Although it is a non-specific procedure that does not identify specific health conditions, it could lead to new biomarkers that clinical laboratories could use for predictive healthcare
Researchers from the Mayo Clinic recently used artificial intelligence (AI) to develop a predictive computational tool that analyzes an individual’s gut microbiome to identify how a person may experience improvement or deterioration in health.
Dubbed the Gut Microbiome Wellness Index 2 (GMWI2), Mayo’s new tool does not identify the presence of specific health conditions but can detect even minor changes in overall gut health.
Built on an earlier prototype, GMWI2 “demonstrated at least 80% accuracy in differentiating healthy individuals from those with any disease,” according to a Mayo news release. “The researchers used bioinformatics and machine learning methods to analyze gut microbiome profiles in stool samples gathered from 54 published studies spanning 26 countries and six continents. This approach produced a diverse and comprehensive dataset.”
“Our tool is not intended to diagnose specific diseases but rather to serve as a proactive health indicator,” said senior study author Jaeyun Sung, PhD (above), a computational biologist at the Mayo Clinic Center for Individualized Medicine: Microbiomics Program in the news release ease. “By identifying adverse changes in gut health before serious symptoms arise, the tool could potentially inform dietary or lifestyle modifications to prevent mild issues from escalating into more severe health conditions, or prompt further diagnostic testing.” For microbiologists and clinical laboratory managers, this area of new knowledge about the human microbiome may lead to multiplex diagnostic assays. (Photo copyright: Mayo Clinic.)
Connecting Specific Diseases with Gut Microbiome
Gut bacteria that resides in the gastrointestinal tract consists of trillions of microbes that help regulate various bodily functions and may provide insights regarding the overall health of an individual. An imbalance in the gut microbiome is associated with an assortment of illnesses and chronic diseases, including cardiovascular issues, digestive problems, and some cancers and autoimmune diseases.
To develop GMWI2, the Mayo scientists provided the machine-learning algorithm with data on microbes found in stool samples from approximately 8,000 people collected from 54 published studies. They looked for the presence of 11 diseases, including colorectal cancer and inflammatory bowel disease (IBS). About 5,500 of the subjects had been previously diagnosed with one of the 11 diseases, and the remaining people did not have a diagnosis of the conditions.
The scientists then tested the efficacy of GMWI2 on an additional 1,140 stool samples from individuals who were diagnosed with conditions such as pancreatic cancer and Parkinson’s disease, compared with those who did not have those illnesses.
The algorithm gives subjects a score between -6 and +6. People with a higher GMWI2 score have a healthier microbiome that more closely resembles individuals who do not have certain diseases.
Likewise, a low GMWI2 score suggests the individual has a gut microbiome that is similar to those who have specific illnesses.
Highly Accurate Results
According to their study, the researchers determined that “GMWI2 achieves a cross-validation balanced accuracy of 80% in distinguishing healthy (no disease) from non-healthy (diseased) individuals and surpasses 90% accuracy for samples with higher confidence,” they wrote in Nature Communications.
Launched in 2020, the original GMWI (Gut Microbiome Wellness Index) was trained on a much smaller number of samples but still showed similar results.
The researchers tested the enhanced GMWI2 algorithm across various clinical schemes to determine if the results were similar. These scenarios included individuals who had previous fecal microbiota transplants and people who had made dietary changes or who had exposure to antibiotics. They found that their improved tool detected changes in gut health in those scenarios as well.
“By being able to answer whether a person’s gut is healthy or trending toward a diseased state, we ultimately aim to empower individuals to take proactive steps in managing their own health,” Sung said in the news release.
The Mayo Clinic team is developing the next version of their tool, which will be known as the Gut Microbiome Wellness Index 3. They plan to train it on at least 12,000 stool samples and use more sophisticated algorithms to decipher the data.
More research and studies are needed to determine the overall usefulness of Mayo’s Gut Microbiome Wellness Index and its marketability. Here is a world-class health institution disclosing a pathway/tool that analyzes the human microbiome to identify how an individual may be experiencing either an improvement in health or a deterioration in health.
The developers believe it will eventually help physicians determine how patients’ conditions are improving or worsening by comparing the patients’ microbiomes to the profiles of other healthy and unhealthy microbiomes. As this happens, it would create a new opportunity for clinical laboratories to perform the studies on the microbiomes of patients being assayed in this way by their physicians.
Use of artificial intelligence in clinical laboratory testing could improve the diagnosis of cancer worldwide
In a proof of concept study, scientists at Shanghai Jiao Tong University in China have developed a clinical laboratory test that utilizes artificial intelligence (AI) to diagnose three types of cancer from a single drop of dried blood. The paper-based test was able to identify patients with colorectal, gastric, and pancreatic cancers and distinguish between patients with and without cancer.
The team’s goal was to develop a way to diagnose cancer while the disease is still in the earlier stages, especially in rural areas.
“Over a billion people across the world experience a high rate of missed disease diagnosis, an issue that highlights the need for diagnostic tools showing increased accuracy and affordability. In addition, such tools could be used in ecologically fragile and energy-limited regions, pointing to the need for developing solutions that can maximize health gains under limited resources for enhanced sustainability,” the researchers wrote in an article published in the journal Nature Sustainability titled, “A Sustainable Approach to Universal Metabolic Cancer Diagnosis.”
The researchers determined that by using less than 0.05 millimeters of dried blood, their test could accurately and quickly identify if a patient had cancer between 82% to 100% of the time.
According to Chaoyuan Kuang, MD, PhD (above), an oncologist at Montefiore Health System and assistant professor at the Albert Einstein College of Medicine, unlike liquid blood, dried serum can be “collected, stored, and transported at much lower cost and with much simpler equipment,” Live Science reported. “This could help democratize the availability of cancer early detection testing across the world,” he added. A paper-based clinical laboratory test that can detect and distinguish one cancer type from another would be a boon to cancer diagnosis worldwide. (Photo copyright: Albert Einstein College of Medicine.)
Improving Cancer Screening in Rural Areas
An earlier study conducted in China in 2022 examined results from 1,570 cancer survivors from both urban and rural areas of China. That study showed that 84.1% of the patients were diagnosed with cancer only after developing symptoms and that urban patients were more likely to be diagnosed in the early stages of cancer. In addition, rural patients also had less screening and treatment options available to them.
The researchers in this latest Chinese study tested their AI model on blood donors with and without cancer and compared the results to traditional liquid-blood biopsy tests.
“Based on modeling they performed, they reported the new tool could reduce the estimated proportion of undiagnosed cases of pancreatic, gastric, and colorectal cancers by about 20% to 50% if it was used for population-level cancer screening in rural China,” Live Science reported.
The scientists used dried serum spots (DSS) and machine learning to perform the research. According to their Nature Sustainability paper, DSS can be challenging in cancer research because sensitive biomarkers in the samples are often degraded or have inadequate amount of blood for proper analysis. To circumvent these issues, the researchers used nanoparticle-enhanced laser desorption/ionization mass spectrometry (NPELDI MS) to increase reliability and sensitivity. Inorganic nanoparticles were applied to the samples to strengthen selectivity and refine metabolic compounds from the samples.
However, the study authors noted that “the adaptation of NPELDI MS to dried spot analysis has not been validated,” Interesting Engineering reported.
A ‘Great Start’
The machine learning algorithm the Chinese scientists created demonstrates that DSS samples can be used to preserve important biological markers and could be beneficial in the diagnosis of cancer.
Their research indicated an overall reduction rate of undiagnosed cancers in the range of 20.35% to 55.10%. The researchers estimated the implementation of their AI tool could reduce the proportion of specific undiagnosed cancer cases in rural China by:
84.30% to 29.20% for colorectal cancer,
77.57% to 57.22% for gastric cancer, and
34.56% to 9.30% for pancreatic cancer.
It’s a “great start,” Chaoyuan Kuang, MD, PhD, an oncologist at Montefiore Health System and assistant professor at the Albert Einstein College of Medicine told Live Science. “This cancer test won’t enter use for a long time,” he said. Nevertheless, the potential of the tool is “immense,” he added, but that “we are still years away from being able to offer this test to patients.
“With further development, it could theoretically be used for the early detection of other types of cancer or for other diseases, or to monitor the progression of disease in patients who have already been diagnosed,” he noted.
Further research and clinical trials are needed before this AI tool can be used in a clinical diagnostic setting. This study is another example of researchers looking for cancer biomarkers in specimen types that are not tissue and further supports the hope that machine learning may one day detect cancer in earlier stages, increase survival rates, and save healthcare costs.
One factor motivating this type of research in China is the fact that the nation has more than 36,000 hospitals and approximately 20,000 anatomic pathologists. Of this total, only a minority of these pathologists have been trained to the standards of North America and Northern Europe.
Like other nations, China’s demand for subspecialist pathology services outstrips its supply of such pathologists. This is the reason why researchers in that country want to develop diagnostic assays for cancer and other diseases that are faster, cheaper, and comparable to a human pathologist in accuracy.
Infection control teams and clinical laboratory managers may want to look at this new product designed to improve the diagnosis and treatment of sepsis
Accurate and fast diagnosis of sepsis for patients arriving in emergency departments is the goal of a new product that was just cleared by the federal Food and Drug Administration (FDA). It is also the newest example of how artificial intelligence (AI) continues to find its way into pathology and clinical laboratory medicine.
Sepsis is one of the deadliest killers in US hospitals. That is why there is interest in the recent action by the FDA to grant marketing authorization for an AI-powered sepsis detection software through the agency’s De Novo Classification Request. The DNCR “provides a marketing pathway to classify novel medical devices for which general controls alone, or general and special controls, provide reasonable assurance of safety and effectiveness for the intended use, but for which there is no legally marketed predicate device,” the FDA’s website states.
Unlike a single analyte assay that is run in a clinical laboratory, Prenosis’ AI/ML software uses 22 diagnostic and predictive parameters, along with ML algorithms, to analyze data and produce a clinically actionable answer on sepsis.
It is important for clinical laboratory managers and pathologists to recognize that this diagnostic approach to sepsis brings together a number of data points commonly found in a patient’s electronic health record (EHR), some of which the lab generated and others the lab did not generate.
“Sepsis is a serious and sometimes deadly complication. Technologies developed to help prevent this condition have the potential to provide a significant benefit to patients,” said Jeff Shuren, MD, JD, Director of the FDA’s Center for Devices and Radiological Health, in a statement. “The FDA’s authorization of the Prenosis Sepsis ImmunoScore software establishes specific premarket and post-market requirements for this device type.” Clinical laboratory EHRs contain some of the data points Prenosis’ diagnostic software uses. (Photo copyright: US Food and Drug Administration.)
How it Works
To assist doctors diagnose sepsis, the ImmunoScore software is first integrated into the patient’s hospital EHR. From there, it leverages 22 parameters including:
White blood cell count to produce a score that informs caregivers of the patient’s risk for sepsis within 24 hours, MedTech Dive reported.
Instead of requiring a doctor or nurse to look at each parameter separately, the SaMD tool uses AI “to evaluate all those markers at once”, CNBC noted. It then produces a risk score and four discrete risk stratification categories (low, medium, high, and very high) which correlate to “a patient’s risk of deterioration” represented by:
By sharing these details—a number from one to 100 for each of the 22 diagnostic and predictive parameters—Sepsis ImmunoScore helps doctors determine which will likely contribute most to the patient’s risk for developing sepsis, MedTech Dive reported.
“A lot of clinicians don’t trust AI products for multiple reasons. We are trying very hard to counter that skepticism by making a tool that was validated by the FDA first, and then the second piece is we’re not trying to replace the clinician,” Bobby Reddy Jr., PhD, Prenosis co-founder and CEO, told MedTech Dive.
Big Biobank and Blood Sample Data
Prenosis, which says its goal is the “enabling [of] precision medicine in acute care” developed Sepsis ImmunoScore using the company’s own biobank and a dataset of more than 100,000 blood samples from more than 25,000 patients.
AI algorithms drew on this biological/clinical dataset—the largest in the world for acute care patients suspected of having serious infections, according to Prenosis—to “elucidate patterns in rapid immune response.”
“It does not work without data, and the data started at Carle,” said critical care specialist Karen White, MD, PhD, Carle Foundation Hospital, St. Louis, MO, in the news release. “The project involved a large number of physicians, research staff, and internal medicine residents at Carle who helped recruit patients, collect data, and samples,” she said.
Opportunity for Clinical Laboratories
Sepsis is a life-threatening condition based on an “extreme response to an infection” that affects nearly 1.7 million adults in the US each year and is responsible for 350,000 deaths, according to US Centers for Disease Control and Prevention (CDC) data.
A non-invasive diagnostic tool like Sepsis ImmunoScore will be a boon to emergency physicians and the patients they treat. Now that the FDA has authorized the SaMD diagnostic tool to go to market, it may not be long before physicians can use the information it produces to save lives.
Clinical laboratory managers inspired by the development of Sepsis ImmunoScore may want to look for similar ways they can take certain lab test results and combine them with other data in an EHR to create intelligence that physicians can use to better treat their patients. The way forward in laboratory medicine will be combining lab test results with other relevant sets of data to create clinically actionable intelligence for physicians, patients, and payers.
Findings could lead to new biomarkers clinical laboratories would use for identifying cancer in patients and monitoring treatments
As DNA “dark matter” (the DNA sequences between genes) continues to be studied, researchers are learning that so-called “junk DNA” (non-functional DNA) may influence multiple health conditions and diseases including cancer. This will be of interest to pathologists and clinical laboratories engaged in cancer diagnosis and may lead to new non-invasive liquid biopsy methods for identifying cancer in blood draws.
This technique could enable non-invasive monitoring of cancer treatment and cancer diagnosis, Technology Networks noted.
“Our study shows that ARTEMIS can reveal genomewide repeat landscapes that reflect dramatic underlying changes in human cancers,” said study co-leader Akshaya Annapragada (above), an MD/PhD student at the Johns Hopkins University School of Medicine, in a news release. “By illuminating the so-called ‘dark genome,’ the work offers unique insights into the cancer genome and provides a proof-of-concept for the utility of genomewide repeat landscapes as tissue and blood-based biomarkers for cancer detection, characterization, and monitoring.” Clinical laboratories may soon have new biomarkers for the detection of cancer. (Photo copyright: Johns Hopkins University.)
Detecting Early Lung, Liver Cancer
Artemis is a Greek word meaning “hunting goddess.” For the Johns Hopkins researchers, ARTEMIS also describes a technique “to analyze junk DNA found in tumors” and which float in the bloodstream, Financial Times explained.
“It’s like a grand unveiling of what’s behind the curtain,” said geneticist Victor Velculescu, MD, PhD, Professor of Oncology and co-director of the Cancer Genetics and Epigenetics Program at Johns Hopkins Kimmel Cancer Center, in the news release.
“Until ARTEMIS, this dark matter of the genome was essentially ignored, but now we’re seeing that these repeats are not occurring randomly,” he added. “They end up being clustered around genes that are altered in cancer in a variety of different ways, providing the first glimpse that these sequences may be key to tumor development.”
ARTEMIS could “lead to new therapies, new diagnostics, and new screening approaches for cancer,” Velculescu noted.
Repeats of DNA Sequences Tough to Study
For some time technical limitations have hindered analysis of repetitive genomic sequences by scientists.
“Genetic changes in repetitive sequences are a hallmark of cancer and other diseases, but characterizing these has been challenging using standard sequencing approaches,” the study authors wrote in their Science Translational Medicine paper.
“We developed a de novok-mer (short sequences of DNA)-finding approach called ARTEMIS to identify repeat elements from whole-genome sequencing,” the researchers wrote.
The scientists put ARTEMIS to the test in laboratory experiments.
The first analysis involved 1,280 types of repeating genetic elements “in both normal and tumor tissues from 525 cancer patients” who participated in the Pan-Cancer Analysis of Whole Genomes (PCAWG), according to Technology Networks, which noted these findings:
A median of 807 altered elements were found in each tumor.
About two-thirds (820) had not “previously been found altered in human cancer.”
Second, the researchers explored “genomewide repeat element changes that were predictive of cancer,” by using machine learning to give each sample an ARTEMIS score, according to the Johns Hopkins news release.
The scoring detected “525 PCAWG participants’ tumors from the healthy tissues with a high performance” overall Area Under the Curve (AUC) score of 0.96 (perfect score being 1.0) “across all cancer types analyzed,” the Johns Hopkins’ release states.
Liquid Biopsy Deployed
The scientists then used liquid biopsies to determine ARTEMIS’ ability to noninvasively diagnose cancer. Researchers used blood samples from:
ARTEMIS classified patients with lung cancer with an AUC of 0.82.
ARTEMIS detected people with liver cancer, as compared to others with cirrhosis or viral hepatitis, with a score of AUC 0.87.
Finally, the scientists used their “ARTEMIS blood test” to find the origin of tumors in patients with cancer. They reported their technique was 78% accurate in discovering tumor tissue sources among 12 tumor types.
“These analyses reveal widespread changes in repeat landscapes of human cancers and provide an approach for their detection and characterization that could benefit early detection and disease monitoring of patients with cancer,” the researchers wrote in Science Translational Medicine.
Large Clinical Trials Planned
Velculescu said more research is planned, including larger clinical trials.
“While still at an early stage, this research demonstrates how some cancers could be diagnosed earlier by detecting tumor-specific changes in cells collected from blood samples,” Hattie Brooks, PhD, Research Information Manager, Cancer Research UK (CRUK), told Financial Times.
Should ARTEMIS prove to be a viable, non-invasive blood test for cancer, it could provide pathologists and clinical laboratories with new biomarkers and the opportunity to work with oncologists to promptly diagnosis cancer and monitor patients’ response to treatment.
This AI platform has the potential to also reduce workload of radiologists, but also of anatomic pathologists and oncologists allowing them to be more productive
When the UK’s National Health Service (NHS) recently tested an artificial intelligence (AI) platform’s ability to analyze mammograms, the AI found early signs of breast cancer that “human doctors” had previously missed, the BBC reported. This level of ability by AI might soon be adapted to aid overworked anatomic pathologists and cancer doctors in the United Kingdom.
Out of 10,000 mammograms MIA analyzed, the AI platform found “tiny signs of breast cancer in 11 women” which had not been spotted during earlier examinations, the BBC noted, adding that the cancers “were practically invisible to the human eye.”
This is a significant development in AI’s role in healthcare. Anatomic pathologists and clinical laboratory leaders will note that ongoing advancements in AI are enabling technology developers to apply their solutions to assessing radiology images, as well as in whole slide imaging used in digital pathology. In the UK, use of AI, the BBC noted, may also help ease doctor’s workloads.
“This is just the beginning of our work with Kheiron,” said Ben Glocker, PhD (above), Professor in Machine Learning for Imaging at Imperial College London and Head of ML Research at Kheiron Medical, in a news release. “We are actively working on new methodologies for the safe deployment and continuous monitoring of MIA to support a US and UK rollout. We are working hard to make sure that as many women as possible will benefit from the use of this new technology within the next year.” AI tools such as MIA may soon take much of the load from anatomic pathologists and radiologists. (Photo copyright: Imperial College London.)
MIA Cloud-based AI Platform
Kheiron was founded in 2016 and MIA was named one of the seven biggest medical breakthroughs in 2023 by ABC News. A study conducted by Imperial College London in 2023 found that MIA “could significantly increase the early detection of breast cancers in a European healthcare setting by up to 13%,” according to an Imperial news release.
“The study was conducted over three phases (two pilot phases and a live roll-out). Overall across the three phases, the AI reader found 24 more cancers than the standard human reading—a 7% relative increase—and resulted in 70 more women recalled (0.28% relative increase),” the news release reported. “Of the additional recalls, six (initial pilot), 13 (extended pilot), and 11 (live use) additional cancers were found, increasing relative cancer detection rate by 13%, 10%, and 5% respectively. [The researchers] found that 83% of the additional cancers detected using MIA in real clinical practice were invasive, showing that MIA can detect cancers where early detection is particularly vital.”
Supported by Microsoft’s Azure Cloud, MIA came together over six years based on training encompassing millions of mammograms worldwide, Healthcare Digital reported.
“AI tools are generally pretty good at spotting symptoms of a specific disease if they are trained on enough data to enable them to be identified. This means feeding the program with as many different anonymized images of those symptoms as possible, from as diverse a range of people as possible,” Sarah Kerruish, Chief Strategy Officer, Kheiron, told Healthcare Digital.
MIA has been trained to “recognize subtle patterns and anomalies” that can point to “cancerous cells even in their earliest stages of development,” Dataconomy reported.
MIA Finds Early Cancer Signs
In the pilot study, MIA examined mammograms from 10,889 women. Each image had previously been reviewed by two radiologists, the BBC reported.
Findings include the following according to Healthcare Digital:
MIA “flagged” all people the physicians previously identified with symptoms.
The AI platform discovered 11 people with cancer the doctors did not identify.
The cancer MIA discovered—and the doctors did not—suggested cancer in early stages.
So, how did the doctors miss the cancer that MIA spotted? Gerald Lip, MD, Clinical Director for Breast Screening in North East Scotland who led the pilot study for the NHS, told Healthcare Digital, “part of the power of AI is it’s not prone to exhaustion or distraction.
“There is an element of fatigue,” he said. “You get disruptions, someone’s coming in, someone’s chatting in the background. There are lots of things that can probably throw you off your regular routine as well. And in those days when you have been distracted, you go, ‘how on earth did I miss that?’ It does happen.”
Lip is also the Chief Investigator in the Mammography Artificial Intelligence Project in the Industrial Center for Artificial Intelligence and Digital Diagnostics in Scotland.
“I see MIA as a friend and an augmentation to my practice,” he told Healthcare Digital. “MIA isn’t perfect. It had no access to patient history so [it] would flag cysts that had already been identified by previous scans and designated harmless.”
AI as a Safety Net
In the 2023 study, researchers from Imperial College London deployed MIA as an extra reader for mammograms of 25,065 women who visited screening sites in Hungary between April 2021 and January 2023, according to a news release.
“Our prospective real-world usage data in Hungary provides evidence for a significant, measurable increase of early breast cancer detection when MIA is used in clinical practice,” said Peter Kecskemethy, PhD, CEO and co-founder of Kheiron Medical, in the news release.
“Our study shows that AI can act as an effective safety net—a tool to prevent subtler signs of cancer from falling through the cracks,” said Ben Glocker, PhD, Professor in Machine Learning for Imaging at Imperial College London and Head of ML Research at Kheiron Medical, in the news release.
More studies are needed before MIA can be used in clinical settings. Nevertheless, use of AI in radiology—specifically mammograms—where the AI tool can identify very small cancers typically undetectable by radiologists, would be a boon to cancer doctors and the patients they treat.
So far, the research suggests that the AI-powered MIA has benefits to deployment in breast cancer screening. Eventually, it may also make impressive contributions to medical diagnosis and patient care, particularly if MIA eventually proves to be effective at analyzing the whole slide images used by anatomic pathologists.
