Status
Conditions
About
Cardiovascular disease (CVD) is one of the prominent diseases that affect many people. One cost-effective solution is to identify people at higher risk of CVD by CVD risk prediction model. China-PAR, TRS-2P, and SMART2 are common risk prediction models for prevention. However, these risk scores were mostly based on the routinely self-check health information and multivariable regression without time-varying consideration. Investigators developed a Machine Learning (ML) based risk prediction model, Personalized CARdiovascular DIsease risk Assessment for Chinese (P-CARDIAC) among a predominantly Chinese population in Hong Kong to estimates the 10 years of secondary recurrent CVD risk for the high-risk individuals. The study objective is to evaluate the accuracy of the P-CARDIAC performance in practice among a large-scale Hong Kong population in medicine specialist outpatient clinic (SOPC) and cardiac clinic. The results will reassure cardiologists that the P-CARDIAC risk score is sensitive to the heart disease symptoms. Investigators anticipate that the results may help to facilitate P-CARDIAC in clinical setting and provide more practical information with the development of P-CARDIAC.
Full description
This study has two parts to conduct for different outcomes.
Part one is a prospective population-based cohort study. Investigators hypothesize that patients with a higher frequency or larger number of types of cardiac event symptoms are likely to have a higher P-CARDIAC risk score. The correlation between P-CARDIAC risk score and the increasing in number of types and in number of times a patient reported to have heart disease symptoms, e.g. chest pain, shortness of breath, fatigue, swelling in legs, ankles or feet will be examined. The results reassure cardiologists that the P-CARDIAC risk score is sensitive to the heart disease symptoms. Patients who are at a higher risk of recurrent CVD event are likely to have more medications and to have a sooner follow up appointment. At such, investigators also hypothesize a positive relationship between P-CARDIAC risk score and the number of class of medications prescribed and a negative correlation between P-CARDIAC risk score and the time to next SOPC follow up appointment.
Part two of the study will involve Delphi technique to determine P-CARDIAC risk thresholds with reference to current clinical management guideline such as American Heart Association (AHA)/American College of Cardiology (ACC) Multisociety Guideline (AHA/ACC) and European Society of Cardiology (ESC) and the European Atherosclerosis Society (EAS) (ESC/EAS), and this part will not involve patient contact. Investigators will compare the P-CARDIAC risk score and the clinician's rating of a patient's recurrent CVD risk on a 10% random subgroup of patients from part one in a "silent deployment" approach, ie, an individual patient's P-CARDIAC risk score will not be communicated to cardiologists. Cardiologists will be invited to rate if they think the patient is at high or low risk according to existing risk scores and the proposed treatment management and target based on treatment guidelines based on their experience. Each patient profile will be reviewed by at least 2 cardiologists and discussed to reach a consensus. Investigators will then evaluate the sensitivity of P-CARDIAC risk score with clinician's judgment. Prior to the commencement of Part two, investigators will collect insights from cardiologists regarding the interpretability of performance metrics and acceptable threshold of model performance for clinical practice. In addition, through this exercise, the research team will assist cardiologists to draft a guideline for follow up for patients at different risk level.
Enrollment
Sex
Ages
Volunteers
Inclusion criteria
Exclusion criteria
Loading...
Data sourced from clinicaltrials.gov
Clinical trials
Research sites
Resources
Legal