Boston University has developed a new AI model to assist hypertension treatment, faster and more accurately than ever.
Image by Myriams-Fotos via Pixabay
The US Centers for Disease Control and Prevention reveal that high blood pressure (hypertension) presents a potential death sentence for nearly half of all Americans, causing close to 700,000 mortalities in 2021. It also greatly increases the risk of stroke and chronic heart failure.
Background
Blood pressure measures the force exerted on arteries when the heart pumps blood. This metric increases and decreases throughout one’s day due to natural environmental factors. For example, stress and exercise could increase blood pressure to get blood to reach the body’s cells quicker. Certain medications and drugs could also trigger this response and cause blood to flow at a higher-than-normal pressure. This fluctuation normalizes between a certain range, but age and other diseases can cause abnormal hypertension, which becomes detrimental to health.
The buildup of blood pressure’s effects can and will lead to heart failure. Prevention strategies including proper exercise and diet can avert these consequences. However, many hypertension cases today are primarily due to genetic causes and cannot be prevented through lifestyle modifications.
Unfortunately, due to limited resources and physicians, treatment is often time-consuming and repetitive for hypertension patients and varies based on their stages of disease.
Solution
However, Boston University scientists have recently developed an AI program to tackle this issue. Aiming to help doctors prescribe effective hypertension medications to patients in real-time, the program considers multiple factors, including demographics, vital signs, patient medical and family histories, and clinical test records to generate a list of suggested medications and their associated likelihood of success.
The model even considers the efficacy of the medication for patients with similar cases, improving clinical workflow while providing quick access to trends related to hypertension.
Using anonymous data from 42,752 hypertensive patients at the Boston Medical Center (BMC), the model was trained and developed to sort patients into relative groups based on data similarities. An evaluation revealed that the program achieved a 70.3% cumulative larger reduction in high systolic blood pressure, leading to the algorithm’s clinical validation.
Despite exceptional results, the widespread integration of the program in clinics is limited by the general physician’s trust in AI. Due to AI’s weak explainability when arriving at conclusions, physicians are often reluctant to welcome AI and the changes that result. This predicament known as the “black box” problem, encompasses the issue of supporting an AI algorithm’s decisions. Humans have years of in-depth training in proper treatment and medical ethics, which artificial intelligence lacks. This mistrust makes the adoption of AI algorithms such as BU’s hypertension model quite difficult in clinical settings.
Furthermore, because of the risk of data confiscation, collecting large amounts of confidential data for the development of algorithms becomes difficult.
Therefore, despite this innovative artificial intelligence algorithm’s current success and potential, the awareness of such technology remains inadvertently low, possibly preventing better outcomes in numerous cases worldwide. The opportunity to serve additional patients more effectively might now be here, but ultimately the choice is ours whether we want to take it.
AI is a tool and reality that must be accepted to accelerate the desired quality of life. Although the potential to improve many fields has been emphasized in the past decade, its role in healthcare remains insignificant due to various complications. Physicians and clinics should accept AI with open arms to achieve better patient outcomes and hasten the clinical workflow while practicing proper regulation to ensure safety.
References:
High Blood Pressure Facts. (2024b, May 15). High Blood Pressure. https://www.cdc.gov/high-blood-pressure/data-research/facts-stats/?CDC_AAref_Val=https://www.cdc.gov/bloodpressure/facts.htm
Understanding Blood Pressure Readings. (2024, May 24). www.heart.org. https://www.heart.org/en/health-topics/high-blood-pressure/understanding-blood-pressure-readings
Stanton, M. (2023, November 6). New Artificial Intelligence Program Could Help Treat Hypertension. Boston University. https://www.bu.edu/articles/2023/new-artificial-intelligence-program-could-help-treat-hypertension/
ScaDS_PubRel. (2023, July 12). Cracking the Code: The Black Box Problem of AI. ScaDS.AI. https://scads.ai/cracking-the-code-the-black-box-problem-of-ai/#:~:text=The%20black%20box%20problem%20refers,This%20poses%20a%20significant%20challenge.