AI found that nearly half of IBD patients died prematurely

Machine learning algorithms reveal the disturbing reality of Canadians with inflammatory bowel disease (IBD): Almost half of them die before their time.

In a new study published Monday in the Journal of the Canadian Medical Association, researchers found that 47% of Ontarians with IBD died before the age of 75 (75) and had certain early life health conditions, greatly increasing the risk.

Between 2010 and 2020, a research team led by scientists from Sickkids Hospital and the University of Toronto analyzed data from 9,200 IBD patients in Ontario. Their goal: to determine whether AI can determine which other combinations of health problems most powerfully predict early death.

“The clinical implication is that, while further causal studies are needed to elucidate this relationship, early chronic diseases may be more important in determining the patient’s health trajectory,” wrote Dr. Eric Benchimol, a pediatric gastroenterologist and senior scientist at the hospital.

This discovery is a critical moment in Canada, one of the highest speeds in the world of IBD. By 2035, approximately 470,000 Canadians (1 in 91) are expected to have Crohn’s disease or ulcerative colitis, two main forms of IBD.

Conditions for early onset mark greater risks

The study shows that developing certain conditions before the age of 60 specifically predicts premature death. The AI ​​model also performs well when analyzing whether patients have other health problems and how young they are when these conditions first occur.

Key risk factors include early emotional disorders, arthritis, other mental health conditions, and hypertension. The risk of IBD patients is higher than that of women, while 50% of men are 44% compared with women.

By the end of life, most IBD patients are managing multiple chronic diseases. The most common are various types of arthritis (77%), hypertension (73%), mood disorders (69%), renal failure (50%) and cancer (46%).

“While our insights are not causal insights, they can identify patients at a higher risk of premature death, so who may benefit from more coordinated care for IBD and other chronic diseases,” Benchimol noted.

AI performs better than traditional methods

Using sophisticated machine learning algorithms, the research team achieved significant accuracy in predicting which patients died prematurely. The best model shows that the area under the receiver operating curve (AUC) is 0.95, a nearly perfect discrimination reflecting between people who die before age 75 and before age 75.

This performance outperforms traditional statistical methods, illustrating the ability of AI to detect complex patterns that may be hidden in large data sets.

Gemma Postill, a medical student at the University of Toronto’s Metty University School of Medicine, co-leading author of the study, and Dr. Laura Rosella, has developed several predictive models to explore different aspects of the relationship between chronic disease and early death in patients with IBD.

The authors stressed in publication: “Because the results more directly identify opportunities for health system improvement, premature death is used because premature death can be avoided through appropriate prevention or early effective treatment.”

These findings highlight key gaps in current care models for patients with IBD who often face broken health care that cannot meet the full range of their health needs.

Despite established multidisciplinary IBD care goals, structural barriers, including funding restrictions and organizational challenges, often prevent implementation. The study shows that coordinated care that addresses mental health, arthritis, and cardiovascular diseases with IBD management may prolong life.

“These findings provide scientific support for the provision of multidisciplinary and integrated health care throughout the life cycle, especially during adulthood,” the author concluded.

For nearly half a million Canadians who are expected to live with IBD over the next decade, the study has both a warning and a roadmap: Early identification and coordinated treatment of other health conditions may be key to avoiding premature death.

While the study did not determine that these diseases directly lead to early death in IBD patients, it provides compelling evidence that they are important warning signals – signals that healthcare systems can potentially be used to identify high-risk patients and to intervene before it is too late.