individuals who want gender-affirming voice care. Researchers analyzed 278 audio samples and found that their system correctly identified the gender of voices 92% of the time, doing especially well with female voices (94% accuracy) compared to male voices (87% accuracy). This is important because it shows potential for using AI to measure the success of voice treatment for transgender patients more objectively. Who this helps: This benefits transgender individuals seeking voice therapy and healthcare providers who support their treatment.

Plain English
This study looked at how small farmers in Sumatra, Indonesia, feel about blocking drainage canals on their peatland farms as part of a government effort to restore over 2 million hectares of degraded land. Most farmers (over 60%) said they would accept these canal blocks, believing they wouldn’t hurt their crop yields and could reduce the risk of fires. However, some farmers who rely heavily on these canals for farm access were hesitant, concerned that their yields would drop if the blocks were installed. Understanding these mixed feelings is important to ensure that land restoration efforts succeed without harming farmers' livelihoods. Who this helps: This information benefits smallholder farmers and government planners working on land restoration initiatives.

Plain English
This study looked at the connection between hearing loss and depression using advanced computer techniques called machine learning. Researchers found that they could predict depression levels based on hearing loss factors, achieving a mean error rate of about 3.03 points on a depression scale out of 27. Notably, the emotional impact of hearing difficulties, especially during conversations with family and friends, was one of the top predictors of depression scores. Who this helps: This research benefits patients experiencing hearing loss, helping identify those at risk for depression.

Plain English
This study looked at how well an artificial intelligence program could identify fluid in the middle ear of children undergoing surgery for ear problems. The AI correctly classified images of the eardrum 83.8% of the time, which is much better than the accuracy of human doctors in earlier studies. This is important because it can help ensure children get the right treatment for their ear infections, avoiding potential problems from misdiagnosis. Who this helps: Patients with ear infections and their families.

Plain English
This study looked at how well family doctors in Canada recognize and manage sudden sensorineural hearing loss (SSNHL), which is a serious medical condition that needs quick treatment. Out of the 52 doctors surveyed, nearly all (94.2%) agreed that urgent referral to a specialist is necessary for patients with SSNHL, and most (76.9%) would prescribe corticosteroids for treatment. However, tuning fork tests, which can help differentiate types of hearing loss, were not commonly used compared to other methods. Who this helps: This research benefits patients experiencing sudden hearing loss by improving how quickly and effectively they receive care.

Plain English
This study looked at how well doctors followed established guidelines for treating patients with oral cancer by analyzing data from a large U.S. healthcare database. Out of 2,752 patients, only 21.2% received a feeding tube after treatment, and only 18.6% had consultations with multiple specialists, showing that many patients didn’t receive the recommended care. Improving adherence to these guidelines is important because it can lead to better outcomes for patients. Who this helps: Patients diagnosed with oral cancer.

Plain English
This study focused on improving how doctors diagnose and treat Menière's disease, a condition that affects hearing and balance. Researchers identified five important quality indicators to help standardize care, including measures like documenting hearing tests and the use of certain treatments. By implementing these indicators, healthcare providers can offer better, more consistent care which can lead to improved outcomes for patients. Who this helps: Patients with Menière's disease.

Plain English
This study looked at how a new electronic health record (EHR) system helped identify babies at risk for a virus called congenital CMV after they failed initial hearing tests. Out of 19,069 newborns screened, 1,358 failed those tests, and only 38 were tested for the virus, with 2 results coming back positive. The findings showed that the new system led to more babies being tested over time, but many still weren’t followed up on in a timely manner, which is important for their health. Who this helps: This helps newborns and their parents by ensuring better detection of congenital CMV and related hearing issues.

Plain English
This study examined the Glasgow Benefit Inventory (GBI) to see how well it measures the quality of life for adults who received cochlear implants. Researchers found that the average GBI score was 38.6 out of 100, indicating some benefits post-implant, though the score was only weakly linked to improvements measured by other hearing assessments. This matters because while the GBI is a useful tool, it shouldn’t replace more specific measures designed for cochlear implant patients. Who this helps: This helps patients and healthcare providers by providing insights into how cochlear implants might improve quality of life.

Plain English
This study looked at how the COVID-19 pandemic affected the surgery practices of ear and nerve specialists around the world. Researchers surveyed 396 specialists, finding that their average number of surgeries dropped from 5.3 per week before the pandemic to just 0.7 during it, with 71.5% not performing any elective surgeries in that time. These changes are important because they highlight how the pandemic significantly disrupted healthcare for patients needing ear and nerve procedures, as well as the challenges doctors faced with equipment shortages and changes in surgery techniques. Who this helps: This information benefits patients needing ear and nerve surgeries and the doctors who treat them.

Plain English
This study focused on vestibular migraine, which is often overlooked as a cause of balance disorders. Experts identified seven main areas that need improvement, such as better symptom recognition and the development of reliable diagnostic tests and treatment guidelines. They recommended actions like creating a global registry for vestibular migraine and improving how doctors diagnose the condition to ultimately enhance patient care. Who this helps: This helps patients with vestibular migraine by improving diagnoses and treatment options.

Plain English
This study developed a system called 'AutoAudio' that uses deep learning to read and interpret audiogram results, which help diagnose hearing loss. The best performing model achieved an impressive accuracy of 97.5% in identifying different types of hearing loss. This innovation is important because it can improve access to hearing evaluations, especially as the number of people needing these services is expected to rise globally. Who this helps: Patients with hearing loss and healthcare providers.

Plain English
This study looked into how the gut bacteria of elite athletes might affect their athletic performance. Researchers found that focusing on gut health through diet could be beneficial for improving performance, even though there are still few studies directly linking microbiome health to athletic success. This shift in approach suggests that taking care of gut bacteria could give athletes a helpful advantage. Who this helps: This helps athletes and sports nutritionists looking for new ways to enhance performance.