Researchers have developed a new artificial intelligence system designed to detect multiple diseases just by analyzing images of the eye. Known as Reti-Pioneer, this computer program examines scans of the retina, the light-sensitive tissue at the back of the eye. By reviewing these pictures, the AI can check for signs of several different health conditions all at once.
To build this tool, the research team trained the AI using a massive collection of eye images gathered from various healthcare centers, ranging from local community clinics to large, specialized hospitals. This broad mix of data helps ensure the program works well for different types of patients. The creators then tested the system using outside data, confirming that it performs reliably beyond the original images it was trained on.
The system has already been put to the test in real healthcare settings. Through an initial pilot study and a silent trial, where the AI ran in the background without affecting patient care, researchers showed that the tool is fast and practical for doctors to use. However, the study does not yet specify exactly which diseases the AI targets, the exact number of patients tested, or the precise accuracy rates of the diagnoses.
Artificial intelligence is moving fast, and doctors are testing whether popular models can handle complex medical decisions. In a recent study, researchers evaluated whether the AI model DeepSeek could design treatment plans for breast cancer patients. They gathered anonymized medical records from 163 patients, including details like tumor characteristics and past therapies, and fed them into the AI using its "Deep Thinking R1" mode. The goal was to see if the machine could match the expertise of a human multidisciplinary team, which is a group of specialists who collaborate to guide patient care.
The AI proved to be remarkably fast, generating a personalized treatment plan for each patient in under 20 seconds. However, the accuracy left a lot to be desired. Two senior clinical experts reviewed the results and found that only 25.2 percent of the AI-generated plans perfectly matched the regimens created by the human team.
When researchers looked at a specific group of 96 patients to evaluate the AI as a decision support tool, they found an error rate of nearly 47 percent. The authors caution that this is a small, early-stage study conducted at a single hospital over just a two-week period. Still, the findings clearly show that the AI makes significantly different choices than human experts and requires strict professional oversight.
Over the past few years, a health technology company called OpenEvidence has captured the attention of the medical community by offering a free artificial intelligence chatbot. The tool is powered by a large language model, which is a type of artificial intelligence trained on vast amounts of text to understand and generate human-like responses. OpenEvidence designed its specific model to help doctors navigate complex patient care decisions.
Rather than trying to sell this software to hospital administrators, the company offered it directly to medical professionals. The approach was a massive hit. OpenEvidence claims that about 650,000 active physicians in the United States now use the platform, with medical trainees being particularly frequent users.
Now, the company is attempting a major strategic shift. After building a huge user base of individual doctors, OpenEvidence is pivoting to target large hospitals and health institutions directly.
The White House is expanding its prescription drug discount website, TrumpRx, by adding more than 600 generic medications to its catalog. Generic drugs are affordable copies of brand-name medications that have the same active ingredients and clinical effects. This expansion aims to simplify the process of shopping for affordable healthcare by giving consumers a centralized place to search for the best prices.
To build this massive catalog, the government is teaming up with major players in the healthcare savings space. The updated platform relies on new partnerships with Amazon Pharmacy, GoodRx, and entrepreneur Mark Cuban's Cost Plus Drug Company. Instead of checking multiple websites to find a good deal, patients can now use the single government platform to locate the lowest possible cost for their generic prescriptions.
Childhood vaccines are a powerful tool for preventing infectious diseases, but keeping track of which children might miss their scheduled shots is a major challenge for public health officials. To tackle this problem, researchers are turning to artificial intelligence. By analyzing large amounts of patient data, AI programs can hunt for hidden patterns that humans might overlook. A recent comprehensive review of medical literature identified twelve studies that evaluated how well AI can predict whether a child will fall behind on their routine immunizations.
The early results from these algorithms are highly encouraging. In nine of the twelve studies, the artificial intelligence models achieved an accuracy rate of over 80 percent. The majority of this research focused on low-income and middle-income countries, relying on broad survey data to train the software. When the AI processed the information, three main warning signs consistently emerged as top predictors of missed vaccines: a mother's access to prenatal care, the specific setting where the child was delivered, and whether the family had a history of missing previous appointments.
Despite these impressive initial numbers, scientists warn that the technology is still in its early stages. A major limitation is that none of the twelve studies performed external validation, which is the process of testing the AI on completely new, outside data to prove it works in new environments. Without this crucial step, the algorithms might just be memorizing the specific datasets they were trained on. Because of this issue, which data scientists call overfitting, these prediction models are not yet ready for everyday use in clinics and hospitals.
For children battling cancer or recovering from a stem cell transplant, receiving blood transfusions is a regular part of treatment. But in many hospitals, doctors must manually type a new order into the electronic medical record every single time a child needs blood. This repetitive task can slow down care, open the door for mistakes, and contribute to physician burnout. To fix this, researchers at a pediatric hospital tested a new approach: reusable digital therapy plans that allow doctors to set up recurring blood orders in advance.
The new system made a noticeable difference for both patients and medical staff. By tracking more than 2,700 transfusions over a year, researchers found that the reusable plans significantly cut down the time it took to get blood to the patient after a laboratory test flagged the need for a transfusion. The streamlined system also boosted safety. It increased the rate of correct, guideline-approved dosing and ensured patients consistently received irradiated blood products, which are specially treated to prevent dangerous immune reactions.
The streamlined tool was a major hit with the medical team. By preventing an average of four disruptive overnight order requests per night, it eased the administrative burden on doctors and nurses. A survey revealed that 95 percent of the health care workers involved wanted to keep using the reusable plans. However, the researchers note that this study took place at a single academic center focusing solely on children, meaning the results might look different in adult hospitals or other healthcare systems.
When surgeons perform a minimally invasive liver removal, known as a hepatectomy, they have to navigate a highly complex maze of blood vessels and tissue. To get a better map of this delicate anatomy, medical teams are increasingly turning to extended reality. This technology, which includes virtual and mixed reality, allows doctors to visualize the surgical site in three dimensions before they even make an incision and provides crucial guidance during the operation itself.
A recent study looked at the medical records of 102 patients who underwent minimally invasive liver surgeries using either robotic instruments or traditional keyhole approaches, known as laparoscopy. Researchers divided the cases into two groups: those whose surgeries were assisted by extended reality and those planned without it.
The findings revealed an interesting dynamic. Surgeons consistently utilized extended reality tools for significantly more complex operations, including a much higher proportion of complicated anatomical removals. Yet, despite tackling harder cases, the surgeons using extended reality achieved safety levels comparable to those performing easier, traditional surgeries. The patients in the high-tech group did not experience any statistically significant differences in blood loss or severe complications, suggesting the technology effectively leveled the playing field for difficult procedures.
Colorectal cancer is the second leading cause of cancer-related deaths in the United States. Medical guidelines recommend that screening begins at age 45, but getting people to actually complete these tests remains a major challenge, especially in medically underserved communities. To find a better way to reach patients, researchers tested whether simple text messages could nudge adults to get screened. The study tracked more than 4,800 patients across two large community health networks in Texas and California that serve vulnerable populations.
Participants were divided into four groups. One group received no text messages, while others received either a single reminder, a three-week series of messages, or a six-week series. Researchers then tracked whether patients completed one of three common screening methods within 90 days. These methods included a traditional colonoscopy or at-home stool tests, specifically the fecal immunochemical test and Cologuard. The results showed that a moderate approach worked best. Patients receiving the three-week text series were significantly more likely to complete their overall screening than those who received no texts, and even a single message provided a slight boost.
Interestingly, sending too many texts seemed to dilute the benefit. The six-week messaging group did not show a significant overall advantage over the group receiving no messages. Researchers suspect this is due to message fatigue, where patients simply tune out frequent notifications. However, texting was uniquely effective for patients using the Cologuard test, making them up to seven times more likely to complete it. While these findings are promising, the researchers caution that the study was limited to two specific health center networks, meaning the exact results might not perfectly generalize to every hospital or community.
For people navigating a cancer diagnosis, the internet often feels like a lifeline. Online support groups offer a convenient space to find information, share experiences, and connect with others who truly understand. A comprehensive new review of 168 studies looked at exactly how these digital communities impact cancer patients and survivors. The researchers found that when these groups work well, they provide a powerful mix of emotional comfort, practical advice, and a sense of purpose from helping others. This combined support can significantly lower feelings of isolation and boost a patient's overall mental well-being.
However, logging on is not a guaranteed cure for loneliness. The research team discovered that online support groups are not universally beneficial. A community's success heavily depends on how well it matches an individual's specific needs, attitudes, and coping style. Technology features, like whether users can stay anonymous and protect their privacy, play a massive role in whether someone feels safe. Group dynamics, such as active moderation and empathetic interactions, are equally crucial. If a group is a bad fit for a patient, participating can actually increase their distress rather than relieve it.
To make these platforms more effective, the researchers emphasize that digital communities must be carefully designed and consistently monitored. Developers and moderators need to focus on building trust, ensuring privacy, and actively managing conversations so that users feel understood, accepted, and valued.
We all stumble over our words occasionally, filling the gaps with a brief pause or a filler word. Researchers have found that these everyday speech patterns are actually closely tied to our executive function, which is the brain's command center for managing tasks like memory, focus, and planning.
To explore this connection, scientists used artificial intelligence to analyze natural, everyday conversations. They discovered that AI can evaluate these speaking patterns to predict a patient's cognitive performance, offering a unique glimpse into how well the brain is functioning. However, it is important to note that this is still early-stage, proof-of-concept research. The current findings do not yet provide specific quantitative data, such as exact predictive accuracy or the total sample size of patients tested, meaning the technology needs more validation before it becomes a standard medical tool.
Women experience thoughts of suicide and make attempts three times more often than men. Yet, men account for up to 80 percent of all suicide deaths in the United States. This stark contrast highlights a major public health emergency. Historically, experts have attributed this high male mortality rate to a few key factors, including higher impulsivity, a lower reported fear of death, and easy access to firearms.
A recent report from the Crisis Text Line adds another crucial explanation to this tragic puzzle. The data shows that men are significantly less likely to reach out for help than women. Even when struggling with profound mental health crises, men often remain silent rather than connecting with support networks or crisis counselors.
While these insights are revealing, they do come with an early-stage caveat. Because the findings are based on data from a specific text-based crisis lifeline, they primarily reflect the behaviors of people who naturally gravitate toward texting for support. This means the data might not perfectly represent the entire national or global population. However, the trend points to a clear gap in who is utilizing available mental health resources.
When a patient undergoes an MRI scan to check for prostate cancer, a medical professional must carefully line up the machine to capture the exact right area. This process, known as field of view planning, is especially tricky for prostate imaging because the gland needs to be viewed from complex, angled perspectives. As a result, the quality of the scan heavily depends on the individual skill of the operator, leading to variations in how the images turn out.
To solve this problem of inconsistency, researchers developed an artificial intelligence tool to automate the scanner setup. They used a deep learning framework, a type of computer algorithm highly skilled at recognizing patterns in images, and trained it on nearly 1,500 past prostate MRI exams. The team then tested the AI on over 500 new exams from different hospitals to see if it could correctly position the digital image slices without human help.
The results showed that the AI matched the performance of human experts. The automated field of view predictions overlapped with the ideal manual setups by up to 88 percent, and the scanning angles were off by an average of less than five degrees. When clinical experts reviewed the AI's automated setups across different hospitals, they accepted them between 95 and 100 percent of the time, proving the tool is highly accurate.
