About

Heather Allore, PhD, is a Professor of Medicine (Geriatrics) and of Biostatistics at Yale School of Medicine. Her research collaborations and methodological development work focus on the Data Management and Statistics Core of the Yale Alzheimer's Disease Research Center. She previously served as Director of Biostatistics at the Yale Program on Aging for 12 years, where she founded the field of Gerontological Biostatistics. Dr. Allore leads the Design and Statistics Core of the Imbedded Pragmatic Alzheimer’s Disease and AD-Related Dementias Clinical Trials Collaboratory and has co-authored work on health disparities in Alzheimer’s research. Her innovative analytic methods include joint trajectories of cognition, function, and mortality, addressing scientific questions related to older adults and persons with dementia through rigorous biostatistical approaches. With over 300 peer-reviewed articles and continuous NIH funding since 2000, her research emphasizes issues related to the design of clinical trials and studies of older adults, including risk prediction and longitudinal statistical methods such as latent class trajectory models and joint models. Dr. Allore is recognized for adapting new statistical methods to specific clinical questions rather than developing methods in isolation, which has contributed significantly to the field of gerontologic biostatistics.

Research topics

• Medicine
• Gerontology
• Psychiatry
• Internal medicine
• Nursing
• Medical emergency
• Physical therapy
• Psychology
• Emergency medicine
• Physical medicine and rehabilitation
• Pharmacology
• Social psychology
• Demography
• Pathology
• Surgery
• Clinical psychology
• Intensive care medicine

Selected publications

• Uncovering treatment effect heterogeneity in pragmatic gerontology trials

  Experimental Gerontology · 2026-01-31

  articleOpen access

  Detecting heterogeneity in treatment response enriches the interpretation of gerontologic trials. In aging research, estimating the intervention's effect on clinically meaningful outcomes poses analytical challenges when outcomes are truncated by death. For example, in the Whole Systems Demonstrator trial, a large cluster-randomized study evaluating telecare among older adults, the overall effect of the intervention on quality of life was found to be null. However, this marginal intervention estimate obscures potential heterogeneity of individuals responding to the intervention, particularly among those who survive to the end of follow-up. To explore this heterogeneity, we adopt a causal framework grounded in principal stratification, targeting the Survivor Average Causal Effect (SACE)-the treatment effect among "always-survivors," or those who would survive regardless of treatment assignment. We extend this framework using Bayesian Additive Regression Trees (BART), a nonparametric machine learning method, to flexibly model both latent principal strata and stratum-specific potential outcomes. This enables the estimation of the Conditional SACE (CSACE), allowing us to uncover variation in treatment effects across subgroups defined by baseline characteristics. Our analysis reveals that despite the null average effect, some subgroups experience distinct quality of life benefits (or lack thereof) from telecare, highlighting opportunities for more personalized intervention strategies. This study demonstrates how embedding machine learning methods, such as BART, within a principled causal inference framework can offer deeper insights into trial data with complex features including truncation by death and clustering-key considerations in analyzing pragmatic gerontology trials.

  PublisherDOI

• Cardiometabolic Multimorbidity and Dementia Onset Among Middle-Aged and Older Adults: Differences by Race/Ethnicity

  The Journals of Gerontology Series A · 2025-01-14 · 5 citations

  articleOpen access

  BACKGROUND: Racial/ethnic minoritized groups in the United States have a higher prevalence of cardiometabolic multimorbidity and experience a higher risk of dementia. This study evaluates the relationship between cardiometabolic multimorbidity and dementia onset according to racial/ethnic group in a nationally representative cohort of U.S. middle-aged and older adults. METHODS: Data from the Health & Retirement Study (1998-2018, N = 7,960, mean baseline age 59.4 years) and discrete-time survival models were used to estimate differences in the risk of dementia onset, defined by Langa-Weir classification. Models included race/ethnicity (non-Hispanic White, non-Hispanic Black, and Hispanic), chronic disease/multimorbidity categories (no disease, one disease, cardiovascular multimorbidity, metabolic multimorbidity, cardiometabolic multimorbidity, other multimorbidity), age, sex, education, wealth, body-mass index, and proxy status. RESULTS: Over a mean follow-up of 14.6 years, 7.7% of the participants (n = 614) developed dementia. In the fully adjusted model, participants with cardiometabolic multimorbidity had the highest risk of dementia onset (HR:3.27, 95%CI: 2.06, 5.21), followed by metabolic (HR:1.83, 95%CI: 1.14, 2.94), and cardiovascular (HR:1.81, 95%CI: 1.24, 2.64) multimorbidity, relative to participants with no disease. The risk of dementia was significantly greater among Black (HR: 6.40, 95% CI: 3.84, 10.67) and Hispanic participants (HR: 4.90, 95% CI: 2.85, 8.43) with cardiometabolic multimorbidity, compared with White adults with no disease. CONCLUSIONS: Individuals from racial/ethnic minoritized groups have a higher risk of dementia. The risk of dementia onset was significantly greater for Black and Hispanic participants experiencing cardiometabolic multimorbidity, highlighting the value of intervening in cardiometabolic conditions among middle-aged and older adults, in particular, those from racial/ethnic minoritized backgrounds to reduce the risk of developing dementia.

  PublisherOA PDFDOI

• Artificial Intelligence and Clinical Care

  JAMA Internal Medicine · 2025-10-13 · 2 citations

  article
  PublisherDOI

• Multimorbidity patterns and cognitive change among Mexican older adults: findings from the Mexican Health and Aging Study

  Journal of global ageing. · 2025-01-13

  article

  Introduction: Mexico faces unique challenges due to the confluence of population ageing, an increasing burden of chronic conditions and limited resources to address these concerns. Substantial evidence links both independent and co-occurring chronic conditions to the risk of cognitive decline, but the association between common multimorbidity patterns and change in cognitive function has not been examined among older adults in Mexico. Objective: The objective of our research was to identify the most common multimorbidity patterns among ageing Mexican adults in 2012, then examine the association between these multimorbidity patterns and change in cognitive function with advancing age. Methodology: Data from the Mexican Health and Aging Study (2012–21; n = 6,082) was used to identify multimorbidity patterns in 2012 (including hypertension, diabetes, cancer, lung disease, heart attack, stroke, arthritis and depressive symptoms), then latent growth modelling was used to examine associations between the mean cognitive trajectory and the baseline multimorbidity patterns. Results: Several baseline multimorbidity patterns were associated with variation in cognitive function. Diabetes and depressive symptoms were present in the majority of multimorbidity patterns associated with lower cognitive function at baseline, and respondents with either lone diabetes or hypertension+diabetes+arthritis at baseline experienced more rapid cognitive decline than those reporting no conditions at baseline. Conclusion: Our findings suggest that prevention and management of diabetes may reduce cognitive gaps that manifest prior to older adulthood and protect against rapid cognitive decline among older Mexican adults. Further examination of depressive symptoms as a component of multimorbidity and cognitive change is also warranted.

  PublisherDOI

• The story of pain in people with dementia: a rationale for digital measures

  BMC Medicine · 2025-04-17

  articleOpen access

  BACKGROUND: The increasingly older world population presents new aging-related challenges, especially for persons with dementia unable to express their suffering. Pain intensity and the effect of pain treatment are difficult to assess via proxy rating and both under- and overtreatment lead to neuropsychiatric symptoms, inactivity, care-dependency and reduced quality of life. In this debate piece, we provide a rationale on why valid digitalization, sensing technology, and artificial intelligence should be explored to improve the assessment of pain in people with dementia. MAIN TEXT: In dementia care, traditional pain assessment relies on observing the manifestations of typical pain behavior. At the same time, pain treatment is complicated by polypharmacy, potential side effects, and a lack of around-the-clock, timely measures. But proper pain treatment requires objective and accurate measures that capture both the levels of pain and the treatment effects. Sensing systems research for personalized pain assessment is underway, with some promising results regarding associations between physiological signals and pain. Digital phenotyping, making use of everyday sensor data for monitoring health behaviors such as patterns of sleep or movement, has shown potential in clinical trials and for future continuous observation. This emerging approach requires transdisciplinary collaboration between medical and engineering sciences, with user involvement and adherence to ethical practices. CONCLUSION: Digital phenotyping based on physiological parameters and sensing technology may increase pain assessment objectivity in older adults with dementia. This technology must be designed with user involvement and validated; however, it opens possibilities to improve pain relief and care.

  PublisherOA PDFDOI

• The Longitudinal Course of Physical Activity Behavior Poststroke and Its Variation Across Subgroups: The Norwegian Cognitive Impairment After Stroke Study

  Physical Therapy · 2025-05-13 · 1 citations

  articleOpen access

  IMPORTANCE: Despite its importance as a modifiable target poststroke, the longitudinal course of physical activity (PA) is not fully understood. OBJECTIVE: This study aimed to describe the course of poststroke PA behavior from 3 to 36 months and identify subgroups with different PA patterns using multi-trajectory modeling. DESIGN: A prospective multicenter cohort study design was used. SETTING: Follow-up at 3, 18, and 36 months poststroke was community-based. PARTICIPANTS: In total, 277 individuals (age = 70.1 [SD = 10.9]; 116 [41.9%] female) with primarily mild strokes were included. Participants provided at least 2 follow-up periods with accelerometer data each lasting at least 3 consecutive days. MAIN OUTCOMES AND MEASURES: At each follow-up, daily estimates of upright time, time spent in light physical activity (LPA), time spent in moderate physical activity (MPA), step count, and the number of sit-to-stand transitions were measured. RESULTS: Average daily upright time declined by -7.4 min (95% CI = -10.09 to 4.64), and average daily step count declined by -132 steps (95% CI = -176 to -88) each year. Four distinct groups of individuals with different characteristics were identified, following a similar developmental course across PA dimensions over time: one-fourth of the participants (25.6%) were characterized by stable low PA estimates and a tendency to decline over time. Two groups, making up 32.4% and 20.8% of the sample, were characterized by intermediate levels of LPA and MPA, with differing levels of sit-to-stand transitions; and 1 group (21.2% of participants) was characterized by stable high PA duration estimates over time. CONCLUSIONS: The overall course of PA poststroke was characterized by a modest decrease over 3 years. Differing PA trajectory groups characterized by different demographic and clinical features highlight the diverse needs for supporting people living with stroke in becoming more active. RELEVANCE: Findings may help clinicians identify subgroups of people with stroke who need extended professional follow-up in long-term rehabilitation.

  PublisherOA PDFDOI

• Absence of Care Among Community-Living Older Persons with Dementia and Disabilities: A Cross-National Analysis of Population Survey from 22 Countries

  SSRN Electronic Journal · 2025-01-01

  articleOpen access
  PublisherDOI

• Semiparametric joint modeling for biomarker trajectory before disease onset

  Biometrics · 2025-04-02

  article

  Understanding how biomarkers change in relation to disease pathogenesis is a key area in biomedical research. We propose a semiparametric joint model to analyze the temporal evolution of biomarkers prior to the onset of disease. The model allows for a flexible biomarker trajectory that depends on two time scales: a natural time scale such as age and time to disease onset. In practice, the natural time scale often differs from time-on-study, leading to analytical challenges such as left-truncation bias. We introduce a profile kernel estimating equation approach to estimate regression coefficients and unspecified baseline mean trajectory functions. We establish the large-sample properties of the proposed estimators and conduct simulation studies to evaluate their finite-sample performance. Our method is applied to investigate brain biomarker trajectories before the onset of preclinical Alzheimer's disease. We observed a decline in cortical thickness prior to disease onset across brain regions, with APOE4 carriers showing lower levels compared to non-carriers.

  PublisherDOI

• Longitudinal sequencing of cardiometabolic multimorbidity among older adults and association with subsequent dementia onset

  PLoS ONE · 2025-07-10 · 1 citations

  articleOpen access

  BACKGROUND: Patterns of development of cardiometabolic multimorbidity (CMM) and the impact of specific cardiometabolic disease combinations on cognitive function are not well understood. This study utilizes sequence analysis to describe the ordering and timing of cardiometabolic disease accumulation over a five-year period and to assess both sociodemographic predictors and cognitive outcomes of typical cardiometabolic disease sequences. METHODS: We analyzed data from the National Health and Aging Trends Study (2011-2022), including respondents aged ≥65 years without CMM or cognitive impairment at baseline (N = 4956). We used sequence analysis with optimal matching and hierarchical cluster analysis to describe temporal patterns of cardiometabolic disease accumulation and to construct a typology by clustering similar sequences. Sociodemographic predictors of CMM cluster membership were assessed using multinomial logistic regression and discrete time survival analysis was used to examine the association of CMM clusters with subsequent dementia development. RESULTS: 11.8% of respondents developed CMM within 5-years. From a total of 366 distinct cardiometabolic disease sequences, we identified eight cardiometabolic sequence clusters. The first five clusters, "No Cardiometabolic Disease" (N = 2283, 46.1%); "Diabetes Only" (N=642, 13.0%); Heart Disease Only" (N = 297, 6.0%); "MI Only" (N = 145, 2.9%); "Stroke Only" (N = 132, 2.7%), were composed of persons who did not develop CMM over the observation period. The sixth cluster, "Incident CVD with Multimorbidity" (N = 656, 13.2%), was largely composed of persons with no conditions at baseline who developed incident cardiometabolic disease and/or CMM during the observation period (N = 477, 72.7%) and the seventh cluster, "Diabetes Multimorbidity" (N = 333, 6.7%), primarily consisted of persons with diabetes who developed incident CMM. Finally, the eight cluster (N = 468, 9.4%) was characterized by mortality early in the observation period with minimal CMM development during the observation period. Black and Hispanic race/ethnicity, lower wealth, and obesity were associated with increased likelihood of membership in one or both of the clusters characterized by CMM development. We observed increased dementia risk among persons in the Incident CVD with Multimorbidity cluster (HR = 1.32, 95% CI = 1.04-1.67) and the Diabetes MM cluster (HR = 1.88, 95% CI = 1.44,2.44). CONCLUSIONS: Development of cardiometabolic multimorbidity is more likely among minoritized and/or low-income older adults and is associated with increased risk of subsequent dementia. Targeted approaches to cardiometabolic disease prevention and risk reduction may be an effective means of slowing or preventing the onset of cognitive decline among these groups.

  PublisherDOI

• Informal Care Receiving Among Older Adults: The Role of Multimorbidity and Intersectional Social Position

  The Gerontologist · 2025-02-27 · 2 citations

  articleOpen access

  BACKGROUND AND OBJECTIVES: Older adults with distinct multimorbidity combinations may require varying intensities of informal care and these needs may vary in important ways by race/ethnicity and sex. This study aims to examine informal care-receiving characteristics among older adults with varying multimorbidity patterns and race/ethnicity-sex characteristics. RESEARCH DESIGN AND METHODS: A total of 4,875 participants from the National Health and Aging Trends Study were included. Five clinically informed multimorbidity categories (no condition(s), somatic-only, depression without cognitive impairment, cognitive impairment without depression, both depression, and cognitive impairment) and 6 intersectional groups (non-Hispanic White female, non-Hispanic White male, non-Hispanic Black female, non-Hispanic Black male, Hispanic female, and Hispanic male) were assessed. Negative binomial regression was applied to explore the associations among multimorbidity groups, race/ethnicity and sex combinations, and informal care-receiving characteristics. RESULTS: Compared with the somatic-only multimorbidity group, individuals with no condition(s) received assistance with fewer ADL/IADL activities, whereas the cognitive impairment multimorbidity group received assistance with more ADL/IADL activities. Across race/ethnicity-sex groups, non-Hispanic White and Black males received assistance with fewer ADL/IADL activities, and no statistical significance was found for non-Hispanic Black females or Hispanics compared with non-Hispanic White female counterparts. Sensitivity analysis examining assistance with only ADL activities revealed the same pattern. DISCUSSION AND IMPLICATIONS: The study highlighted the complexities of informal care-receiving characteristics among older adults, particularly among those with cognitive difficulties, and the ways in which race/ethnicity and sex are associated with care-receiving patterns. The findings highlight a need for person- and family-centered interventions sensitive to the diverse needs of care-recipients and caregivers.

  PublisherOA PDFDOI

Recent grants

• Ethics and Regulation Core (C)

  NIH · $68.6M · 2019–2030

• NIH Grant P50AG047270

  NIH · $16.6M · 2021

• Administrative Core

  NIH · $42.1M · 2025–2030

• NIH Grant R21AG045148

  NIH · $458k · 2016

• NIH Grant R01AG047891

  NIH · $990k · 2020

Frequent coauthors

• Thomas M. Gill

  Yale University

  190 shared

• Anne B. Newman

  University of Pittsburgh

  89 shared

• Michael E. Miller

  85 shared

• Roger A. Fielding

  Tufts University

  83 shared

• ­Abby C. King

  Stanford University

  83 shared

• Jack M. Guralnik

  University of Maryland, Baltimore

  83 shared

• Anthony P. Marsh

  Southeast Louisiana Veterans Health Care System

  82 shared

• Marco Pahor

  University of Florida

  82 shared

Education

• PhD

  Cornell University

  1996

Awards & honors

• Gerontological Biostatistics (field founded by Heather Allor…