About

Safinah Ali is an Assistant Professor at NYU Steinhardt in the Department of Administration, Leadership, and Technology. She holds a PhD in Media Arts and Sciences from the Massachusetts Institute of Technology's Media Lab, along with a Masters degree in Science from MIT, a Masters degree in Human-Computer Interaction from Carnegie Mellon University, and a Bachelor's degree in Design from the Indian Institute of Technology Guwahati. Her research focuses on using AI tools and agents to support human creativity and learning. She develops child-AI interactions to foster creative learning and creates inclusive AI curricula to empower creators. Safinah has deployed her AI learning resources to teachers worldwide and engaged in extensive student outreach, developing teaching resources, interactive tools, and assessments to make AI concepts accessible and study their influence on student learning. She also develops social robotic agents to provide creativity scaffolding and social-emotional support to children, personalizing interactions for students with diverse needs. Her long-term research demonstrates how these interventions enhance verbal, figural, and constructional creativity among children. Her work has been published in numerous academic journals and peer-reviewed conferences, and she has received awards including the Teaching Excellence Award 2024, the Microsoft Research fellowship (2022-2024), and MIT’s Teaching Development fellowship.

Research topics

• Sociology
• Political Science
• Medicine
• Psychology
• Demography
• Gerontology
• Social Science
• Environmental health
• Psychiatry
• Developmental psychology
• Social psychology
• Business
• Advertising
• Nursing
• Medical education
• Public relations
• Internal medicine

Selected publications

• Using Group Chats to Drive Behavior Change in Digital Health Interventions: Scoping Review and Realist Synthesis

  Journal of Medical Internet Research · 2026-04-13

  articleOpen accessSenior author

  Background: Group chats on platforms such as WhatsApp (Meta Platforms, Inc), WeChat (Tencent Holdings Limited), and Telegram (Telegram FZ-LLC) are central to everyday communication in many settings, including across low- and middle-income countries and among groups often overlooked by one-to-one or app-based digital health tools. Yet their roles and underlying mechanisms as intentionally designed health interventions have not been comprehensively examined. Objective: This scoping review with realist synthesis aimed to map the characteristics and reported outcomes of group chat-based health interventions and contextual conditions shaping their outcomes. It mapped intervention characteristics and identified context-mechanism-outcome (CMO) configurations driving behavior change. Methods: We included empirical studies in which group chats were the primary digital component of the health intervention. Searches of PubMed, Embase, MEDLINE, Web of Science, and Scopus (2005-2026) and reference screening identified eligible studies. Data were extracted on study characteristics, health domains, settings, participants, platform, intervention features, and outcome domains. For the realist synthesis, CMO configurations were analyzed using reflexive thematic analysis informed by social cognitive theory and synthesized into a final program theory. Results: Eighty-one studies were included. Publications increased sharply after 2020, with evidence clustering in mental health (25/81, 30.9%) and maternal and child health (20/81, 24.7%). Most studies enrolled adult participants, including pregnant or postpartum women, caregivers, and people managing chronic conditions. WhatsApp (38/81, 46.9%) and WeChat (21/81, 25.9%) were the most frequently studied platforms. Interventions were commonly delivered in clinical or community settings and typically involved mixed membership groups that included health professionals alongside participants (68/81, 84.0%). The realist synthesis identified 12 recurring CMO configurations across 5 domains: capability and actionability, confidence and motivation, modeling and norms, safe and supportive environment and access, and self-regulation and maintenance. These mechanisms were activated or suppressed by facilitation quality, group composition, cultural alignment, technological access, and social norms. The final program theory depicts group chats as dynamic systems where access, facilitation, and group structure shape how the domains amplify or dampen one another, explaining shifts from low engagement to high trust spaces sustaining behavior change. Conclusions: This review integrates descriptive mapping of interventions with realist program theory to explain how context and facilitation activate mechanisms of change, unlike prior reviews that have focused narrowly on effectiveness of digital modalities. This dual approach provides a practical, mechanism-focused basis for designing and scaling group chat interventions in practice, particularly given their potential as a low-cost, high-reach strategy embeddable within broader digital health programs. Realizing this potential depends on treating group chats as purposefully designed social environments, with deliberate attention to facilitation quality, equity-oriented access, and alignment between group structure, norms, and communication practices.

  PublisherDOI

• Real-world outcomes of hybrid obesity care using digital coaching and GLP-1 therapy in a multi-ethnic Asian setting

  International Journal of Obesity · 2026-04-09

  articleOpen access1st author

  BACKGROUND: Obesity remains a major health challenge globally and in Asia, driving cardio-metabolic disease risks. Glucagon-like peptide-1 receptor agonists (GLP-1 RAs) and mobile health (mHealth) coaching each demonstrate weight loss efficacy, but real-world evidence for hybrid models combining these treatments remains limited, especially in multi-ethnic Asian settings. METHODS: We evaluated real-world outcomes among 708 adults enrolled in NOVI Optimum Plus, a physician-led obesity program in Singapore that integrates GLP-1 RA pharmacotherapy with app-based lifestyle coaching. Data on weight, metabolic indicators, and engagement were extracted from clinical records and the mHealth platform. Linear mixed models estimated changes over 6-18 months, stratified by engagement, metabolic status, and ethnicity. RESULTS: Participants (mean age 42.1 years; 64.1% female) were primarily East Asian (45.5%), European (26.8%), South Asian (13.3%), and Southeast Asian (10.3%). Most received semaglutide (86% oral 14 mg). At 12 months, mean weight loss was 12.7% (95% CI: -14.0, -11.3) and BMI dropped by 4.1 points, with further weight loss reaching 14.7% at 18 months. Systolic blood pressure decreased by 11.5 mmHg, body fat percentage by 8.8%, waist-to-hip ratio improved from 0.83 to 0.80, and HbA1c declined by 0.6%. Greater app engagement was linked to 2.0-2.2% additional weight loss, 0.72 kg/m² more BMI reduction, and up to 2.9 mmHg greater systolic BP drop. More frequent health coach contact contributed modest added improvements for weight and BMI. Weight loss was significantly lower among East Asians and those with hyperglycemia. CONCLUSION: In this real-world Asian setting, hybrid obesity care combining GLP-1 RAs with digital coaching produced clinically meaningful, sustained weight and metabolic improvements. Higher engagement consistently enhanced outcomes, supporting scalable integrated models tailored for diverse populations.

  PublisherOA PDFDOI

• Reorienting health promotion in remote Pacific Island communities: building family and child-centered agent-of-change models for sustainable behavior change

  The Lancet Regional Health - Western Pacific · 2026-04-27

  articleOpen access1st authorCorresponding
  PublisherDOI

• How Different Forms of Parent-Child Interactions Shape Healthy Aging: Evidence from over 100,000 Parent-Child Relationships on Lifestyle Behaviors Among Chinese Older Adults (2011–2020)

  Behavioral Medicine · 2026-04-07

  article1st authorCorresponding

  As family structures evolve in China, adult children's engagement with aging parents has become increasingly complex, yet little is known about how distinct forms of interaction shape older adults' health behaviors. This study examined how visitation, interaction frequency, co-residence, and financial support relate to four lifestyle behavior domains: total and leisure physical activity, smoking, alcohol use, and sleep. Data were drawn from 22,114 older adults and 138,243 parent-child dyads in the China Health and Retirement Longitudinal Study (2011 to 2020) and analyzed using multivariable mixed-effects models that accounted for repeated observations and clustering at both the respondent and child levels. Weekly interaction increased from 86.0% to 89.1%, while in-person visitation declined from 60.4% to 55.2%. Interaction was more common with healthier, more highly educated, and female children; male children were more likely to co-reside. Weekly interaction was associated with lower odds of low total physical activity, low leisure activity, and drinking. Co-residence reduced odds of drinking, and increased odds of low total activity, low leisure activity, and smoking. Financial support predicted lower odds of low total activity and low leisure activity, while weekly interaction was linked to lower odds of suboptimal sleep. Even as face-to-face contact declines, interaction continues to grow, likely through digital platforms, and distinctly shapes lifestyle behaviors. This study lays important groundwork for more intensive mixed methods research to explore mechanisms underlying how different engagement patterns influence health behaviors in aging populations.

  PublisherDOI

• Integration of Continuous Glucose Monitoring With HbA1c to Improve the Detection of Prediabetes in Asian Individuals: Model Development Study

  JMIR Diabetes · 2026-04-27

  articleOpen access

  Background: Glycated hemoglobin (HbA1c) is a convenient tool to evaluate glycemic status but its ability to detect individuals at risk for type 2 diabetes is limited. Objective: Exploiting the glycemic variability captured in continuous glucose monitoring (CGM), we used a well-characterized Asian cohort study from Singapore to assess whether utilizing CGM features in a machine learning model can improve the detection of prediabetes as compared to using HbA1c alone. Methods: In this study, 406 nondiabetic Asian participants underwent an oral glucose tolerance test and had their fasting and 2-hour plasma glucose concentrations measured, together with HbA1c, to classify them as with normoglycemia or prediabetes. They also wore a CGM sensor for 14 days. CGM profile features were extracted and prediction models were constructed with random subsampling validation to evaluate predictive efficacy. The use of CGM and HbA1c data alone or in combination was assessed for the ability to correctly distinguish prediabetes from normoglycemia. Results: In this cohort (N=406), 189 (46.6%) individuals had prediabetes. The majority of the cohort were women (n=236, 58.1%) and of Chinese ethnicity (n=267, 65.8%). Those with prediabetes were slightly older, heavier, and had higher glucose levels with more variability than the normoglycemia group. A 2-step approach was used where those with HbA1c ≥5.7% were automatically categorized as having prediabetes; the model then focused on the prediction capability of the CGM features among individuals with HbA1c <5.7%. The prediction models with CGM outperformed the benchmark for comparison defined by HbA1c ≥5.7%, where they yielded an area under the receiver operating characteristic curve of 0.866-0.876, with a lower specificity of 78%-80% but a vastly improved sensitivity of 76%-78%. Conclusions: Adding CGM to HbA1c in a 2-step approach greatly improved the sensitivity of detecting prediabetes in an Asian population. Given the benefits to optimizing lifestyle behaviors and its growing acceptability among the nondiabetic population, CGM is a promising alternative for type 2 diabetes mellitus risk screening.

  PublisherDOI

• “The first person to reach you becomes dearest to you”: Mapping health information diffusion in Myanmar's polycrisis through digital strategies, networked pathways, and innovative solutions

  Social Science & Medicine · 2026-03-25

  articleSenior authorCorresponding
  PublisherDOI

• Empower families to lead the design of their ageing loved ones’ health care

  Nature · 2025-03-26 · 3 citations

  articleOpen access1st authorCorresponding
  PublisherOA PDFDOI

• How We Live: Characteristics of Multigenerational Households among Asian Americans (2006-2018)

  Journal of Asian Health · 2025-04-29 · 1 citations

  articleOpen access

  Background: Household structure is an important social determinant of health. Asian Americans are the race/ethnicity group most likely to live in multigenerational households, but little is known about which characteristics are associated with living in a multigenerational household among Asian Americans as a whole, or as disaggregated subgroups. Objective: We examined the characteristics associated with living in a multigenerational household among Asian Americans (aggregated and subgroups), compared to Non-Hispanic Whites. Design: Cross-sectional study using the National Health Interview Survey (2006-2018). Participants: Our sample included 572,783 adults: 515,420 Non-Hispanic Whites, 11,113 Asian Indians, 11,864 Chinese, 13,000 Filipino, and 21,386 Other Asians. Main Measures: We used binary logistic regression to examine how living in multigenerational households (outcome) was associated with race and sociodemographic characteristics among the entire population and by race/ethnic subgroup. Key Results: Approximately 15% of the study population lived in multigenerational households. 12.4% and 24.0% of Non-Hispanic Whites and Asians respectively lived in multigenerational households. Filipinos had the highest (30.5%) and Asian Indians had the lowest (19.5%) proportion of people living in multigenerational households. Aggregated Asians had twice the odds compared to Non-Hispanic Whites of living in multigenerational households (OR = 2.32, 95% CI: [2.17-2.49]) Foreign born Asian Indians compared to U.S. born Asian Indians were less likely (OR = 0.67, 95% CI: [0.51-0.87]) to live in a multigenerational household. Foreign born Filipinos compared to U.S.-born Filipinos had twice the odds (OR = 2.04, 95% CI: [1.72-2.42]) of living in multigenerational households. Conclusions: Asian Americans are more likely to live in multigenerational households compared to Non-Hispanic Whites, although proportions vary by Asian subgroups. Characteristics associated with living in a multigenerational vary. Understanding the characteristics of multigenerational households in Asian Americans can inform public health practice.

  PublisherOA PDFDOI

• Diabetes distress among immigrants of south Asian descent living in New York City: baseline results from the DREAM randomized control trial

  BMC Public Health · 2025-02-02 · 2 citations

  articleOpen access

  BACKGROUND: Type 2 diabetes (T2D) disproportionately affects individuals of South Asian descent. Additionally, diabetes distress (DD) may lead to complications with diabetes management. This study examines the prevalence of DD among foreign-born individuals of South Asian descent in New York City (NYC) and its association with sociodemographic and clinical factors. METHODS: Baseline data was collected from the Diabetes Research, Education, and Action for Minorities (DREAM) Initiative, an intervention designed to reduce hemoglobin A1c (HbA1c) among South Asian individuals with uncontrolled T2D at primary care practices in NYC. The Diabetes Distress Scale (DDS) measured DD, and Core Healthy Days Measures assessed physical and mental healthy days. Sociodemographic variables were analyzed using descriptive statistics, Chi-square tests assessed categorical variables, and Wilcoxon Rank Sum tests evaluated continuous variables (Type I error rate = 0.05). Logistic regression models examined associations between HbA1c, mental health, and other covariates with dichotomized DD subscales. RESULTS: Overall, 414 participants completed the DDS at baseline (median age = 55.2 years; SD = 9.8). All were born outside of the US; the majority were born in Bangladesh (69.8%) followed by India, Pakistan, and Nepal (24.7%) and Guyana and Trinidad and Tobago (5.5%). High emotional burden, regimen-related distress and physician-related distress were reported by 25.9%, 21.9%, and 6.2% of participants, respectively. In adjusted analyses, individuals with ≥ 1 day of poor mental health had higher odds of overall distress (OR:3.8, p = 0.013), emotional burden (OR:4.5, p < 0.001), and physician-related distress (OR:4.6, p = 0.007) compared to individuals with no days of poor mental health. Higher HbA1c (OR:1.45, p = < 0.001) was associated with regimen-related distress; and lower emotional support was associated with overall distress (OR:0.92, p < 0.001) and regimen-related distress (OR:0.95, p = 0.012). Individuals born in Bangladesh had significantly lower odds of overall distress, emotional burden, and regimen-related distress compared to individuals born in Guyana and Trinidad and Tobago. CONCLUSIONS: Findings highlight the rate and risk factors of DD among individuals of South Asian descent living in NYC. Screening for DD in patients with prediabetes or diabetes should be integrated to address mental and physical health needs. Future research can benefit from a longitudinal analysis of the impact of DD on diabetes self-management and health outcomes. TRIAL REGISTRATION: This study uses baseline data from "Diabetes Management Intervention for South Asians" (NCT03333044), which was registered with clinicaltrials.gov on 6/11/2017.

  PublisherOA PDFDOI

• Empowering Young Adults as Agents of Household Dietary Change: Findings From a Pilot Study Involving a Digital, Family-Led Sodium Reduction Intervention in Singapore

  Current Developments in Nutrition · 2025-08-05 · 1 citations

  articleOpen accessSenior author

  <h2>Abstract</h2><h3>Background</h3> Unhealthy dietary habits such as high sodium intake are socially embedded and often resistant to individual-level interventions. Family-led approaches, where one member initiates change within the household, may offer a more effective alternative. <h3>Objective</h3> This pilot study assessed the feasibility and preliminary impact of a digitally delivered, young adult-led sodium reduction intervention on household-level knowledge, attitudes, and behaviors in Singapore. <h3>Methods</h3> In a pre-post, single-group design, 35 young adults (mean age: 24.4) completed a co-created, self-paced online course featuring video lessons, interactive assignments, and personalized feedback. Over two weeks, participants developed sodium-reduction goals and implemented them through tailored four-week action plans. Weekly reflections and course metadata captured goal progress, effort, strategies, and barriers. Family members (n=79, mean age: 43.0) completed parallel pre- and post-intervention surveys, although they did not receive the intervention directly. Surveys assessed constructs from the Theory of Planned Behavior. Multivariable linear mixed models evaluated changes over time, adjusting for demographic and health characteristics. <h3>Results</h3> Overall, 35 young adults (mean age:24.4, SD:3.1) and 79 family members (mean age:43.0, SD:15.5) completed the intervention. Young adults took an average of 7.7 days to complete the course, with most crafted goals focusing on reducing sodium when eating out. Participants reported higher effort and success with personal goals than family-oriented ones. Perceived behavioral control showed the greatest improvement amongst both young adults (+2.64, 95%CI:2.05–3.22) and family members (+1.82, 95%CI:1.42–2.22). Significant gains were also observed in knowledge, behaviors, subjective norms, and behavioral intentions for all participants (all p<0.001). Engagement metrics (e.g. time spent on course, effort put into action plans) were not associated with differential changes in most outcomes. <h3>Conclusion</h3> A young adult-led, family-focused digital intervention was feasible and demonstrated preliminary improvements in household sodium-related outcomes, warranting further evaluation in larger, more diverse populations.

  PublisherOA PDFDOI

Frequent coauthors

• Ralph J. DiClemente

  New York University

  34 shared

• Nadia Islam

  Mayo Clinic in Arizona

  24 shared

• Angela Trude

  New York University

  22 shared

• Joshua Foreman

  Centre for Eye Research Australia

  21 shared

• Yuxuan Gu

  18 shared

• W.H. Wilson Tang

  Gates (United States)

  18 shared

• Caitlin M. Lowery

  University of North Carolina at Chapel Hill

  18 shared

• Ariadna Capasso

  Health Resources in Action

  18 shared

Education

• PhD, School of Global Public Health

  New York University

• Bachelor of Arts (Public Health, Political Science)

  Johns Hopkins University

  2019

• Coursework

  University of Queensland

  2016

Awards & honors

• Teaching Excellence Award 2024
• Microsoft Research fellowship (2022-2024)
• MIT’s Teaching Development fellowship