NIH: AI Tool Could Help Doctors Detect Disease Earlier from CT Scans
A recent advancement in artificial intelligence (AI), funded by the National Institutes of Health (NIH), promises to enhance doctors’ ability to glean insights from routine medical scans. This innovative technology could even enable the prediction of certain diseases years before symptoms manifest.
The AI system, known as Merlin, is a machine-learning model that evaluates CT scans, a widely used form of medical imaging. CT scans, or computed tomography scans, compile multiple X-ray images taken from different angles to produce comprehensive images of the body’s internal structures. Physicians commonly employ these scans to identify injuries, tumors, infections, and other health concerns.
Typically, a radiologist—an expert trained in interpreting medical imagery—reviews these scans and prepares a report detailing their observations. However, this process can be time-consuming and may require additional tests for a definitive diagnosis. With a notable shortage of healthcare providers in the United States, researchers are investigating ways to enhance the efficiency of scan evaluations.
The new AI model aims to simplify this review process.
A Powerful New Type of Medical AI
Merlin is classified as a “foundation model,” a novel category of AI systems trained on vast datasets to perform diverse tasks rather than just one specific function.
To develop Merlin, researchers analyzed over 15,000 abdominal CT scans alongside their corresponding radiology reports and nearly one million diagnosis codes from Stanford University. These codes are standardized labels used by healthcare practitioners to categorize diseases and medical conditions within health records.
By examining both the visual data and the written reports, Merlin was able to discern patterns in the scans associated with specific health conditions.
Researchers further tested the system using more than 50,000 CT scans from four hospitals, which Merlin had not encountered previously. The primary aim was to evaluate how effectively Merlin could interpret the scans compared to human-generated medical conclusions.
Predicting Disease Risk Years in Advance
The outcomes were encouraging, according to the NIH. Across a range of medical conditions, Merlin accurately predicted the scan most likely linked to a specific diagnosis over 81 percent of the time, outperforming several other AI systems tailored for this task.
Additionally, the model demonstrated potential in forecasting future disease risk. When reviewing CT scans of individuals who appeared healthy at the time, Merlin could identify those at higher risk for conditions such as diabetes, osteoporosis (bone thinning), or heart disease within the next five years with approximately 75 percent accuracy.
Scientists believe the AI may be detecting subtle warning signs—often referred to as biomarkers—that are challenging for human clinicians to observe.
Gut Bacteria May Influence Cholesterol Levels, New Study Suggests
Emerging research indicates that the bacteria residing in our digestive systems may significantly impact heart health. A recent study published in Microbiology Spectrum revealed that individuals with dyslipidemia—an unhealthy fat balance in the blood—exhibit different gut bacteria and microbial functions than those with normal cholesterol levels.
Dyslipidemia pertains to abnormal lipid levels in the bloodstream, including elevated cholesterol or triglycerides, and serves as a critical risk factor for cardiovascular disease, which encompasses conditions like heart attacks and strokes.
In the study, researchers analyzed stool samples from 1,384 adults, which included 895 individuals with dyslipidemia and 489 without. Utilizing a method known as shotgun metagenomic sequencing, the scientists examined the types of bacteria present in the gut and their biological functions.
Differences in Gut Bacteria
The study revealed significant disparities in the gut microbiome—the collection of microorganisms in the digestive tract.
Specifically, one bacterial species, Bacteroides caccae, was more prevalent in people with dyslipidemia. Earlier research has connected this bacterium to inflammation, which might contribute to metabolic disturbances and alterations in fat processing.
Conversely, individuals without dyslipidemia showed higher levels of other beneficial bacteria, such as Coprococcus eutactus, Coprococcus catus, and Blautia obeum. These microbes are instrumental in producing short-chain fatty acids (SCFAs)—beneficial molecules generated when gut bacteria metabolize dietary fiber.
SCFAs play a role in regulating metabolism, supporting gut health, and may also influence cholesterol levels. A reduction in these beneficial bacteria could disrupt lipid regulation.
Changes in ‘Metabolic Pathways’
In addition to differences in bacterial species, the study also highlighted alterations in the metabolic pathways of gut microbes in individuals with dyslipidemia. Metabolic pathways are crucial chemical processes through which cells generate energy and synthesize molecules.
The participants with dyslipidemia exhibited decreased activity in pathways involved in glycogen production—a key element for energy storage—and peptidoglycan biosynthesis, a process pertinent to bacterial cell wall formation that might influence immune responses.
Researchers also noted reduced levels of pseudouridine, a molecule essential for RNA metabolism and normal cellular function.
Another observation was an increase in genes associated with pathways linked to bacterial pathogenicity—indicating the potential for these microbes to cause disease.
Possible Link to Antibiotic Resistance
The study also examined the gut resistome, which is the collection of antibiotic-resistance genes present in gut bacteria.
Those with dyslipidemia showed a slight increase in the tetQ gene, which provides resistance to tetracycline antibiotics. While the increase was modest, it hints at a possible relationship between metabolic health and antibiotic resistance within the microbiome.
What It Means for Heart Health
These findings do not establish a direct causal link between gut bacteria and dyslipidemia. However, they reinforce the notion that the microbiome may impact lipid metabolism and cardiovascular risk.
The researchers suggest that these results could eventually inform microbiome-based strategies for managing metabolic diseases. This may include dietary adjustments to promote beneficial bacteria, probiotics, or other treatments aimed at modifying the gut microbiome.
For the present, experts stress that maintaining heart health remains reliant on established lifestyle choices, including a balanced diet, regular physical activity, and medical interventions when deemed necessary.
Only Half of U.S. Adults Know About At-Home Colorectal Cancer Screenings, Survey Finds
Colorectal cancer is currently the leading cause of cancer-related fatalities among adults under 50 in the United States. A new national survey has unveiled considerable gaps in public awareness and screening efforts, contributing to this pressing health crisis.
The survey, commissioned by the Colorectal Cancer Alliance—a nonprofit organization—revealed that, while many cancers are experiencing declining mortality rates, the incidence of colorectal cancer continues to rise. To address this alarming trend, the organization is advocating for action during March, designated National Colorectal Cancer Awareness Month.
Gaps in Knowledge and Care
The survey uncovered troubling insights. One-third of all Americans, and nearly half of those under 45, reported that their stomach or bowel symptoms had been dismissed by a healthcare provider. This finding indicates that concerns raised by patients, especially younger individuals, may not always be taken seriously.
Additionally, even when screenings are recommended, over a quarter of adults fail to follow through. A significant knowledge gap may be at play here: only half of Americans are aware that convenient at-home screening tests are available, which could lead to detrimental delays in diagnosis.
Furthermore, many individuals lack awareness of fundamental facts about colorectal cancer. Approximately 52 percent do not know that this cancer type is highly preventable, and 37 percent are unfamiliar with the various screening options available, which include colonoscopies, stool tests (conducted at home), and CT scans of the colon.
Overcoming Barriers to Screening
What are the barriers preventing adequate screening? Commonly cited reasons include financial concerns, fear, and a lack of knowledge about the screening process. However, the survey also identified what motivates individuals to take action. Over 60 percent of respondents indicated they would be more likely to seek screening if they had a better understanding of their personal risk factors.
The silver lining is that a growing number of people are willing to engage in conversations about screening. A significant majority of respondents expressed their readiness to share their screening experiences to inspire others, viewing this as an act of love.
Know the Risks and Take Action
While the most prevalent symptom of colorectal cancer is the absence of symptoms, it’s crucial to recognize potential warning signs. These may include changes in bowel habits, rectal bleeding, abdominal discomfort, fatigue, or unexplained weight loss.
Screening is generally recommended to commence at age 45. However, it is wise to consult a healthcare provider about starting earlier if specific risk factors are present, such as a family history of the disease, certain genetic predispositions, or chronic inflammatory bowel disorders like Crohn’s disease.
Experts underscore that early detection significantly increases the curability of this type of cancer. Having open discussions with your doctor, taking symptoms seriously, and timely screening are the most effective approaches to save lives.
Learn more about colorectal screenings.
Tagged Under
CT Scan, gut bacteria, cholesterol, colorectal cancer, cancer screening
