Harnessing technology and molecular analysis to understand the development of cardiovascular diseases in Asia: a prospective cohort study (SingHEART)

Stroke Demographics, Risk Factors, Subtypes, Syndromes, Mechanisms and Inter-Ethnic Differences between Chinese, Malays and Indians in Singapore-A Hospital-Based Study

Disparities in stroke may be due to socioeconomics, demographics, risk factors (RF) and ethnicity. Asian data are scant. This retrospective hospital-based study aimed to explore demographics, RF, stroke subtypes and mechanisms among the Chinese, Malays and Indians in Singapore. Stroke was subtyped into haemorrhagic stroke (HS) and ischaemic stroke (IS). For IS, the clinical syndrome was classified using the Oxfordshire Community Stroke Project (OCSP) classification while the stroke mechanism was categorised using the Trial of Org 10172 in Acute Stroke Treatment (TOAST) classification. During the study period 1 June 2015 to 31 December 2023, data were collected on 1165 patients, with a mean age of 65.6 ± 12.9 yr; 47.4% were female, 83% were Chinese and hypertension (63.5%) and hyperlipidaemia (60.3%) were the most common RF. HS comprised 23.5% (95%CI 21.1-26.1%) (intracerebral 21.7%, subarachnoid 1.3%) of the patients, while IS comprised 76.5% (95%CI 73.9-78.9%) (small artery occlusion 29.0%, cardioembolism 13.3%, large artery atherosclerosis 9.4%, stroke of other determined aetiology 6.2%, stroke of undetermined aetiology 18.6%); 55% of patients had lacunar syndrome. A multivariable analysis showed that HS was associated with ethnicity (p = 0.044), diabetes mellitus (OR 0.27, 95%CI 0.18-0.41, p < 0.001) and smoking (OR 0.47, 95%CI 0.34-0.64, p < 0.001). There were no significant inter-ethnic differences by the OCSP (p = 0.31) or TOAST (p = 0.103) classification. While differences in stroke subtype in Asia may be due to RF, ethnicity has a role. More studies are needed to further explore this.

Improving the predictive capability of Framingham Risk Score for the risk of myocardial infarction based on coronary artery calcium score in healthy Singaporeans

INTRODUCTION

Cardiovascular disease was the top cause of deaths and disability in Singapore in 2018, contributing extensively to the local healthcare burden. Primary prevention identifies at-risk individuals for the swift implementation of preventive measures. This has been traditionally done using the Singapore-adapted Framingham Risk Score (SG FRS). However, its most recent recalibration was more than a decade ago. Recent changes in patient demographics and risk factors have undermined the accuracy of SG FRS, and the rising popularity of wearable health metrics has led to new data types with the potential to improve risk prediction.

METHODS

In healthy Singaporeans enrolled in SingHEART study (absence of any clinical outcomes), we investigated improvements in SG FRS to predict myocardial infarction risk based on high/low classification of the Agatston score (surrogate outcome). Logistic regression, receiver operating characteristic and net reclassification index (NRI) analyses were conducted.

RESULTS

We demonstrated a significant improvement in the area under curve (AUC) of SG FRS (AUC = 0.641) after recalibration and incorporation of additional variables (fasting blood glucose and wearable-derived activity levels) (AUC = 0.774) ( P < 0.001). SG FRS++ significantly increases accuracy in risk prediction (NRI = 0.219, P = 0.00254).

CONCLUSION

Existing Singapore cardiovascular disease risk prediction guidelines should be updated to improve risk prediction accuracy. Recalibrating existing risk functions and utilising wearable metrics that provide a large pool of objective health data can improve existing risk prediction tools. Lastly, activity levels and prediabetic state are important factors for coronary heart disease risk stratification, especially in low-risk individuals.

The Singapore National Precision Medicine Strategy

Precision medicine promises to transform healthcare for groups and individuals through early disease detection, refining diagnoses and tailoring treatments. Analysis of large-scale genomic-phenotypic databases is a critical enabler of precision medicine. Although Asia is home to 60% of the world's population, many Asian ancestries are under-represented in existing databases, leading to missed opportunities for new discoveries, particularly for diseases most relevant for these populations. The Singapore National Precision Medicine initiative is a whole-of-government 10-year initiative aiming to generate precision medicine data of up to one million individuals, integrating genomic, lifestyle, health, social and environmental data. Beyond technologies, routine adoption of precision medicine in clinical practice requires social, ethical, legal and regulatory barriers to be addressed. Identifying driver use cases in which precision medicine results in standardized changes to clinical workflows or improvements in population health, coupled with health economic analysis to demonstrate value-based healthcare, is a vital prerequisite for responsible health system adoption.

High-Resolution Digital Phenotypes From Consumer Wearables and Their Applications in Machine Learning of Cardiometabolic Risk Markers: Cohort Study

BACKGROUND

Consumer-grade wearable devices enable detailed recordings of heart rate and step counts in free-living conditions. Recent studies have shown that summary statistics from these wearable recordings have potential uses for longitudinal monitoring of health and disease states. However, the relationship between higher resolution physiological dynamics from wearables and known markers of health and disease remains largely uncharacterized.

OBJECTIVE

We aimed to derive high-resolution digital phenotypes from observational wearable recordings and to examine their associations with modifiable and inherent markers of cardiometabolic disease risk.

METHODS

We introduced a principled framework to extract interpretable high-resolution phenotypes from wearable data recorded in free-living conditions. The proposed framework standardizes the handling of data irregularities; encodes contextual information regarding the underlying physiological state at any given time; and generates a set of 66 minimally redundant features across active, sedentary, and sleep states. We applied our approach to a multimodal data set, from the SingHEART study (NCT02791152), which comprises heart rate and step count time series from wearables, clinical screening profiles, and whole genome sequences from 692 healthy volunteers. We used machine learning to model nonlinear relationships between the high-resolution phenotypes on the one hand and clinical or genomic risk markers for blood pressure, lipid, weight and sugar abnormalities on the other. For each risk type, we performed model comparisons based on Brier scores to assess the predictive value of high-resolution features over and beyond typical baselines. We also qualitatively characterized the wearable phenotypes for participants who had actualized clinical events.

RESULTS

We found that the high-resolution features have higher predictive value than typical baselines for clinical markers of cardiometabolic disease risk: the best models based on high-resolution features had 17.9% and 7.36% improvement in Brier score over baselines based on age and gender and resting heart rate, respectively (P<.001 in each case). Furthermore, heart rate dynamics from different activity states contain distinct information (maximum absolute correlation coefficient of 0.15). Heart rate dynamics in sedentary states are most predictive of lipid abnormalities and obesity, whereas patterns in active states are most predictive of blood pressure abnormalities (P<.001). Moreover, in comparison with standard measures, higher resolution patterns in wearable heart rate recordings are better able to represent subtle physiological dynamics related to genomic risk for cardiometabolic disease (improvement of 11.9%-22.0% in Brier scores; P<.001). Finally, illustrative case studies reveal connections between these high-resolution phenotypes and actualized clinical events, even for borderline profiles lacking apparent cardiometabolic risk markers.

CONCLUSIONS

High-resolution digital phenotypes recorded by consumer wearables in free-living states have the potential to enhance the prediction of cardiometabolic disease risk and could enable more proactive and personalized health management.

Application of ensemble machine learning algorithms on lifestyle factors and wearables for cardiovascular risk prediction

This study looked at novel data sources for cardiovascular risk prediction including detailed lifestyle questionnaire and continuous blood pressure monitoring, using ensemble machine learning algorithms (MLAs). The reference conventional risk score compared against was the Framingham Risk Score (FRS). The outcome variables were low or high risk based on calcium score 0 or calcium score 100 and above. Ensemble MLAs were built based on naive bayes, random forest and support vector classifier for low risk and generalized linear regression, support vector regressor and stochastic gradient descent regressor for high risk categories. MLAs were trained on 600 Southeast Asians aged 21 to 69 years free of cardiovascular disease. All MLAs outperformed the FRS for low and high-risk categories. MLA based on lifestyle questionnaire only achieved AUC of 0.715 (95% CI 0.681, 0.750) and 0.710 (95% CI 0.653, 0.766) for low and high risk respectively. Combining all groups of risk factors (lifestyle survey questionnaires, clinical blood tests, 24-h ambulatory blood pressure and heart rate monitoring) along with feature selection, prediction of low and high CVD risk groups were further enhanced to 0.791 (95% CI 0.759, 0.822) and 0.790 (95% CI 0.745, 0.836). Besides conventional predictors, self-reported physical activity, average daily heart rate, awake blood pressure variability and percentage time in diastolic hypertension were important contributors to CVD risk classification.

Clinical- and omics-based models of subclinical atherosclerosis in healthy Chinese adults: a cross-sectional exploratory study

BACKGROUND

Classical risk factors, such as fasting cholesterol, blood pressure (BP), and diabetes status are used today to predict the risk of developing cardiovascular disease (CVD). However, accurate prediction remains limited, particularly in low-risk groups such as women and younger individuals. Growing evidence suggests that biomarker concentrations following consumption of a meal challenge are better and earlier predictors of disease development than biomarker concentrations.

OBJECTIVE

To test the hypothesis that postprandial responses of circulating biomarkers differ between healthy subjects with and without subclinical atherosclerosis (SA) in an Asian population at low risk of coronary artery disease (CAD).

METHODS

One hundred healthy Chinese subjects (46 women, 54 men) completed the study. Subjects consumed a mixed-meal test and 164 blood biomarkers were analyzed over 6 h by using a combination of chemical and NMR techniques. Models were trained using different methodologies (including logistic regression, elastic net, random forest, sparse partial least square) on a random 75% subset of the data, and their performance was evaluated on the remaining 25%.

RESULTS

We found that models based on baseline clinical parameters or fasting biomarkers could not reliably predict SA. By contrast, an omics model based on magnitude and timing of postprandial biomarkers achieved high performance [receiving operating characteristic (ROC) AUC: 91%; 95% CI: 77, 100). Investigation of key features of this model enabled derivation of a considerably simpler model, solely based on postprandial BP and age, with excellent performance (AUC: 91%; 95% CI: 78, 100).

CONCLUSION

We report a novel model to detect SA based on postprandial BP and age in a population of Asian subjects at low risk of CAD. The use of this model in large-scale CVD prevention programs should be explored. This trial was registered at ClinicalTrials.gov as NCT03531879.

Impact of Age and Sex on Subclinical Coronary Atherosclerosis in a Healthy Asian Population

Background

The influence of age and sex on clinical atherosclerotic cardiovascular disease is well reported, but literature remains sparse on whether these extend to the disease in its preclinical stage.

Objectives

The purpose of this study was to report the prevalence, risk factors, and impact of age and sex on the burden of subclinical coronary atherosclerosis in a healthy Asian population.

Methods

Healthy subjects age 30 to 69 years, with no history of cardiovascular disease or diabetes were recruited from the general population. Subclinical coronary atherosclerosis was quantified via the coronary artery calcium score (CAC) with CAC of 0 indicating absence of calcified plaque, 1 to 10 minimal plaque, 11 to 100 mild plaque, and >100 moderate to severe plaque.

Results

A total of 663 individuals (mean age 49.4 ± 9.2 years; 44.8% men) were included. The prevalence of any CAC was 29.3%, with 9% having CAC >100. The prevalence was significantly higher in men than women (43.1% vs 18.0%; P < 0.001). Multivariable analysis revealed significant associations of increasing age, male sex, higher blood pressure, increased glucose levels, and higher low-density lipoprotein cholesterol levels with the presence of any CAC. Low-density lipoprotein cholesterol was more significantly associated with CAC in women compared with men (P_interaction = 0.022).

Conclusions

The prevalence of preclinical atherosclerosis increased with age, and was higher in men, with sex-specific differences in associated risk factors. These results will better inform individualized future risk management strategies to prevent the development and progression of coronary artery disease within healthy individuals.

Deep learning and the electrocardiogram: review of the current state-of-the-art

In the recent decade, deep learning, a subset of artificial intelligence and machine learning, has been used to identify patterns in big healthcare datasets for disease phenotyping, event predictions, and complex decision making. Public datasets for electrocardiograms (ECGs) have existed since the 1980s and have been used for very specific tasks in cardiology, such as arrhythmia, ischemia, and cardiomyopathy detection. Recently, private institutions have begun curating large ECG databases that are orders of magnitude larger than the public databases for ingestion by deep learning models. These efforts have demonstrated not only improved performance and generalizability in these aforementioned tasks but also application to novel clinical scenarios. This review focuses on orienting the clinician towards fundamental tenets of deep learning, state-of-the-art prior to its use for ECG analysis, and current applications of deep learning on ECGs, as well as their limitations and future areas of improvement.

Family history assessment significantly enhances delivery of precision medicine in the genomics era

BACKGROUND

Family history has traditionally been an essential part of clinical care to assess health risks. However, declining sequencing costs have precipitated a shift towards genomics-first approaches in population screening programs rendering the value of family history unknown. We evaluated the utility of incorporating family history information for genomic sequencing selection.

METHODS

To ascertain the relationship between family histories on such population-level initiatives, we analysed whole genome sequences of 1750 research participants with no known pre-existing conditions, of which half received comprehensive family history assessment of up to four generations, focusing on 95 cancer genes.

RESULTS

Amongst the 1750 participants, 866 (49.5%) had high-quality standardised family history available. Within this group, 73 (8.4%) participants had an increased family history risk of cancer (increased FH risk cohort) and 1 in 7 participants (n = 10/73) carried a clinically actionable variant inferring a sixfold increase compared with 1 in 47 participants (n = 17/793) assessed at average family history cancer risk (average FH risk cohort) (p = 0.00001) and a sevenfold increase compared to 1 in 52 participants (n = 17/884) where family history was not available (FH not available cohort) (p = 0.00001). The enrichment was further pronounced (up to 18-fold) when assessing only the 25 cancer genes in the American College of Medical Genetics (ACMG) Secondary Findings (SF) genes. Furthermore, 63 (7.3%) participants had an increased family history cancer risk in the absence of an apparent clinically actionable variant.

CONCLUSIONS

These findings demonstrate that the collection and analysis of comprehensive family history and genomic data are complementary and in combination can prioritise individuals for genomic analysis. Thus, family history remains a critical component of health risk assessment, providing important actionable data when implementing genomics screening programs.

TRIAL REGISTRATION

ClinicalTrials.gov NCT02791152 . Retrospectively registered on May 31, 2016.