An individually adjusted approach for communicating epidemiological results on health and lifestyle to patients

If scientific research on modifiable risk factors was more accessible to the general population there is a potential to prevent disease and promote health. Mobile applications can automatically combine individual characteristics and statistical models of health to present scientific information as individually tailored visuals, and thus there is untapped potential in incorporating scientific research into apps aimed at promoting healthier lifestyles. As a proof-of-concept, we develop a statistical model of the relationship between Self-rated-health (SRH) and lifestyle-related factors, and a simple app for conveying its effects through a visualisation that sets the individual as the frame of reference. Using data from the 6th (n = 12 981, 53.4% women and 46.6% men) and 7th (n = 21 083, 52.5% women and 47.5% men) iteration of the Tromsø population survey, we fitted a mixed effects linear regression model that models mean SRH as a function of self-reported intensity and frequency of physical activity (PA), BMI, mental health symptoms (HSCL-10), smoking, support from friends, and HbA1c ≥ 6.5%. We adjusted for socioeconomic and demographic factors and comorbidity. We designed a simple proof-of-concept app to register relevant user information, and use the SRH-model to translate the present status of the user into suggestions for lifestyle changes along with predicted health effects. SRH was strongly related to modifiable health factors. The strongest modifiable predictors of SRH were mental health symptoms and PA. The mean adjusted difference in SRH between those with 10-HSCL index = 1.85 (threshold for mental distress) and HSCL-10 = 1 was 0.59 (CI 0.61-0.57). Vigorous physical activity (exercising to exhaustion ≥ 4 days/week relative to sedentary) was associated with an increase on the SRH scale of 0.64 (CI 0.56-0.73). Physical activity intensity and frequency interacted positively, with large PA-volume (frequency ⨯ intensity) being particularly predictive of high SRH. Incorporating statistical models of health into lifestyle apps have great potential for effectively communicating complex health research to a general audience. Such an approach could improve lifestyle apps by helping to make the recommendations more scientifically rigorous and personalised, and offer a more comprehensive overview of lifestyle factors and their importance.

Algorithm for predicting valvular heart disease from heart sounds in an unselected cohort

Objective

This study aims to assess the ability of state-of-the-art machine learning algorithms to detect valvular heart disease (VHD) from digital heart sound recordings in a general population that includes asymptomatic cases and intermediate stages of disease progression.

Methods

We trained a recurrent neural network to predict murmurs from heart sound audio using annotated recordings collected with digital stethoscopes from four auscultation positions in 2,124 participants from the Tromsø7 study. The predicted murmurs were used to predict VHD as determined by echocardiography.

Results

The presence of aortic stenosis (AS) was detected with a sensitivity of 90.9%, a specificity of 94.5%, and an area under the curve (AUC) of 0.979 (CI: 0.963-0.995). At least moderate AS was detected with an AUC of 0.993 (CI: 0.989-0.997). Moderate or greater aortic and mitral regurgitation (AR and MR) were predicted with AUC values of 0.634 (CI: 0.565-703) and 0.549 (CI: 0.506-0.593), respectively, which increased to 0.766 and 0.677 when clinical variables were added as predictors. The AUC for predicting symptomatic cases was higher for AR and MR, 0.756 and 0.711, respectively. Screening jointly for symptomatic regurgitation or presence of stenosis resulted in an AUC of 0.86, with 97.7% of AS cases (n = 44) and all 12 MS cases detected.

Conclusions

The algorithm demonstrated excellent performance in detecting AS in a general cohort, surpassing observations from similar studies on selected cohorts. The detection of AR and MR based on HS audio was poor, but accuracy was considerably higher for symptomatic cases, and the inclusion of clinical variables improved the performance of the model significantly.