Machine learning programs predict risk of death based on results from routine hospital tests

Computer analysis can help the health-care system “learn” by providing information vital to improving care, says researcher.

Padma Kaul with research team members Sunil Kalmady Vasu, Nariman Sepehrvand and Weijie Sun.

(From left) Padma Kaul with research team members Sunil Kalmady Vasu, Nariman Sepehrvand and Weijie Sun. The team developed a machine learning program that can accurately predict a patient’s risk of death within a month, a year and five years based on results from routine hospital tests. (Photo: Supplied)

If you’ve ever been admitted to hospital or visited an emergency department, you’ve likely had an electrocardiogram, or ECG, a standard test involving tiny electrodes taped to your chest that checks your heart’s rhythm and electrical activity. 

Hospital ECGs are usually read by a doctor or nurse at your bedside, but now researchers are using artificial intelligence to glean even more information from those results to improve your care and the health-care system all at once.

In recently published findings, the research team built and trained machine learning programs based on 1.6 million ECGs done on 244,077 patients in northern Alberta between 2007 and 2020. The algorithm predicted the risk of death from that point for each patient from all causes within one month, one year and five years with an 85 per cent accuracy rate, sorting patients into five categories from lowest to highest risk. The predictions were even more accurate when demographic information (age and sex) and six standard laboratory blood test results (creatinine, kidney function, sodium, troponin, hemoglobin and potassium) were included. 

The study is a proof-of-concept for using routinely collected data to improve individual care and allow the health-care system to “learn” as it goes, according to principal investigator Padma Kaul, professor of medicine and co-director of the Canadian VIGOUR Centre.

“We wanted to know whether we could use new artificial intelligence and machine learning methods to analyze the data and identify patients who are at higher risk for mortality,” Kaul explains. 

“These findings illustrate how machine learning models can be employed to convert data collected routinely in clinical practice to knowledge that can be used to augment decision-making at the point of care as part of a learning health-care system,” the researchers conclude in the study.

Toward a “learning health-care system”

A clinician will order an electrocardiogram if you have high blood pressure or symptoms of heart disease, such as chest pain, shortness of breath or an irregular heartbeat. The first phase of the study examined ECG results in all patients, but Kaul and her team hope to refine these models for particular subgroups of patients. They also plan to focus the predictions beyond all-cause mortality to look specifically at heart-related causes of death.

Kaul says Alberta is in a good position to analyze population-level data because so much is collected and archived within the publicly funded health-care system. 

“There is a big push to see how we can use AI to improve the delivery of health care,” says Kaul, who also holds the Heart & Stroke Chair in Cardiovascular Research and Canadian Institutes of Health Research Chair, Sex and Gender Differences in Diabetes, and is an adjunct professor with the School of Public Health. “For Albertans, this is really a demonstration of their data at work.”

The advantage of using high-powered computing is that, unlike humans, it can see the patterns in a multitude of data points at once, Kaul says. 

“We want to take data generated by the health-care system, convert it into knowledge and feed it back into the system so that we can improve care and outcomes. That’s the definition of a learning health-care system.”

The project was supported by the Canadian Institutes of Health Research and was a collaboration between the Canadian VIGOUR Centre in the Faculty of Medicine & Dentistry and the computing science department in the Faculty of Science. Kaul gives particular credit to co-authors Weijie Sun, a doctoral student in computing science, Sunil Vasu Kalmady, senior machine learning specialist, and Nariman Sepehrvand, research associate. Padma Kaul is a member of the Alberta Diabetes Institute, Cardiovascular Research Insitute, and Women and Children’s Health Research Institute.