Survey and Evaluation of Hypertension Machine Learning Research

Area of Science:

• Cardiology
• Medical Informatics
• Artificial Intelligence

Background:

• Machine learning (ML) applications are expanding across research fields, with increasing use in hypertension studies.
• However, the breadth and depth of ML applications in specialized areas like hypertension remain limited.
• This study addresses the growing body of ML research in hypertension, evaluating its quality and identifying barriers to clinical adoption.

Purpose of the Study:

• To survey machine learning (ML) research in hypertension.
• To critically evaluate the reporting quality of these studies.
• To identify barriers hindering the integration of ML into hypertension care.

Main Methods:

• A systematic review of 63 hypertension-related ML research articles published between January 2019 and September 2021.
• Utilized the Harmonious Understanding of Machine Learning Analytics Network (HUMAN) survey questionnaire.
• Analyzed research topics, reporting standards, model validation, and ethical considerations.

Main Results:

• Blood pressure prediction was the most common topic (38%), followed by general hypertension research (22%).
• Reporting quality was variable: only 46% detailed study populations, 40% reported calibration measures, and 48% mentioned ethical compliance.
• Crucially, only 14% used distinct validation datasets, and algorithmic bias was unaddressed in all studies, with only 6 acknowledging its risk.

Conclusions:

• Current ML research in hypertension is largely exploratory and suffers from significant reporting deficiencies.
• Inadequate model validation and a lack of attention to algorithmic bias are major shortcomings.
• Addressing these areas is crucial for realizing ML's potential to transform hypertension management and care.