Development and external validation of a novel dementia risk prediction score in the UK Biobank cohort

Abstract

Background: Globally, up to 40% of dementia cases may be prevented if several key risk factors, such as low education and obesity, are targeted. This has motivated interest into the development of risk scores that aim to quantify an individual's risk of developing dementia within a given time frame. However, translation to a clinical setting has been hampered by either a lack of external validation, poor out-of-sample performance, or the integration of measures (e.g., MRI, cognitive testing) that are costly or time intensive to administer.This study aimed to develop a novel dementia risk score suitable for a primary care setting and compare its discrimination accuracy with other risk scores. Method: 208,108 UK Biobank (UKB) participants were examined, with cases of incident dementia (mean time to diagnosis ± SD = 5.91 years ± 2.02) identified through self-report, hospital inpatient records and death certificates (n = 1,270, 0.6%). The dataset was randomly split into a training (80%) and test set (20%). Potential predictors (n = 30) included Apolipoprotein E4 (APOE4) status and cardiovascular, medical and lifestyle factors collected at baseline. Logistic LASSO regression was employed for variable selection and logistic regression was then used to derive beta coefficients for each predictor included in the UKB dementia risk score (UKB-DRS). The area under the curve (i.e., AUC) was used to compare the discriminative ability of the UKB-DRS to the CAIDE, ANU-ADRI, Dementia Risk Score (i.e., DRS) and Framingham risk score, in both the test set and the Whitehall II study (WHIII, n = 2,974), which served as an external dataset. Result: The UKB-DRS consisted of age, sex, education, APOE4 status, a medical history of diabetes, stroke and depression and a family history of dementia. The UKB-DRS demonstrated good discrimination accuracy in both the test set (AUC [95% ] = 0.79 [0.77, 0.82]) and external dataset (AUC [95% CI] = 0.83 [0.79,0.87]). Additionally, the UKB-DRS outperformed all other risk scores (the DRS had the next best AUC [95% CI]: UKB = 0.77 [0.74, 0.8]; WHII = 0.77 [0.73, 0.81]). Conclusion:This study contributes a novel dementia risk score that may be used to guide primary prevention strategies.

Learning the natural history of human disease with generative transformers

“…Delphi-2M’s disease predictions on the longitudinal validation data in the UK Biobank are comparable to clinical risk models like the Framingham 7 risk score for cardiovascular diseases, the UKBDRS 27 risk score for dementia and the Charlson 28 comorbidity index for mortality ( Figure 2d ). Delphi-2M’s performance also appears stable when evaluated over varying prediction horizons ranging from 1-24 months, albeit with slightly decreasing performance over longer time periods ( Extended Data Figure 3c ).…”

Feasibility of multiorgan risk prediction with routinely collected diagnostics: a prospective cohort study in the UK Biobank

“…Diagnoses and dates for the 10 disease outcomes (itemised above) were collated across multiple UK Biobank sources including self-report, linked hospital and primary care records and deaths, using published code lists where available 35–37. Incident outcomes were defined by first occurrence of disease after baseline recruitment.…”

“…For incident stroke risk we considered QStroke38 and CHA2DS2-VASc,39 a score comprising congestive heart failure, hypertension, age, diabetes, prior stroke or transient ischaemic attack, vascular disease and sex. For all-cause dementia, we included three dementia risk scores; one developed using the CAIDE study (Cardiovascular Risk Factors, Aging and Dementia),40 the Lifestyle for Brain Health score (LIBRA)41 and the recently developed UK Biobank Dementia Risk Score (UKB-DRS) 37. For myocardial infarction and heart failure, we considered Framingham Risk Score (with and without blood lipids),11 the Pooled Cohort Equations to Prevent Heart Failure (PCP-HF risk score 42) and QRISK3 10.…”

“…Extending the health information framework described above, we categorise any risk score that is comprised only of inputs that can be collected remotely as a remote risk score. UKB-DRS37 (dementia) and Framingham risk score (using body mass index, BMI)11 are examples of this. With the most restricted set of inputs, we would expect these models to be the least powerful and have the lowest predictive performance.…”

CogDrisk, ANU-ADRI, CAIDE, and LIBRA Risk Scores for Estimating Dementia Risk

“… 6 , 7 , 8 , 9 , 10 Perhaps because these tools differ in their construction and the methodology used in their development, direct comparisons have thus far been limited. While a few studies have compared risk tools in a single cohort, 11 , 12 , 13 to our knowledge, there has been no thorough comparison of the tools across multiple independent cohorts. In order to facilitate the implementation of risk scores in primary care and public health practice, we need to understand what each tool requires to allow calculation of a full score, and how well each tool performs in estimating dementia or AD risks in a range of populations.…”