Scientists trained machine learning models to use electrocardiogram (ECG) results, demographics data, and six standard lab tests to estimate a patient’s risk of mortality. This new AI-based learning system could help healthcare providers make faster, more accurate medical decisions.
Predicting the risk of mortality could help healthcare providers streamline and prioritize the care and treatment of their patients. But current methods of diagnosing patients and determining treatment plans are limited to reading results from tests such as ECGs, X-rays, and bloodwork.
Now, researchers from Canada have trained machine learning programs to read and analyze data from those tests and estimate a patient's mortality risk.
Their findings, published in npj Digital Medicine, demonstrate that these artificial intelligence-based models can quickly provide valuable insight for doctors by identifying patients who are at high risk for short- or longer-term mortality — right at the patient’s bedside.
To investigate the mortality prediction performance of machine learning, the research team developed and trained machine learning programs using 1.6 million ECGs from 244,077 patients in Canada between 2007 and 2020.
The machine’s algorithm predicted mortality risk with an 85% accuracy for each patient at one month, one year, and five years. The models also categorized the patients into very low, low, medium, high, and very high risk groups.
However, when the team added demographic information like age and sex and six standard blood test results into the system — including hemoglobin, glomerular filtration rate (GFR), troponin I, creatinine, sodium, and potassium — mortality predictions were even more accurate.
In a news release, lead investigator Padma Kaul says, "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 healthcare system."
According to the study, the team’s future investigations will center around adding other labs to the machine learning models, such as AST, ALT, and HbA1c. They would also like to fine-tune the models for specific patient subgroups and focus on predicting heart-related mortality. In addition, the team plans to study the feasibility of the mortality risk assessment models in clinical practice.