Newswise — A new UCLA study suggests that a new machine learning technique known as “causal forests” is about five times more efficient than current clinical practice in treating hypertensive patients.
In current practice, physicians treat patients with high blood pressure under the assumption that those at highest risk of adverse outcomes or death from the disease will benefit most from lowering blood pressure. . However, recent advances have been made in how machine learning predicts the adverse health effects of hypertension based on individual characteristics, allowing clinicians to examine and adjust for how the effects of interventions vary between individuals. can. Treatment tailored to the patient’s needs.
Using a technique called the “forest of causes,” researchers found that people at the highest risk of cardiovascular disease, such as hypertension and hypertension, did not always benefit most from intensive pressure control. said lead author Kosuke Inoue, Ph.D. UCLA Fielding School of Public He embarked on this research while an epidemiology graduate student in Health and is currently an Associate Professor of Social Epidemiology at Kyoto University.
“We found that a significant number of people without high blood pressure would benefit from lowering their blood pressure,” he said. We found that this improved population health compared to traditional, high-risk approaches.”
The study is published in the peer-reviewed International Journal of Epidemiology.
Causal Forest is a machine learning algorithm that estimates the impact of interventions on outcomes based on individual characteristics, enabling personalized predictions of how each individual will benefit from a particular treatment. The current standard practice is to treat all people at high risk of adverse health effects, such as stroke in hypertensives, using options that do not necessarily benefit everyone. . Pressure-lowering medications, as you may have other underlying health problems that are more important risk factors for cardiovascular disease such as diabetes.
In the current study, researchers used data from approximately 10,700 people from the Systolic Blood Pressure Intervention Trial (SPRINT) and the Behaviors to Control Cardiovascular Risk in Diabetic Blood Pressure (ACCORD-BP) trial. They also used data on her 14,600 US adults from the 1999–2018 National Health and Nutrition Examination Survey (NHANES) to estimate the effects of these approaches.
SPRINT showed that intensive systolic blood pressure therapy aimed at keeping blood pressure below 120 mmHg was associated with reduced risk of cardiovascular disease events and all-cause mortality in people without diabetes. rice field. As a result, the diagnostic blood pressure threshold has been lowered from 140 mmHg to 130 mmHg, resulting in approximately half of the US population being diagnosed with hypertension. The ACCORD-BP trial, which focused on people with diabetes, found no evidence that intensive blood pressure control was associated with a reduced probability of adverse cardiovascular events.
The researchers applied the causal forest method to data from participants in these two trials to predict what the effect of individualized treatment to reduce adverse cardiovascular outcomes would be after three years. bottom. We then compared the performance of the two approaches to each other. High-benefit approaches (targeting individuals with high treatment response) and high-risk approaches (targeting high-risk individuals, such as hypertension and high cardiovascular risk scores).
They found that about 80% of people with blood pressure readings of 130 mmHg or higher benefited from intensive blood pressure control. The higher profit approach outperformed the higher risk approach by nearly 8 percent. When transferred to NHANES data, the results were the same.
The study has some limitations, the researchers wrote. For example, we could not rule out individual characteristics that might influence treatment. Baseline characteristics were self-reported by participants, which may lead to measurement errors. It is also possible that the study design, rather than social and physical mechanisms, influenced the results, due to differences in testing criteria between SPRINT and ACCORD-BP.
Although more research is needed, the findings suggest that a high-benefit approach targeting individuals with putative health benefits from lowering blood pressure could significantly enhance treatment efficacy and improve population health outcomes. suggesting that it could be improved. It would change current therapeutic strategies in clinical practice and political decision-making, researchers note.
“Our findings shine a light on powerful machine learning algorithms to identify individuals who would benefit most from tight blood pressure control, a key enabler of precision medicine.” Dr. Tsugawa is an associate professor of general medicine and health services research at UCLA’s David Geffen College of Medicine and an associate professor of health policy and management at The Fielding School.
Susan Athey of Stanford University was an additional co-author.
This research was partially supported by the Japan Agency for Medical Research and Development (JP22rea522107). The study authors also received funding from the National Diabetes Institute and Digestive and Kidney Diseases (F99 DK126119). National Institute on Minority Health and Health Disparities (R01MD013913); National Institute on Aging (R01AG068633); Golub Capital Social Impact Lab. Office of Naval Research (N00014-17–1-2131); Mercatas Center; Microsoft Research; Japan Society for the Promotion of Science (21K20900 and 22K17392); For non-research-related tasks, the next-generation global-minded leading scientists development program (L-INSIGHT) sponsored by Japan’s Ministry of Education, Culture, Sports, Science and Technology.
