Washington : According to new investigation led by researchers from Harvard Medical School and the University of Copenhagen, in collaboration with the VA Boston Healthcare System, Dana-Farber Cancer Instituteand the Harvard T.H. Chan School of Public Healtha artificial intelligence tool has successfully identified people most at risk of pancreatic cancer up to three years before diagnosis using only the medical records of the patients.
The findings, published May 8 in Nature Medicine, suggest that AI-based population assessment could be valuable in finding people at elevated risk of disease and could speed up diagnosis of a condition that is all too often in the advanced stages when treatment is less effective and results are grim, the researchers said. Pancreatic cancer It is one of the world’s deadliest cancers and its toll is expected to increase.
Currently, there are no population-based tools to broadly detect pancreatic cancer. Those with a family history and certain genetic mutations that predispose them to pancreatic cancer are specifically evaluated. But such targeted assessments may miss other cases that fall outside of those categories, the researchers said.
“One of the biggest decisions doctors face every day is who is at high risk of disease and who would benefit from further testing, which can also mean more invasive and costly procedures that carry their own risks,” said the study co-author. -Senior Research Fellow Chris Sander, faculty member in the Department of Systems Biology at the Blavatnik Institute at HMS. “An artificial intelligence tool that can zero in on people at higher risk of pancreatic cancer who will benefit most from more testing could go a long way toward improving clinical decision making.”
Applied at scale, Sander added, this approach could speed detection of pancreatic cancer, lead to earlier treatment, and improve outcomes and prolong the lives of patients.
“Many types of cancer, especially those that are difficult to identify and treat early, have a disproportionate effect on patients, families and the broader health care system,” said study co-author Soren Brunak, a professor of biology. of disease systems and director of research. at the Novo Nordisk Foundation Center for Protein Research at the University of Copenhagen. “AI-based detection is an opportunity to alter the trajectory of pancreatic cancer, an aggressive disease that is notoriously difficult to diagnose early and treat promptly when the chances of success are highest.”
In the new study, the AI algorithm was trained on two separate data sets totaling 9 million patient records from Denmark and the United States. The researchers “asked” the AI model to look for telltale signs based on the data contained in the logs. Based on combinations of disease codes and time of onset, the model was able to predict which patients were likely to develop pancreatic cancer in the future. In particular, many of the symptoms and disease codes were not directly related to or derived from the pancreas.
The researchers tested different versions of the AI models to determine their ability to detect people at elevated risk of developing disease at different time scales, 6 months, one year, two years, and three years. Overall, each version of the AI algorithm was substantially more accurate in predicting who would develop pancreatic cancer than current estimates of the incidence of the disease in the entire population, defined as the frequency with which a condition develops in a population over a period of time. a specific period of time. The researchers said they believe the model is at least as accurate in predicting disease onset as current genetic sequencing tests, which are typically available for only a small subset of patients in data sets.
The “angry organ”
Screening for certain common cancers, such as breast, cervical, and prostate cancers, relies on relatively simple and highly effective techniques, a mammogram, Pap smear, and blood test, respectively. These detection methods have transformed the outcomes of these diseases by ensuring early detection and intervention during the most treatable stages.
By comparison, pancreatic cancer is more difficult and expensive to detect and evaluate. Doctors primarily look at family history and the presence of genetic mutations which, while important indicators of future risk, are often missed by many patients. One particular advantage of the AI tool is that it could be used on any and all patients for whom health records and medical history are available, not just those with a known family history or genetic predisposition to the disease. This is especially important, the researchers add, because many high-risk patients may not even be aware of their genetic predisposition or family history.
In the absence of symptoms and no clear indication that someone is at high risk for pancreatic cancer, doctors are understandably wary of recommending more sophisticated and expensive tests, such as CT scans, MRIs, or endoscopic ultrasound. When these tests are used and suspicious lesions are discovered, the patient must undergo a procedure to obtain a biopsy. Located deep in the abdomen, the organ is difficult to access and easy to provoke and inflame. His irritability has earned him the nickname “the angry organ.”
An artificial intelligence tool that identifies people most at risk for pancreatic cancer would ensure that doctors are testing the right population, while sparing others unnecessary tests and additional procedures, the researchers said.
About 44 percent of people diagnosed in the early stages of pancreatic cancer survive five years after diagnosis, but only 12 percent of cases are diagnosed that early. The survival rate drops from 2 to 9 percent in those whose tumors have grown beyond their site of origin, the researchers estimate.
“This low survival rate occurs despite marked advances in surgical techniques, chemotherapy and immunotherapy,” Sander said. “So, in addition to sophisticated treatments, there is a clear need for better detection, more targeted testing, and earlier diagnosis, and this is where the AI-based approach presents itself as the critical first step on this continuum.”
Previous diagnoses portend future risk
For the current study, the researchers designed several versions of the AI model and trained them on the health records of 6.2 million patients in Denmark’s national health system over 41 years. Of those patients, 23,985 developed pancreatic cancer over time. During training, the algorithm discerned patterns indicative of future pancreatic cancer risk based on disease trajectories—that is, whether the patient had certain conditions that occurred in a certain sequence over time.
For example, diagnoses such as gallstones, anemia, type 2 diabetes, and other problems related to the gastrointestinal system portended an increased risk of pancreatic cancer within 3 years of evaluation. Less surprisingly, inflammation of the pancreas was a strong predictor of future pancreatic cancer in an even shorter time frame of two years. The researchers caution that none of these diagnoses by themselves should be considered indicative or causative of future pancreatic cancer. However, the pattern and sequence in which they occur over time offer clues to an AI-based surveillance model and could prompt clinicians to monitor people at high risk more closely or test accordingly. .
Next, the researchers tested the best-performing algorithm on a completely new set of patient records it hadn’t encountered before—a US Veterans Health Administration dataset of nearly 3 million records spanning 21 years and containing 3864 people diagnosed with pancreatic cancer. The predictive accuracy of the tool was somewhat lower on the US dataset.
This is most likely because the US dataset was collected for a shorter time and contained somewhat different patient population profiles, the entire population of Denmark in the Danish dataset vs. current and former military in the Veterans Affairs data set. When the algorithm was retrained from scratch on the US dataset, its predictive accuracy improved. This, the researchers said, underscores two important points: first, ensuring that AI models are trained on high-quality, rich data.
Second, the need to access large representative data sets of aggregated clinical registries at the national and international level. In the absence of such globally valid models, AI models must be trained on local health data to ensure that their training reflects the idiosyncrasies of local populations.
