Adlin Pinheiro
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Adlin Pinheiro has a total of 37 co-authors. Here are the top co-authors by number of shared publications:
- Charles DeCarli
Department of Neurology, University of California, Davis
2 shared publications - José Rafael Romero
Department of Neurology, Boston University Chobanian & Avedisian School of Medicine, MA (J.R.R.).
2 shared publications - Oluchi Ekenze
Boston University Chobanian & Avedisian School of Medicine
2 shared publications - Serkalem Demissie
Department of Biostatistics, Boston University School of Public Health, Boston, MA
2 shared publications - Sudha Seshadri
Boston University and National Heart, Lung, and Blood Institute’s Framingham Study, MA (A.B., A.S.B., E.J.B., R.S.V., S.S.).
2 shared publications - Alexa S. Beiser
Boston University School of Public Health
1 shared publication
Latest Publications
- Open Access
- 23 December 2025
- Published by Wiley in Alzheimer's & Dementia
- Vol. 21 (S6) , e098476
Predicting cognitive impairment using novel functional features of spatial proximity and circularity in the digital clock drawing test
Background: The digital clock drawing test (dCDT) is a cognitive screening tool employing a digital pen to capture high‐resolution pen movements. Traditional dCDT approaches to predict cognitive outcomes often rely on many summary features (e.g. time to completion, mean pressure, clock face area, etc.) which involve subjective decisions such as feature selection and imputation of missing data. To address these limitations, we introduce novel dCDT features, expressed as mathematical functions. These functional features include the G‐function (measuring spatial proximity of points), pressure density function (measuring variability in pressure exerted on the paper), and radius function (measuring circularity of the clock face). We compare the performance of these functional features to 45 commonly used summary features. Methods: We included dCDTs from 3,415 stroke‐free participants from the Framingham Heart Study with at least one command or copy task. Random forest models with five‐fold cross‐validation were trained to distinguish participants with MCI or dementia from cognitively intact participants. Predictors included combinations of demographics, time‐based features, summary features, and functional features. Functional features underwent dimension reduction using functional principal components analysis. Area under the receiver operating curve (AUC), sensitivity, and specificity assessed model performance. Results: Average age was 63.6 years (SD: 13.9), with 1,546 (45%) males, 785 (23%) having at least one APOE‐e4 allele, and 1,957 (57%) having a college education. At the time of the dCDT, 122 (4%) had mild cognitive impairment and 74 (2%) had dementia. When combined with demographics and time‐based features, functional features from command and copy tests related to spatial proximity (AUC: 0.90; 95% CI: 0.88‐0.92) and circularity (AUC: 0.91; 95% CI: 0.89‐0.93) were as predictive as summary features (AUC: 0.90; 95% CI: 0.88‐0.92). Similar results were found when analyzing command and copy features separately. Conclusions: Functional features of the dCDT offer several advantages such as robustness to missing data, decreased subjectivity in feature selection, and avoidance of reducing complex information into summary statistics. In addition, there is potential to make inferences about cognition based on the shape and behavior of the functions. Further research is required to better understand what these functional features reveal about cognition.
- 23 December 2025
- Published by Wiley in Alzheimer's & Dementia
- Vol. 21 (S6) , e098476
Open AccessPredicting cognitive impairment using novel functional features of spatial proximity and circularity in the digital clock drawing test
Background: The digital clock drawing test (dCDT) is a cognitive screening tool employing a digital pen to capture high‐resolution pen movements. Traditional dCDT approaches to predict cognitive outcomes often rely on many summary features (e.g. time to completion, mean pressure, clock face area, etc.) which involve subjective decisions such as feature selection and imputation of missing data. To address these limitations, we introduce novel dCDT features, expressed as mathematical functions. These functional features include the G‐function (measuring spatial proximity of points), pressure density function (measuring variability in pressure exerted on the paper), and radius function (measuring circularity of the clock face). We compare the performance of these functional features to 45 commonly used summary features. Methods: We included dCDTs from 3,415 stroke‐free participants from the Framingham Heart Study with at least one command or copy task. Random forest models with five‐fold cross‐validation were trained to distinguish participants with MCI or dementia from cognitively intact participants. Predictors included combinations of demographics, time‐based features, summary features, and functional features. Functional features underwent dimension reduction using functional principal components analysis. Area under the receiver operating curve (AUC), sensitivity, and specificity assessed model performance. Results: Average age was 63.6 years (SD: 13.9), with 1,546 (45%) males, 785 (23%) having at least one APOE‐e4 allele, and 1,957 (57%) having a college education. At the time of the dCDT, 122 (4%) had mild cognitive impairment and 74 (2%) had dementia. When combined with demographics and time‐based features, functional features from command and copy tests related to spatial proximity (AUC: 0.90; 95% CI: 0.88‐0.92) and circularity (AUC: 0.91; 95% CI: 0.89‐0.93) were as predictive as summary features (AUC: 0.90; 95% CI: 0.88‐0.92). Similar results were found when analyzing command and copy features separately. Conclusions: Functional features of the dCDT offer several advantages such as robustness to missing data, decreased subjectivity in feature selection, and avoidance of reducing complex information into summary statistics. In addition, there is potential to make inferences about cognition based on the shape and behavior of the functions. Further research is required to better understand what these functional features reveal about cognition.
- Open Access
- 29 July 2025
- Published by MDPI AG in Brain Sciences
- Vol. 15 (8) , 813
Prevalent Cardiovascular Disease and Atrial Fibrillation in Relation to Cerebral Small Vessel Disease Burden
Background/Objectives: Cardiovascular disease (CVD) contributes to stroke and dementia. Individuals with CVD have high risk for adverse cognitive outcomes and stroke, possibly due to shared risk factors between CVD, stroke, and dementia, which may be attributed to cerebral small vessel disease (CSVD). We aim to determine the association between prevalent CVD and atrial fibrillation (AF) with CSVD. Methods: Composite of CVD [coronary heart disease, heart failure (HF)], its individual components, and AF were assessed. Multi-marker CSVD score was used to reflect increasing CSVD burden (cerebral microbleeds (CMBs), high-burden perivascular spaces, extensive white matter hyperintensity, cortical superficial siderosis, or covert brain infarcts were assigned 1 point each, with a range of 0–5). We related prevalent CVD, its individual components, and AF to multi-marker CSVD score and individual CSVD markers using logistic regression analyses adjusted for age, sex, FHS cohort, time between MRI and clinic exam (model-1), and vascular risk factors (model-2). Results: In 3413 participants (mean age: 59 ± 14 years, 53.4% women), 11% had prevalent CVD or AF, 8% had prevalent CVD, and 4% had prevalent AF. One CSVD marker was seen in 23% participants, and 9% had ≥ 2 markers. In multivariable-adjusted analyses, composite prevalent CVD and AF was associated with the presence of one CSVD marker (OR: 1.38, 95% confidence interval [CI]: 1.05–1.84). The association with ≥2 CSVD markers was not significant. Only CMBs were associated with components of CVD and AF, with the highest odds of association with HF. Conclusions: Prevalent CVD (including AF) is associated with the presence of CSVD, with all components associated with CMBs.
- 29 July 2025
- Published by MDPI AG in Brain Sciences
- Vol. 15 (8) , 813
Open AccessPrevalent Cardiovascular Disease and Atrial Fibrillation in Relation to Cerebral Small Vessel Disease Burden
Background/Objectives: Cardiovascular disease (CVD) contributes to stroke and dementia. Individuals with CVD have high risk for adverse cognitive outcomes and stroke, possibly due to shared risk factors between CVD, stroke, and dementia, which may be attributed to cerebral small vessel disease (CSVD). We aim to determine the association between prevalent CVD and atrial fibrillation (AF) with CSVD. Methods: Composite of CVD [coronary heart disease, heart failure (HF)], its individual components, and AF were assessed. Multi-marker CSVD score was used to reflect increasing CSVD burden (cerebral microbleeds (CMBs), high-burden perivascular spaces, extensive white matter hyperintensity, cortical superficial siderosis, or covert brain infarcts were assigned 1 point each, with a range of 0–5). We related prevalent CVD, its individual components, and AF to multi-marker CSVD score and individual CSVD markers using logistic regression analyses adjusted for age, sex, FHS cohort, time between MRI and clinic exam (model-1), and vascular risk factors (model-2). Results: In 3413 participants (mean age: 59 ± 14 years, 53.4% women), 11% had prevalent CVD or AF, 8% had prevalent CVD, and 4% had prevalent AF. One CSVD marker was seen in 23% participants, and 9% had ≥ 2 markers. In multivariable-adjusted analyses, composite prevalent CVD and AF was associated with the presence of one CSVD marker (OR: 1.38, 95% confidence interval [CI]: 1.05–1.84). The association with ≥2 CSVD markers was not significant. Only CMBs were associated with components of CVD and AF, with the highest odds of association with HF. Conclusions: Prevalent CVD (including AF) is associated with the presence of CSVD, with all components associated with CMBs.
- Open Access
- 14 September 2023
- Published by MDPI AG in Brain Sciences
- Vol. 13 (9) , 1323
Relation of MRI-Visible Perivascular Spaces and Other MRI Markers of Cerebral Small Vessel Disease
Perivascular spaces (PVS) visible on brain MRI signal cerebral small vessel disease (CSVD). The coexistence of PVS with other CSVD manifestations likely increases the risk of adverse neurological outcomes. We related PVS to other CSVD manifestations and brain volumes that are markers of vascular brain injury and neurodegeneration. Framingham Heart Study (FHS) participants with CSVD ratings on brain MRI were included. PVS were rated in the basal ganglia (BG) and centrum semiovale (CSO) into grades I–IV and a category reflecting high burden in single or mixed CSO-BG regions. We related PVS to covert brain infarcts (CBI), white matter hyperintensities (WMH), cerebral microbleeds (CMB), total brain, hippocampal, and cortical gray matter volumes using adjusted multivariable regression analyses. In 2454 participants (mean age 54 ± 12 years), we observed that higher PVS burden in both BG and CSO was related to CMB in lobar and deep brain regions and increased WMH. Greater CSO PVS burden was associated with decreased total cortical gray volumes. PVS are associated with ischemic markers of CSVD and neurodegeneration markers. Further studies should elucidate the causality between PVS and other CSVD manifestations.
- 14 September 2023
- Published by MDPI AG in Brain Sciences
- Vol. 13 (9) , 1323
Open AccessRelation of MRI-Visible Perivascular Spaces and Other MRI Markers of Cerebral Small Vessel Disease
Perivascular spaces (PVS) visible on brain MRI signal cerebral small vessel disease (CSVD). The coexistence of PVS with other CSVD manifestations likely increases the risk of adverse neurological outcomes. We related PVS to other CSVD manifestations and brain volumes that are markers of vascular brain injury and neurodegeneration. Framingham Heart Study (FHS) participants with CSVD ratings on brain MRI were included. PVS were rated in the basal ganglia (BG) and centrum semiovale (CSO) into grades I–IV and a category reflecting high burden in single or mixed CSO-BG regions. We related PVS to covert brain infarcts (CBI), white matter hyperintensities (WMH), cerebral microbleeds (CMB), total brain, hippocampal, and cortical gray matter volumes using adjusted multivariable regression analyses. In 2454 participants (mean age 54 ± 12 years), we observed that higher PVS burden in both BG and CSO was related to CMB in lobar and deep brain regions and increased WMH. Greater CSO PVS burden was associated with decreased total cortical gray volumes. PVS are associated with ischemic markers of CSVD and neurodegeneration markers. Further studies should elucidate the causality between PVS and other CSVD manifestations.
- Open Access
- 17 January 2023
- Published by Springer Nature in Journal of General Internal Medicine
- Vol. 38 (9) , 2113-2122
Yoga Versus Education for Veterans with Chronic Low Back Pain: a Randomized Controlled Trial
Background Yoga is effective for chronic low back pain (cLBP) in civilians but understudied among Veterans. Objective Determine whether yoga is more effective than an educational book for improving disability and pain among Veterans with cLBP. Design, Setting, and Participants Veterans diagnosed with cLBP at a VA medical center enrolled in a randomized controlled trial from March to December of 2015. Interventions Twelve weekly hatha yoga classes or education using The Back Pain Helpbook. Measures Co-primary outcomes were changes from baseline at 12 weeks in back-related disability on the modified Roland Morris Disability Questionnaire and pain on the Defense & Veterans Pain Rating Scale. Secondary outcomes were global improvement, patient satisfaction, pain medication use, and post-traumatic stress symptoms. An intention-to-treat approach was used in primary analyses. Results One hundred twenty Veterans (mean age, 55.5 [SD = 16.9]; 11 [9%] women; mean number of chronic conditions, 5.5) were randomized to yoga (n = 62) and education (n = 58). At 12 weeks, reductions in back-related disability in yoga (mean difference [MD] = − 3.50, 95% CI: − 5.03, − 1.97) were not significantly different than education (MD = − 2.55, 95% CI: − 4.10, − 0.99; between-group difference: − 0.95 [95% CI: − 3.14, 1.23], p = 0.39). For pain, there was no significant difference between yoga (MD = − 1.01, 95% CI: − 1.67, − 0.35) and education (MD = − 0.81, 95% CI: − 1.36, − 0.27; between-group difference: − 0.20, 95% CI: − 1.06, 0.66, p = 0.65). More yoga than education participants reported being very much or extremely improved (39% vs 19%, OR = 3.71, 95% CI: 1.37, 10.02, p = 0.01) and very satisfied with treatment (60% vs 31%, OR = 4.28, 95% CI: 1.70, 10.77, p = 0.002). No differences in pain medication use or post-traumatic stress symptoms were observed at 12 weeks. No serious adverse events were reported in either group. Conclusion Twelve weekly yoga classes were not more effective than an education intervention for improving pain or disability outcomes among mostly older male Veterans with cLBP and multiple comorbid health conditions. ClinicalTrials.gov Identifier NCT02224183
- 17 January 2023
- Published by Springer Nature in Journal of General Internal Medicine
- Vol. 38 (9) , 2113-2122
Open AccessYoga Versus Education for Veterans with Chronic Low Back Pain: a Randomized Controlled Trial
Background Yoga is effective for chronic low back pain (cLBP) in civilians but understudied among Veterans. Objective Determine whether yoga is more effective than an educational book for improving disability and pain among Veterans with cLBP. Design, Setting, and Participants Veterans diagnosed with cLBP at a VA medical center enrolled in a randomized controlled trial from March to December of 2015. Interventions Twelve weekly hatha yoga classes or education using The Back Pain Helpbook. Measures Co-primary outcomes were changes from baseline at 12 weeks in back-related disability on the modified Roland Morris Disability Questionnaire and pain on the Defense & Veterans Pain Rating Scale. Secondary outcomes were global improvement, patient satisfaction, pain medication use, and post-traumatic stress symptoms. An intention-to-treat approach was used in primary analyses. Results One hundred twenty Veterans (mean age, 55.5 [SD = 16.9]; 11 [9%] women; mean number of chronic conditions, 5.5) were randomized to yoga (n = 62) and education (n = 58). At 12 weeks, reductions in back-related disability in yoga (mean difference [MD] = − 3.50, 95% CI: − 5.03, − 1.97) were not significantly different than education (MD = − 2.55, 95% CI: − 4.10, − 0.99; between-group difference: − 0.95 [95% CI: − 3.14, 1.23], p = 0.39). For pain, there was no significant difference between yoga (MD = − 1.01, 95% CI: − 1.67, − 0.35) and education (MD = − 0.81, 95% CI: − 1.36, − 0.27; between-group difference: − 0.20, 95% CI: − 1.06, 0.66, p = 0.65). More yoga than education participants reported being very much or extremely improved (39% vs 19%, OR = 3.71, 95% CI: 1.37, 10.02, p = 0.01) and very satisfied with treatment (60% vs 31%, OR = 4.28, 95% CI: 1.70, 10.77, p = 0.002). No differences in pain medication use or post-traumatic stress symptoms were observed at 12 weeks. No serious adverse events were reported in either group. Conclusion Twelve weekly yoga classes were not more effective than an education intervention for improving pain or disability outcomes among mostly older male Veterans with cLBP and multiple comorbid health conditions. ClinicalTrials.gov Identifier NCT02224183