Background. Timely identification of risk factors for arterial hypertension (AH) is essential for prevention, but in busy primary care settings history taking is often incomplete. Artificial intelligence (AI) tools may help collect a more complete history in less time.
Aim. To evaluate the effectiveness of an AI-based history-taking system for identifying AH and its risk factors in primary health care.
Methods. A total of 106 patients aged 20–50 years who attended a pre-physician consultation room for a preventive examination or for elevated blood pressure were randomly assigned to history taking by a physician using a standard protocol (n = 53) or by an AI-based adaptive questionnaire system (n = 53). The identification of five groups of risk factors, the completeness of recommendations, interview duration and patient-rated convenience were compared.
Results. The AI system identified each risk factor more often than the traditional interview; the differences were significant for smoking (83.0% vs 56.6%; p = 0.006) and for excessive salt intake or unbalanced diet (79.2% vs 54.7%; p = 0.013). Interview duration was 6.4 ± 1.1 minutes with the AI system and 12.1 ± 1.2 minutes with the traditional method (p < 0.001). Patients in the AI group more often received complete recommendations on blood pressure self-monitoring (77.4% vs 50.9%; p = 0.009), additional examinations (83.0% vs 62.3%; p = 0.029) and long-term follow-up (86.8% vs 60.4%; p = 0.004). Mean convenience scores were 4.5 and 3.8 on a five-point scale, respectively.
Conclusion. An AI-based history-taking system can improve the identification of AH risk factors, particularly behavioral ones, halve the time needed for history taking and provide a basis for individualized preventive recommendations in primary care.