BABY STEPS IN ARTIFICIAL INTELLIGENCE: DEVELOPMENT OF A JOS CARDIOVASCULAR DISEASE RISK APP TO IMPROVE SCREENING FOR CARDIOVASCULAR DISEASES.

Abstract

Introduction/background: Assessing cardiovascular disease (CVD) risk is necessary in preventive cardiology. Studies have imputed CVD risk factors in algorithms to predict ASCVD. These various scores were derived from risk equations acquired from other populations. In our research, we found that abdominal height measured with our locally conceptualized appliance the Abdominometer predicted ASCVD better than established anthropometric indices.

Objectives: We, therefore, decided to build it into a risk equation and come up with a new algorithm that will not require data generated from invasive procedures.

Methods: We secondarily analysed our data and generated an algorithm utilizing 10 risk factors: one of which was our new anthropometric index of abdominal height (AH). Using the CIMT as a standard with a cut of value of ≥0.078 cm for high atherosclerotic risk we compared our new tool with the Framingham Risk Score (FRS).

Results: With our new algorithm, 24/221 (10.9%) were at high risk with 109 and 88 at low and intermediate risks respectively. Using the FRS, 218/221 were at low risk; only 3 being in the intermediate and high risk. Both risk algorithms correlated significantly with CIMT-determined risk but the correlation coefficient was more for the new (0.448) than the FRS (0.300).

Conclusions: We found that with sub-clinical atherosclerosis indexed by carotid intima-media thickness as standard, our new Jos App as well as the Framingham Risk score correlated positively and significantly. However, interestingly the level of correlation was higher with our new risk estimation App. We have input this into smart devices for pilot clinical studies.