Brain age prediction based on imaging data and machine learning (ML) methods has great potential to provide insights into the development of cognition and mental disorders. Though different ML models have been proposed, a systematic comparison of ML models in combination with imaging features derived from different modalities is still needed. In this study, we evaluate the prediction performance of 36 combinations of imaging features and ML models including deep learning. We utilize single and multimodal brain imaging data including MRI, DTI, and rs‐fMRI from a large data set with 839 subjects. Our study is a follow‐up to the initial work (Liang et al., 2019. Human Brain Mapping) to investigate different analytic strategies to combine data from MRI, DTI, and rs‐fMRI with the goal to improve brain age prediction accuracy. Additionally, the traditional approach to predicting the brain age gap has been shown to have a systematic bias. The potential nonlinear relationship between the brain age gap and chronological age has not been thoroughly tested. Here we propose a new method to correct the systematic bias of brain age gap by taking gender, chronological age, and their interactions into consideration. As the true brain age is unknown and may deviate from chronological age, we further examine whether various levels of behavioral performance across subjects predict their brain age estimated from neuroimaging data. This is an important step to quantify the practical implication of brain age prediction. Our findings are helpful to advance the practice of optimizing different analytic methodologies in brain age prediction. 相似文献
AimsDiabetes related distress (DRD) is a negative emotional reaction to stresses associated with diabetes mellitus (DM) and its management. This study estimated the burden of DRD and self-reported adherence to treatment (SRAT) among patients with DM and investigated their relationship with glycemic control.MethodsA cross sectional study of consented 157 diabetics was conducted using the17-item Diabetes Distress Scale (DDS). It measures distress at four subscales: Emotional Burden (EB), Physician-related (PD), Regimen-related (RD) and Interpersonal Distress (ID). SRAT was assessed using Morisky’s scale. Glycemic control was assessed using the most recent HbA1c results. Multivariable linear regression analysis was used for adjustment of confounders and bootstrap Confidence Interval was used to test for the occurrence of mediating effect.ResultsAverage age was 44.5 ± 16.0 years, 65% were females, 79% had type 2 DM and nearly 55% has had DM for more than 7 years and the average HbA1c was 8.9 ± 2.2%. Clinically significant DRD was reported by 37% of the participants, EB and RD in 40.8%, PD in 46.5%, and ID among 32.5%. Younger patients showed higher level of stress compared to older participants and patients with type 1 DM showed higher level of stress in all DRD domains. Only 46% of patients were defined as having satisfactory SRAT and improvement of SRAT significantly enhanced the glycemic control (r = ?0.32, p < 0.01). DRD and low SRAT negatively correlated with HbA1c; increasing the DRD by one point may increase the HbA1c on average by 0.41 (C.I. 0.02–0.80) and will indirectly raise the HbA1c by 0.24 (C.I. 0.04–0.47) through the mediating effect of low SRAT.ConclusionDRD and low SRAT are commonly reported among DM patients and both are indirectly correlated. The mediating effect of low SRAT highlights the clinical role of DRD and clarifies the process by which distress affect the outcome of DM management. 相似文献
BackgroundGastric cancer (GC) is a leading cause of cancer-related death in the world and most patients have advanced disease upon presentation. The effect of age on prognosis in GC is controversial. We aimed to determine the impact of age on survival in patients with GC.MethodsThis was a retrospective study of the medical records of Lebanese patients diagnosed with GC at the American University of Beirut Medical Center (AUBMC) between 2005 and 2014. Patients were divided into young (<65 years) and older groups (≥65 years). A multivariate analysis was done to determine the independent predictors of survival. Kaplan-Meier method was used for analysis of long-term survival outcomes.ResultsThe sample consisted of 156 patients. The mean age was 62.15 (SD 13.54). Most patients presented with stage 4 disease (62.2%) and poorly differentiated histology (66.4%). The most common symptoms were abdominal pain and weight loss. On bivariate analysis, advanced stage (P=0.02) and higher grade (P=0.04) were associated with increased mortality. Patients <65 years of age were significantly more likely to have poorly differentiated tumours, while patients ≥65 years had more comorbidities (P=0.001). The 5-year DFS were 35% and 37% for patients <65 years of age and ≥65 years of age, respectively (P=0.15).ConclusionsHigher grade and advanced stage are associated with worse survival in patients with GC, but age did not seem to have an impact. Screening high risk patients and early diagnosis are necessary to improve survival. 相似文献
ObjectiveDegenerative thoracic stenosis has been shown to most frequently involve the lower thoracic segments (T9-T12) where there is greater mobility and vulnerability due to flexion, extension and rotation of the spine. The thoracolumbar junction is considered anatomically to be T12-L1; the anatomical transition between the relatively immobile thoracic spine and relatively mobile lumbar spine. From anecdotal experience at our institution, we hypothesise that the true thoracolumbar junction is higher, at T10-11; the point of transition from floating to false ribs resulting in increased mobility at T10-11.Methods and materialsA retrospective review was performed of MRI lumbar and whole spine performed on patients aged 10–40 years in our institution over a 5-year period. Patients with previous surgery, chronic spinal disorders and congenital abnormalities were excluded from the study. Intervertebral discs from T8-9 to L1-2 were assessed for evidence of degeneration using the Pfirrmann grading system. Data obtained from a study using computer-based models to assess mean resultant loads in flexion, sitting and standing from T8-9 to L1-2 on patients aged 18–35 years was also analysed. The mean load gradients between two consecutive discs from T8 to L2 were analysed. Statistical analysis was performed with SPSS (p < 0.05 was considered statistically significant).ResultsThree-hundred and twenty-two MRI studies were reviewed. Mean Pfirrmann grade was highest at T8-9 and T9-10 (1.35 ± 0.99 and 1.32 ± 0.93 respectively).Pfirrmann grade differed significantly at each level (χ2 = 45.137 p = 0.001). Difference in mean load gradient from T9 to T11 was significantly higher than mean load gradient across T11 to L1 in both sitting and standing (0.095 ± 0.062 vs 0.050 ± 0.044 kN; p = 0.007, and 0.101 ± 0.061 kN vs 0.040 ± 0.054 kN; p = 0.007).ConclusionThe changes in segmental loads and more severe disc degeneration at T9-11 compared to T11-L1 support our hypothesis that the true thoracolumbar transition is higher than expected, at T10-11; where the rib cage transitions from floating to false ribs. 相似文献
AbstractObservational and interventional studies have unequivocally demonstrated that “present”, i.e. single-occasion, blood pressure is one of the key determinants of cardiovascular disease risk. Over the past two decades, however, numerous publications have suggested that longitudinal blood pressure data and assessment of long-term blood pressure exposure provide incremental prognostic value over present blood pressure. These studies have used several different indices to quantify the overall exposure to blood pressure, such as time-averaged blood pressure, cumulative blood pressure, blood pressure trajectory patterns, and age of hypertension onset. This review summarises existing research on the association between these indices and hard cardiovascular outcomes, outlines the strengths and weaknesses of these indices, and provides an overview of how longitudinal blood pressure changes can be measured and used to improve cardiovascular disease risk prediction.
KEY MESSAGES
Numerous recent publications have examined the relation between cardiovascular disease and long-term blood pressure (BP) exposure, quantified using indices such as time-averaged BP, cumulative BP, BP trajectory patterns, and age of hypertension onset.
This review summarises existing research on the association between these indices and hard cardiovascular outcomes, outlines the strengths and weaknesses of these indices, and provides an overview of how longitudinal BP changes can be measured and used to improve cardiovascular disease risk prediction.
Although longitudinal BP indices seem to predict cardiovascular outcomes better than present BP, there are considerable differences in the clinical feasibility of these indices along with a limited number of prospective data.