WHWeiting Huang

Weiting Huang

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article

Previous Organizations / Institutes

The organizations / institutes from the scholar's other articles in the past ten years


Weiting Huang has a total of 164 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Physical activity phenotypes from volume-based versus intensity-based wearable metrics: discordant classification and divergent metabolomic, body composition, and cardiac magnetic resonance associations

    by Sarah Shi Sien Goh, Khung Keong Yeo, Jonathan Yap, Siew Ching Kong, Weng Khong Lim, Mark Kei Fong Wong, Eddie Yin Kwee Ng, Xiayan Shen, Alex Weixian Tan, Weiting Huang

    Background: Consumer wearables report direct volume-based metrics (steps, distance, floors climbed) alongside device-derived metrics generated by manufacturer algorithms (activity-intensity minutes, energy expenditure). Both are widely used to characterize individuals into activity phenotypes, yet whether these representations yield concordant activity phenotypes and biological associations is unclear. Objective: To compare activity phenotypes derived from direct accelerometry with those derived from device-derived intensity metrics, and to evaluate each against metabolomic, body composition, and cardiac magnetic resonance (CMR) phenotypes. Methods: Using the SingHEART cohort, two hierarchical cluster analyses using Ward linkage yielded five phenotypes each for volume-based and intensity-based metrics. Agreement was quantified by the adjusted Rand index. For analyses, the two high-variability phenotypes were combined, giving four groups: sedentary, average movers, regular exercisers, and high-variability movers. Serum ¹H-NMR metabolomics was analyzed by PERMANOVA with Holm adjustment across pairwise contrasts. Body and CMR associations were assessed using linear regression with false discovery rate correction. Results: In total, 984 adults (mean age 45.6 years, 48.3% male) wore accelerometers for a mean of 7.1 valid days. The two classifications agreed poorly: adjusted Rand index was 0.18, with only 46.2% assigned identically. Volume-based phenotypes differed in metabolomic profiles ( P < .001). Active participants (high-variability movers plus regular exercisers) differed from sedentary participants across 106 of 250 biomarkers (FDR q < 0.05), including higher HDL, lower VLDL, LDL and lower GlycA. Conversely, although intensity-based phenotypes showed a nominally significant omnibus metabolomic association ( P = .003), no metabolomic difference was observed between active and sedentary groups defined by intensity-based phenotyping ( P = .29). Across 16 body composition and CMR outcomes, volume-based phenotypes were associated with 11 and intensity-based phenotypes with 6, in divergent directions: volume-based high-variability movers showed lower body fat, higher skeletal muscle, and larger ventricular volumes and mass, whereas intensity-based high-variability movers showed higher body mass index and waist circumference. Conclusions: Volume-based metrics and intensity-based metrics assign the same individuals to discordant activity phenotypes. Volume-based phenotypes showed more consistent associations across metabolomic, body composition, and cardiac imaging measures. These findings suggest that the choice of wearable-derived metrics substantially influences digital activity phenotyping and its biological interpretation.

  • Open Access

    Prevalence of hypertension phenotypes in an urban population with low cardiovascular risk in Singapore: a cross-sectional study

    by Aminath Shiwaza Moosa, Shera Chaterji, Ding Xuan Ng, Barath Kumar, Siew Ching Kong, Weiting Huang, Jonathan Yap, Khung Keong Yeo, Chirk Jenn Ng

    Background: Global reports on hypertension phenotypes exist, but population‑based evidence remains limited. A significant gap also persists in understanding these phenotypes within multi‑ethnic urban populations. Thus, this study aimed to determine the prevalence of hypertension phenotypes and their associated factors in a low-risk population in Singapore. Methods: The cross-sectional study analysed data from 981 adults aged 21–69 years with a low 10-year cardiovascular risk (based on the 2023 Singapore modified Framingham risk score) and not on any antihypertensives. Participants were recruited from public or clinic settings using convenience sampling. Following consent, participants completed clinical examinations and investigations using structured questionnaires and validated instruments, such as office BP and 24-hour ambulatory BP monitoring (ABPM) devices. Results: The mean age of the participants was 45.5 years (SD 11.8 years). Most were females (51.6%) and of Chinese ethnicity (91.2%). Normal body mass index (BMI) and waist circumference were observed in 47.0% and 60.7% of the cohort, respectively. Lifestyle data revealed that most participants were nonsmokers (90.9%) and engaged in physical activity (81.3%). The prevalence of hypertension was 25.5% by office BP and 22.7% by 24-hour ABPM, with 38.5% diagnosed via any ABPM target. Nocturnal hypertension (NH) was the most prevalent hypertension phenotype (36.1%), while white-coat hypertension had the lowest prevalence (6.5%). Daytime hypertension and masked hypertension (MH) were observed in 17.4% and 19.4% of participants, respectively. A non-dipping (ND) BP pattern was identified in 59.7% of the cohort. Conclusion: This study revealed a high prevalence of NH and ND, with MH emerging as the next most common phenotype in a low-risk, urban Singaporean population. These findings support the potential role of ABPM to unmask hypertension phenotypes. Longitudinal follow up of this cohort could determine their prognostic significance and implications for cardiovascular risk stratification.

  • Open Access

    Grip Strength and Five-Year Mortality Risk Across Three International Cohorts

    by David Shengyang Zhang, Tsubasa Tsutsumi, Nicholas L. Syn, Mei Chin Lim, Ganakirthana Kalpenya Devi, Agil Singaraselvan, Aditi Athreya, Jordan Jun Yi Low, Lin Aung Yin, Naing Naing Mon, et al.

    Background: Grip strength is widely used to identify low muscle strength in sarcopenia, but fixed diagnostic thresholds might not reflect its continuous, age-dependent association with mortality. We assessed its prognostic information beyond contemporary diagnostic thresholds and developed and externally validated age- and sex-specific values corresponding to five-year mortality risk.Methods: We analysed three population-based cohorts from the United States and China: NHANES, HRS, and CHARLS. Threshold analyses included 40 048 adults aged 50–80 years; prognostic analyses included 46 173 adults aged 45–80 years. Maximum measured grip strength was evaluated continuously and relative to AWGS 2025 and EWGSOP2 thresholds. Outcomes included five-year all-cause mortality, discontinuities at diagnostic thresholds, grip-strength values corresponding to 5%, 10%, and 15% five-year mortality risk, incremental discrimination, and external calibration. The prognostic model was developed in NHANES and externally validated in HRS and CHARLS.Findings: Prognostic analyses included 5590 NHANES participants, 22 136 HRS participants, and 18 447 CHARLS participants, with 533, 1718, and 885 deaths within five years, respectively. Mortality declined continuously with increasing grip strength, without reproducible discontinuities at diagnostic thresholds. Adding grip strength to age and sex increased Harrell’s C by 0·034 (95% CI 0·023–0·046) in NHANES, 0·037 (0·029–0·044) in HRS, and 0·014 (0·007–0·020) in CHARLS. The unchanged NHANES equation retained discrimination in HRS (C=0·718) and CHARLS (C=0·739). Observed-to-expected mortality ratios were 1·05 in HRS and 0·69 in CHARLS; baseline-hazard recalibration improved the CHARLS ratio to 1·02.Interpretation: Grip strength provided continuously graded prognostic information beyond binary diagnostic classification. Age- and sex-specific five-year mortality risk estimates might complement existing sarcopenia criteria, although population-specific recalibration and prospective evaluation are needed before clinical implementation. FundingNo external funding.

  • Open Access

    Integrating Cardiovascular Risk Screening into Routine Diabetic Retinopathy Screening: Cost-Effectiveness of a Retinal Imaging–Based Artificial Intelligence System (JARVIS) in Singapore

    by Sameera Senanayake, Weiting Huang, Sanjeewa Kularatna, See Pdf

    Background: People with diabetes have increased risk of major adverse cardiovascular events (MACE). Conventional cardiovascular risk tools, such as the Framingham Risk Score (FRS), may have limited accuracy in Asian populations. JARVIS (Joint Automated Retinal and Vascular Imaging System) is a retinal imaging-based artificial intelligence approach for cardiovascular risk stratification. We evaluated the cost-effectiveness of JARVIS versus FRS among adults with diabetes in Singapore. Methods: A five-state Markov model estimated costs and quality-adjusted life-years (QALYs) over 20 years from the Singapore healthcare system perspective. Adults with diabetes erwent screening with JARVIS or FRS. True-positive and false-positive individuals received preventive treatment. Inputs were derived from published literature, Singapore cost sources, and age-weighted estimates of high MACE risk. Parameter uncertainty was assessed using 1,000 Monte Carlo simulations. The willingness-to-pay threshold was S$45,000 per QALY gained. Findings: JARVIS had higher costs and greater health benefits than FRS with an an incremental cost-effectiveness ratio of S$4,536 per QALY gained. Its probability of being cost-effective was 88%. With confirmatory coronary artery calcium scoring for test-positive individuals, JARVIS remained cost-effective (S$13,293 per QALY gained). In 100,000 individuals, JARVIS was projected to prevent 4,727 MACE events and avert 887 deaths over 20 years. Interpretation: JARVIS is likely cost-effective for cardiovascular risk screening among adults with diabetes in Singapore. Integration within retinal screening pathways may improve early risk detection, reduce cardiovascular events, and provide good value for money.

  • Open Access

    Mosaic Loss-of-Y Chromosome in Men Is Associated With Aortic Enlargement But Not Aortic Stenosis

    by Matteo Morello, Jesse D. Cochran, Bethany A. Gholson, Christopher R. Xie, Weiting Huang, Sahithi Mikkilineni, Skyler Nelson, Coleen McNamara, Kenneth Walsh, Jonathan R. Lindner

  • Open Access

    Clinical and Genetic Determinants of Subclinical Coronary Atherosclerosis in Healthy Asian People

    by Chris Lenselink, Jonathan Yap, Weng Khong Lim, Jing Xian Teo, Erik Lipsic, Adriaan Voors, Weiting Huang, Swee Yaw Tan, Patrick Tan, Khung Keong Yeo

    Background: Genetic determinants may offer incremental value over routine clinical risk factors for the prediction of coronary atherosclerosis, possibly with different outcomes across populations. Methods: In Asian subjects without known cardiovascular disease and/or diabetes, coronary atherosclerosis was quantified by coronary artery calcium (CAC), with a score of ≥100 indicating moderate-severe calcification. Relevant single nucleotide polymorphisms (SNPs) were identified through published studies with genotypes extracted through whole genome sequencing. Multivariate logistic regression modelling was performed to derive a clinical, genetic and combined model. Models’ performance was assessed using the area under the curve (AUC), Akaike’s information criterion (AIC), and Brier score. Results: Of 845 participants, 74 individuals (8.8%) had a CAC score of at least 100. Independent clinical predictors of a CAC score ≥100 were higher age, male sex, higher systolic blood pressure, higher glucose, longer time asleep, larger number of daily steps and family history of ischemic heart disease. This clinical model had an AUC of 0.89, AIC of 323, and Brier score of 0.0570. The genetic model, with independent SNPs rs515135, rs2047009 and rs10757272, had an AUC of 0.64 (p<0.001 versus clinical model), AIC of 444, and Brier score of 0.0831. The combined model had an AUC of 0.91 (p=0.0026 versus clinical model), AIC of 280, and Brier score of 0.0570. Conclusion: In this healthy Asian cohort, a purely genetic prediction model for moderate-severe subclinical coronary atherosclerosis did not perform as well as its clinical counterpart. Incorporating both clinical and genetic variables allowed for only minor incremental predictive capabilities.

  • Open Access

    Prevalence, Trends, and Mortality of Low Muscle Mass and Its Association With Metabolic Diseases in the United States Population

    by Jing Kai Tioh, Glenn Jun Kit Ho, Ethan Kai Jun Tham, Saori Hieda, Hee Seung Yang, Ruhhan Rekhi, Talia Yi Hui Chan, Mark Dinesh Muthiah, Hirokazu Takahashi, Shadab Siddiqui Mohammad, et al.

    Background: Low muscle mass is increasingly recognized to extend beyond aging and is closely linked to metabolic dysfunction. Its population‐level burden, ethnic disparities, metabolic clustering and mortality impact in the United States remain incompletely characterized. Methods: We analysed data from NHANES 1999–2016, including adults aged ≥ 18 years with dual energy X‐ray absorptiometry (DEXA) measures. Low muscle mass was defined using established appendicular lean mass criteria adjusted for body size. Prevalence estimates were weighted to represent the US population. Temporal trends were assessed using Joinpoint regression, and mortality outcomes (all‐cause, cardiovascular and cancer‐related) were evaluated. Results: Low muscle mass affected 8.35% (95% CI: 7.84% to 8.90% of US adults and remained stable over time (average annual percent change [AAPC]: −1.475, 95% CI: −3.853 to 0.961, p = 0.233). Mexican Americans had the highest prevalence (18.60%, 95% confidence interval [CI]: 15.64 to 21.98, p < 0.001), while non‐Hispanic Blacks had the lowest (2.96%, 95% CI: 2.36 to 3.70, p < 0.001). Individuals with low muscle mass were older, had higher body mass index (BMI) and exhibited greater metabolic dysfunction, including hypertension (50.44%), diabetes (23.10%) and Metabolic Dysfunction‐Associated Steatotic Liver Disease (MASLD) (56.58%). Over a median follow‐up of 166 months, low muscle mass was associated with increased all‐cause mortality (adjusted hazard ratio: 1.43, 95% CI: 1.30 to 1.58, p < 0.001), though cardiovascular and cancer mortality associations attenuated after adjustment. The mortality risk was amplified when low muscle mass coexisted with diabetes, obesity, hypertension or hyperlipidaemia. Conclusions: Low muscle mass affects nearly one in 12 US adults and is strongly intertwined with metabolic dysfunction. Our findings suggest that low muscle mass functions primarily as a marker of adverse metabolic health rather than an isolated predictor of mortality. Its coexistence with cardiometabolic disease confers substantially increased mortality risk, underscoring the need to identify, diagnose and manage low muscle mass within a metabolic disease framework rather than as an isolated muscle disorder.


  • Open Access

    Relationship Between Activity Tracker Metrics and the Physical Activity Index and Their Association With Cardiometabolic Phenotypes, Subclinical Atherosclerosis, and Cardiac Remodeling: Cross-Sectional Study

    by Weiting Huang, Mark Kei Fong Wong, Enver De Wei Loh, Tracy Koh, Alex Weixian Tan, Xiayan Shen, Onur Varli, Siew Ching Kong, Calvin Woon Loong Chin, Swee Yaw Tan, et al.

    Background: Consumer wearable technology quantifies physical activity; however, the association between these metrics and cardiometabolic health requires further elucidation. Objective: This study identified latent factors derived from Fitbit heart rate metrics and their relationship with cross-sectional cardiovascular phenotypes. Methods: This cross-sectional analysis included 457 participants from the SingHEART study, a multiethnic, population-based study of Asian individuals aged 21 to 69 years recruited in Singapore. Participants wore the Fitbit Charge HR for 7 days, and data on physical activity metrics, self-reported physical activity index (PAI), blood tests, coronary artery calcium scores, and cardiac magnetic resonance imaging were collected. Exploratory factor analysis identified latent factors from Fitbit metrics, and multivariate regression analysis assessed associations with blood and cardiovascular imaging phenotypes. Results: Higher levels of self-reported PAI were significantly associated with a higher number of calories burned (P=.008), number of steps and floors climbed, distance, number of activity calories, and number of very active minutes (P<.001). However, there was no association between PAI and other Fitbit metrics. Using exploratory factor analysis, we identified three latent factors measured by Fitbit metrics: (1) elevated metabolic equivalents of task (METs; calories burned per day, minutes per day spent fairly active in 3-6 METs and very active in ≥6 METs, and activity calories), (2) total activity (steps per day, distance in kilometers per day, and number of floors per day), and (3) others, all with a Cronbach α of >0.7. Higher total activity was associated with increased high-density lipoprotein levels (β=0.06; P<.001), decreased triglyceride levels (β=−0.10; P=.006), and lower BMI (β=−0.63; P<.001) after adjustment for age, gender, systolic blood pressure, total cholesterol, and family history of heart disease. The interaction between total activity and elevated METs was associated with lower fasting glucose (β=−0.07; P=.004). Elevated METs were associated with higher log(coronary artery calcium+1) and higher BMI (P<.001). Total activity was significantly associated with higher indexed biventricular systolic (P=.01 for left and P=.006 for right) and diastolic volumes (P<.001) and higher indexed left ventricular mass (P=.005). Conclusions: We identified 3 groups of wearable metrics with distinct characteristics. While total activity had a significant relationship with self-reported PAI, most metrics of elevated METs did not. Total activity had a consistent and favorable association with lipid and glucose profiles and a dose-dependent association with cardiac remodeling. Elevated METs alone did not appear to have a significant association with favorable cardiovascular profiles. This study suggests that the total activity metrics are robust and dependable when interpreting an individual’s activity levels, with construct validity according to self-reported PAI and a positive association with lipid and glucose profiles, and demonstrate dose-dependent associations with cardiac remodeling after adjustment for demographics and risk factors. Findings related to elevated METs may be due to the Hawthorne effect and require further studies.

  • Open Access

    Association of blood pressure and left ventricular mass with subclinical coronary atherosclerosis

    by Enver De Wei Loh, Weiting Huang, Rehena Sultana, Siew Ching Kong, Chirk Jenn Ng, Swee Yaw Tan, Calvin Chin, Jonathan Jiunn Liang Yap, Khung K Yeo

    Background: Left ventricular (LV) mass is closely associated with atherosclerotic heart disease, but the mechanisms are not well defined. This study aimed to evaluate the risk factors associated with LV mass and subclinical coronary atherosclerosis, in an Asian population free of baseline cardiovascular disease. Methods: The SingHEART study is a population-based cohort in which individuals underwent ambulatory blood pressure (BP) monitoring, cardiac MRI to measure indexed LV mass index (LVMI) and coronary artery calcium (CAC) scoring. Individuals were stratified based on LVMI and the presence of CAC, and intergroup differences in risk factors were analysed. Logistic regression models were used to assess the interaction of BP and LVMI on prevalent CAC. Results: The study included 880 subjects (mean age 45.8±11.7 years, 51.4% women). Individuals with high LVMI had higher BP than those with normal LVMI. Across all LVMI groups, higher BP was associated with the presence of CAC. Compared with individuals with normotensive BP and normal LVMI, those with normotensive BP and high LVMI had no increased risk of prevalent CAC (p=0.530); however, risk was progressively higher in those with hypertensive BP and normal LVMI (risk ratio (RR) 1.47, 95% CI 1.13 to 1.91), or hypertensive BP and high LVMI (RR 1.72, 95% CI 1.26 to 2.36). Conclusions: In this healthy asymptomatic population, hypertension was the strongest risk factor for high LVMI and prevalent CAC. LV hypertrophy was a risk modifier, and its prognostic significance in adults without hypertension requires further study.


  • Open Access

    Relationship Between Activity Tracker Metrics and the Physical Activity Index and Their Association With Cardiometabolic Phenotypes, Subclinical Atherosclerosis, and Cardiac Remodeling: Cross-Sectional Study (Preprint)

    by Weiting Huang, Mark Kei Fong Wong, Enver De Wei Loh, Tracy Koh, Alex Weixian Tan, Xiayan Shen, Onur Varli, Siew Ching Kong, Calvin Woon Loong Chin, Swee Yaw Tan, et al.

    BACKGROUND Consumer wearable technology quantifies physical activity; however, the association between these metrics and cardiometabolic health requires further elucidation. OBJECTIVE This study identified latent factors derived from Fitbit heart rate metrics and their relationship with cross-sectional cardiovascular phenotypes. METHODS This cross-sectional analysis included 457 participants from the SingHEART study, a multiethnic, population-based study of Asian individuals aged 21 to 69 years recruited in Singapore. Participants wore the Fitbit Charge HR for 7 days, and data on physical activity metrics, self-reported physical activity index (PAI), blood tests, coronary artery calcium scores, and cardiac magnetic resonance imaging were collected. Exploratory factor analysis identified latent factors from Fitbit metrics, and multivariate regression analysis assessed associations with blood and cardiovascular imaging phenotypes. RESULTS Higher levels of self-reported PAI were significantly associated with a higher number of calories burned (P=.008), number of steps and floors climbed, distance, number of activity calories, and number of very active minutes (P<.001). However, there was no association between PAI and other Fitbit metrics. Using exploratory factor analysis, we identified three latent factors measured by Fitbit metrics: (1) elevated metabolic equivalents of task (METs; calories burned per day, minutes per day spent fairly active in 3-6 METs and very active in ≥6 METs, and activity calories), (2) total activity (steps per day, distance in kilometers per day, and number of floors per day), and (3) others, all with a Cronbach α of >0.7. Higher total activity was associated with increased high-density lipoprotein levels (β=0.06; P<.001), decreased triglyceride levels (β=−0.10; P=.006), and lower BMI (β=−0.63; P<.001) after adjustment for age, gender, systolic blood pressure, total cholesterol, and family history of heart disease. The interaction between total activity and elevated METs was associated with lower fasting glucose (β=−0.07; P=.004). Elevated METs were associated with higher log(coronary artery calcium+1) and higher BMI (P<.001). Total activity was significantly associated with higher indexed biventricular systolic (P=.01 for left and P=.006 for right) and diastolic volumes (P<.001) and higher indexed left ventricular mass (P=.005). CONCLUSIONS We identified 3 groups of wearable metrics with distinct characteristics. While total activity had a significant relationship with self-reported PAI, most metrics of elevated METs did not. Total activity had a consistent and favorable association with lipid and glucose profiles and a dose-dependent association with cardiac remodeling. Elevated METs alone did not appear to have a significant association with favorable cardiovascular profiles. This study suggests that the total activity metrics are robust and dependable when interpreting an individual’s activity levels, with construct validity according to self-reported PAI and a positive association with lipid and glucose profiles, and demonstrate dose-dependent associations with cardiac remodeling after adjustment for demographics and risk factors. Findings related to elevated METs may be due to the Hawthorne effect and require further studies. CLINICALTRIAL