Development of a novel dementia risk prediction model in the general population: A large, longitudinal, population-based machine-learning study

Predicting Alzheimer's disease from cognitive footprints in mid and late life: How much can register data and machine learning help?

BACKGROUND

Real-world data with decades-long medical records are increasingly available alongside the growing adoption of machine learning in healthcare research. We evaluated the performance of machine learning models in predicting the risk of Alzheimer's disease (AD) using data from the Finnish national registers.

METHODS

We conducted a case-control study using data from the Finnish MEDALZ (Medication use and Alzheimer's disease) study. Altogether 56,741 individuals with incident AD diagnosis (age ≥ 65 years at diagnosis and born after 1922) and their 1:1 age-, sex-, and region of residence-matched controls were included. The association of risk factors, evaluated at different age periods (45-54, 55-64, 65+), and AD were assessed with logistic regression. Predictive accuracies of logistic regressions were compared with seven machine learning models (L1-regularized logistic regression, Naive bayes, Decision tree, Random Forest, Multilayer perceptron, XGBoost, and LightGBM).

FINDINGS

63.5 % of cases and controls were females and the mean age was 79.1 (SD = 5.1). The strongest associations with AD were observed for head injuries at age 55-64 (OR, 95 % CI 1.33, 1.19-1.48) and 65+ (1.31, 1.23-1.40), followed by antidepressant use (1.30, 1.22-1.38) at 55-64 and antipsychotic use (1.27, 1.19-1.35) at 65+. The predictive accuracies of all models were low, with the best performance (AUC 0.603) observed in Random Forest for predicting AD onset at age 65-69.

INTERPRETATION

Although significant associations were identified between many risk factors and AD, the low predictive accuracies suggest that specialised healthcare diagnosis data is not sufficient for predicting AD and linkage with other data sources is needed.

Incremental Value of Multidomain Risk Factors for Dementia Prediction: A Machine Learning Approach

OBJECTIVE

The current evidence regarding how different predictor domains contributes to predicting incident dementia remains unclear. This study aims to assess the incremental value of five predictor domains when added to a simple dementia risk prediction model (DRPM) for predicting incident dementia in older adults.

DESIGN

Population-based, prospective cohort study.

SETTING

UK Biobank study.

PARTICIPANTS

Individuals aged 60 or older without dementia.

MEASUREMENTS

Fifty-five dementia-related predictors were gathered and categorized into clinical and medical history, questionnaire, cognition, polygenetic risk, and neuroimaging domains. Incident dementia (all-cause) and the subtypes, Alzheimer's disease (AD) and vascular dementia (VaD), were determined through hospital and death registries. Ensemble machine learning (ML) DRPMs were employed for prediction. The incremental values of risk predictors were assessed using the percent change in Area Under the Curve (∆AUC%) and the net reclassification index (NRI).

RESULTS

The simple DRPM which included age, body mass index, sex, education, diabetes, hyperlipidaemia, hypertension, depression, smoking, and alcohol consumption yielded an AUC of 0.711 (± 0.008 SD). The five predictor domains exhibited varying levels of incremental value over the basic model when predicting all-cause dementia and the two subtypes. Neuroimaging markers provided the highest incremental value in predicting all-cause dementia (∆AUC% +9.6%) and AD (∆AUC% +16.5%) while clinical and medical history data performed the best at predicting VaD (∆AUC% +12.2%). Combining clinical and medical history, and questionnaire data synergistically enhanced ML DRPM performance.

CONCLUSION

Combining predictors from different domains generally results in better predictive performance. Selecting predictors involves trade-offs, and while neuroimaging markers can significantly enhance predictive accuracy, they may pose challenges in terms of cost or accessibility.

Uveal melanoma distant metastasis prediction system: A retrospective observational study based on machine learning

Uveal melanoma (UM) patients face a significant risk of distant metastasis, closely tied to a poor prognosis. Despite this, there is a dearth of research utilizing big data to predict UM distant metastasis. This study leveraged machine learning methods on the Surveillance, Epidemiology, and End Results (SEER) database to forecast the risk probability of distant metastasis. Therefore, the information on UM patients from the SEER database (2000-2020) was split into a 7:3 ratio training set and an internal test set based on distant metastasis presence. Univariate and multivariate logistic regression analyses assessed distant metastasis risk factors. Six machine learning methods constructed a predictive model post-feature variable selection. The model evaluation identified the multilayer perceptron (MLP) as optimal. Shapley additive explanations (SHAP) interpreted the chosen model. A web-based calculator personalized risk probabilities for UM patients. The results show that nine feature variables contributed to the machine learning model. The MLP model demonstrated superior predictive accuracy (Precision = 0.788; ROC AUC = 0.876; PR AUC = 0.788). Grade recode, age, primary site, time from diagnosis to treatment initiation, and total number of malignant tumors were identified as distant metastasis risk factors. Diagnostic method, laterality, rural-urban continuum code, and radiation recode emerged as protective factors. The developed web calculator utilizes the MLP model for personalized risk assessments. In conclusion, the MLP machine learning model emerges as the optimal tool for predicting distant metastasis in UM patients. This model facilitates personalized risk assessments, empowering early and tailored treatment strategies.

Forecasting Myopic Maculopathy Risk Over a Decade: Development and Validation of an Interpretable Machine Learning Algorithm

Purpose

The purpose of this study was to develop and validate prediction model for myopic macular degeneration (MMD) progression in patients with high myopia.

Methods

The Zhongshan High Myopia Cohort for model development included 660 patients aged 7 to 70 years with a bilateral sphere of ≤-6.00 diopters (D). Two hundred twelve participants with an axial length (AL) ≥25.5 mm from the Chinese Ocular Imaging Project were used for external validation. Thirty-four clinical variables, including demographics, lifestyle, myopia history, and swept source optical coherence tomography data, were analyzed. Sequential forward selection was used for predictor selection, and binary classification models were created using five machine learning algorithms to forecast the risk of MMD progression over 10 years.

Results

Over a median follow-up of 10.9 years, 133 patients (20.2%) showed MMD progression in the development cohort. Among them, 69 (51.9%) developed newly-onset MMD, 11 (8.3%) developed patchy atrophy from diffuse atrophy, 54 (40.6%) showed an enlargement of lesions, and 9 (6.8%) developed plus signs. Top six predictors for MMD progression included thinner subfoveal choroidal thickness, longer AL, worse best-corrected visual acuity, older age, female gender, and shallower anterior chamber depth. The eXtreme Gradient Boosting algorithm yielded the best discriminative performance (area under the receiver operating characteristic curve [AUROC] = 0.87 ± 0.02) with good calibration in the training cohort. In a less myopic external validation group (median -5.38 D), 48 patients (22.6%) developed MMD progression over 4 years, with the model's AUROC validated at 0.80 ± 0.008.

Conclusions

Machine learning model effectively predicts MMD progression a decade ahead using clinical and imaging indicators. This tool shows promise for identifying "at-risk" high myopes for timely intervention and vision protection.

Identifying dementia from cognitive footprints in hospital records among Chinese older adults: a machine-learning study

Background

By combining theory-driven and data-driven methods, this study aimed to develop dementia predictive algorithms among Chinese older adults guided by the cognitive footprint theory.

Methods

Electronic medical records from the Clinical Data Analysis and Reporting System in Hong Kong were employed. We included patients with dementia diagnosed at 65+ between 2010 and 2018, and 1:1 matched dementia-free controls. We identified 51 features, comprising exposures to established modifiable factors and other factors before and after 65 years old. The performances of four machine learning models, including LASSO, Multilayer perceptron (MLP), XGBoost, and LightGBM, were compared with logistic regression models, for all patients and subgroups by age.

Findings

A total of 159,920 individuals (40.5% male; mean age [SD]: 83.97 [7.38]) were included. Compared with the model included established modifiable factors only (area under the curve [AUC] 0.689, 95% CI [0.684, 0.694]), the predictive accuracy substantially improved for models with all factors (0.774, [0.770, 0.778]). Machine learning and logistic regression models performed similarly, with AUC ranged between 0.773 (0.768, 0.777) for LASSO and 0.780 (0.776, 0.784) for MLP. Antipsychotics, education, antidepressants, head injury, and stroke were identified as the most important predictors in the total sample. Age-specific models identified different important features, with cardiovascular and infectious diseases becoming prominent in older ages.

Interpretation

The models showed satisfactory performances in identifying dementia. These algorithms can be used in clinical practice to assist decision making and allow timely interventions cost-effectively.

Funding

The Research Grants Council of Hong Kong under the Early Career Scheme 27110519.

Identifying a group of factors predicting cognitive impairment among older adults

BACKGROUND

Cognitive impairment has multiple risk factors spanning several domains, but few studies have evaluated risk factor clusters. We aimed to identify naturally occurring clusters of risk factors of poor cognition among middle-aged and older adults and evaluate associations between measures of cognition and these risk factor clusters.

METHODS

We used data from the National Health and Nutrition Examination Survey (NHANES) III (training dataset, n = 4074) and the NHANES 2011-2014 (validation dataset, n = 2510). Risk factors were selected based on the literature. We used both traditional logistic models and support vector machine methods to construct a composite score of risk factor clusters. We evaluated associations between the risk score and cognitive performance using the logistic model by estimating odds ratios (OR) and 95% confidence intervals (CI).

RESULTS

Using the training dataset, we developed a composite risk score that predicted undiagnosed cognitive decline based on ten selected predictive risk factors including age, waist circumference, healthy eating index, race, education, income, physical activity, diabetes, hypercholesterolemia, and annual visit to dentist. The risk score was significantly associated with poor cognitive performance both in the training dataset (OR Tertile 3 verse tertile 1 = 8.15, 95% CI: 5.36-12.4) and validation dataset (OR Tertile 3 verse tertile 1 = 4.31, 95% CI: 2.62-7.08). The area under the receiver operating characteristics curve for the predictive model was 0.74 and 0.77 for crude model and model adjusted for age, sex, and race.

CONCLUSION

The model based on selected risk factors may be used to identify high risk individuals with cognitive impairment.

An Explainable Artificial Intelligence Model to Predict Malignant Cerebral Edema after Acute Anterior Circulating Large-Hemisphere Infarction

INTRODUCTION

Malignant cerebral edema (MCE) is a serious complication and the main cause of poor prognosis in patients with large-hemisphere infarction (LHI). Therefore, the rapid and accurate identification of potential patients with MCE is essential for timely therapy. This study utilized an artificial intelligence-based machine learning approach to establish an interpretable model for predicting MCE in patients with LHI.

METHODS

This study included 314 patients with LHI not undergoing recanalization therapy. The patients were divided into MCE and non-MCE groups, and the eXtreme Gradient Boosting (XGBoost) model was developed. A confusion matrix was used to measure the prediction performance of the XGBoost model. We also utilized the SHapley Additive exPlanations (SHAP) method to explain the XGBoost model. Decision curve and receiver operating characteristic curve analyses were performed to evaluate the net benefits of the model.

RESULTS

MCE was observed in 121 (38.5%) of the 314 patients with LHI. The model showed excellent predictive performance, with an area under the curve of 0.916. The SHAP method revealed the top 10 predictive variables of the MCE such as ASPECTS score, NIHSS score, CS score, APACHE II score, HbA1c, AF, NLR, PLT, GCS, and age based on their importance ranking.

CONCLUSION

An interpretable predictive model can increase transparency and help doctors accurately predict the occurrence of MCE in LHI patients not undergoing recanalization therapy within 48 h of onset, providing patients with better treatment strategies and enabling optimal resource allocation.

ANU-ADRI scores, tau pathology, and cognition in non-demented adults: the CABLE study

BACKGROUND

It has been reported that the risk of Alzheimer's disease (AD) could be predicted by the Australian National University Alzheimer Disease Risk Index (ANU-ADRI) scores. However, among non-demented Chinese adults, the correlations of ANU-ADRI scores with cerebrospinal fluid (CSF) core biomarkers and cognition remain unclear.

METHODS

Individuals from the Chinese Alzheimer's Biomarker and LifestyLE (CABLE) study were grouped into three groups (low/intermediate/high risk groups) based on their ANU-ADRI scores. The multiple linear regression models were conducted to investigate the correlations of ANU-ADRI scores with several biomarkers of AD pathology. Mediation model and structural equation model (SEM) were conducted to investigate the mediators of the correlation between ANU-ADRI scores and cognition.

RESULTS

A total of 1078 non-demented elders were included in our study, with a mean age of 62.58 (standard deviation [SD] 10.06) years as well as a female proportion of 44.16% (n = 476). ANU-ADRI scores were found to be significantly related with MMSE (β = -0.264, P < 0.001) and MoCA (β = -0.393, P < 0.001), as well as CSF t-tau (β = 0.236, P < 0.001), p-tau (β = 0.183, P < 0.001), and t-tau/Aβ42 (β = 0.094, P = 0.005). Mediation analyses indicated that the relationships of ANU-ADRI scores with cognitive scores were mediated by CSF t-tau or p-tau (mediating proportions ranging from 4.45% to 10.50%). SEM did not reveal that ANU-ADRI scores affected cognition by tau-related pathology and level of CSF soluble triggering receptor expressed on myeloid cells 2 (sTREM2).

CONCLUSION

ANU-ADRI scores were associated with cognition and tau pathology. We also revealed a potential pathological mechanism underlying the impact of ANU-ADRI scores on cognition.

Identifying Predictive Risk Factors for Future Cognitive Impairment Among Chinese Older Adults: Longitudinal Prediction Study

Background

The societal burden of cognitive impairment in China has prompted researchers to develop clinical prediction models aimed at making risk assessments that enable preventative interventions. However, it is unclear what types of risk factors best predict future cognitive impairment, if known risk factors make equally accurate predictions across different socioeconomic groups, and if existing prediction models are equally accurate across different subpopulations.

Objective

This paper aimed to identify which domain of health information best predicts future cognitive impairment among Chinese older adults and to examine if discrepancies exist in predictive ability across different population subsets.

Methods

Using data from the Chinese Longitudinal Healthy Longevity Survey, we quantified the ability of demographics, instrumental activities of daily living, activities of daily living, cognitive tests, social factors and hobbies, psychological factors, diet, exercise and sleep, chronic diseases, and 3 recently published logistic regression-based prediction models to predict 3-year risk of cognitive impairment in the general Chinese population and among male, female, rural-dwelling, urban-dwelling, educated, and not formally educated older adults. Predictive ability was quantified using the area under the receiver operating characteristic curve (AUC) and sensitivity-specificity curves through 20 repeats of 10-fold cross-validation.

Results

A total of 4047 participants were included in the study, of which 337 (8.3%) developed cognitive impairment 3 years after baseline data collection. The risk factor groups with the best predictive ability in the general population were demographics (AUC 0.78, 95% CI 0.77-0.78), cognitive tests (AUC 0.72, 95% CI 0.72-0.73), and instrumental activities of daily living (AUC 0.71, 95% CI 0.70-0.71). Demographics, cognitive tests, instrumental activities of daily living, and all 3 recreated prediction models had significantly higher AUCs when making predictions among female older adults compared to male older adults and among older adults with no formal education compared to those with some education.

Conclusions

This study suggests that demographics, cognitive tests, and instrumental activities of daily living are the most useful risk factors for predicting future cognitive impairment among Chinese older adults. However, the most predictive risk factors and existing models have lower predictive power among male, urban-dwelling, and educated older adults. More efforts are needed to ensure that equally accurate risk assessments can be conducted across different socioeconomic groups in China.

Leveraging electronic health records and knowledge networks for Alzheimer's disease prediction and sex-specific biological insights

Identification of Alzheimer's disease (AD) onset risk can facilitate interventions before irreversible disease progression. We demonstrate that electronic health records from the University of California, San Francisco, followed by knowledge networks (for example, SPOKE) allow for (1) prediction of AD onset and (2) prioritization of biological hypotheses, and (3) contextualization of sex dimorphism. We trained random forest models and predicted AD onset on a cohort of 749 individuals with AD and 250,545 controls with a mean area under the receiver operating characteristic of 0.72 (7 years prior) to 0.81 (1 day prior). We further harnessed matched cohort models to identify conditions with predictive power before AD onset. Knowledge networks highlight shared genes between multiple top predictors and AD (for example, APOE, ACTB, IL6 and INS). Genetic colocalization analysis supports AD association with hyperlipidemia at the APOE locus, as well as a stronger female AD association with osteoporosis at a locus near MS4A6A. We therefore show how clinical data can be utilized for early AD prediction and identification of personalized biological hypotheses.

Mitochondrial DNA haplogroup D and brain microstructure regulate cognitive function among community-dwelling older adults

INTRODUCTION

Maintaining physical and cognitive function among older adults is important. These functional states are affected by mitochondria through various mechanisms, such as cellular energy production and oxidative stress control. Owing to its involvement in the relations among the brain, cognition, and physical function, mitochondrial function may be affected by mitochondrial DNA (mtDNA) haplogroups. This study explored the effect of mtDNA haplogroups and brain microstructure on physical and cognitive functions among community-dwelling older adults.

METHODS

This study was a community-based cross-sectional research. A total of 128 subjects aged 65 years and older without dementia completed several assessments, including mtDNA sequencing, physical and cognitive function tests, and magnetic resonance imaging (MRI) scans. Cognitive function and impairment were assessed by the MMSE and AD8 questionnaires. mtDNA haplogroups were classified by HaploGrep 2 software, and white matter microstructural integrity was scanned by 3T MRI.

RESULTS

The mean age of the subjects was 77.3 years. After the adjustment for covariates, the mtDNA haplogroup D carriers showed significantly lower mini-mental state examination (MMSE) scores than other carriers (p = 0.047). Further considering the brain microstructure, the mtDNA haplogroup D (p = 0.002) and white matter volumes in the left precuneus corrected for total intracranial volumes (p = 0.014) were found to be independently influencing factors of the MMSE scores.

CONCLUSIONS

The mtDNA haplogroup D and white matter microstructure regulated the cognitive function among community-dwelling older adults. The findings provide new insights into the research gap. Scientists must further venture into this field.

Plasma proteomic profiles predict future dementia in healthy adults

The advent of proteomics offers an unprecedented opportunity to predict dementia onset. We examined this in data from 52,645 adults without dementia in the UK Biobank, with 1,417 incident cases and a follow-up time of 14.1 years. Of 1,463 plasma proteins, GFAP, NEFL, GDF15 and LTBP2 consistently associated most with incident all-cause dementia (ACD), Alzheimer's disease (AD) and vascular dementia (VaD), and ranked high in protein importance ordering. Combining GFAP (or GDF15) with demographics produced desirable predictions for ACD (area under the curve (AUC) = 0.891) and AD (AUC = 0.872) (or VaD (AUC = 0.912)). This was also true when predicting over 10-year ACD, AD and VaD. Individuals with higher GFAP levels were 2.32 times more likely to develop dementia. Notably, GFAP and LTBP2 were highly specific for dementia prediction. GFAP and NEFL began to change at least 10 years before dementia diagnosis. Our findings strongly highlight GFAP as an optimal biomarker for dementia prediction, even more than 10 years before the diagnosis, with implications for screening people at high risk for dementia and for early intervention.

Plasma metabolic profiles predict future dementia and dementia subtypes: a prospective analysis of 274,160 participants

BACKGROUND

Blood-based biomarkers for dementia are gaining attention due to their non-invasive nature and feasibility in regular healthcare settings. Here, we explored the associations between 249 metabolites with all-cause dementia (ACD), Alzheimer's disease (AD), and vascular dementia (VaD) and assessed their predictive potential.

METHODS

This study included 274,160 participants from the UK Biobank. Cox proportional hazard models were employed to investigate longitudinal associations between metabolites and dementia. The importance of these metabolites was quantified using machine learning algorithms, and a metabolic risk score (MetRS) was subsequently developed for each dementia type. We further investigated how MetRS stratified the risk of dementia onset and assessed its predictive performance, both alone and in combination with demographic and cognitive predictors.

RESULTS

During a median follow-up of 14.01 years, 5274 participants developed dementia. Of the 249 metabolites examined, 143 were significantly associated with incident ACD, 130 with AD, and 140 with VaD. Among metabolites significantly associated with dementia, lipoprotein lipid concentrations, linoleic acid, sphingomyelin, glucose, and branched-chain amino acids ranked top in importance. Individuals within the top tertile of MetRS faced a significantly greater risk of developing dementia than those in the lowest tertile. When MetRS was combined with demographic and cognitive predictors, the model yielded the area under the receiver operating characteristic curve (AUC) values of 0.857 for ACD, 0.861 for AD, and 0.873 for VaD.

CONCLUSIONS

We conducted the largest metabolome investigation of dementia to date, for the first time revealed the metabolite importance ranking, and highlighted the contribution of plasma metabolites for dementia prediction.

Association of biological age with health outcomes and its modifiable factors

Identifying the clinical implications and modifiable and unmodifiable factors of aging requires the measurement of biological age (BA) and age gap. Leveraging the biomedical traits involved with physical measures, biochemical assays, genomic data, and cognitive functions from the healthy participants in the UK Biobank, we establish an integrative BA model consisting of multi-dimensional indicators. Accelerated aging (age gap >3.2 years) at baseline is associated incident circulatory diseases, related chronic disorders, all-cause, and cause-specific mortality. We identify 35 modifiable factors for age gap (p < 4.81 × 10-4 ), where pulmonary functions, body mass, hand grip strength, basal metabolic rate, estimated glomerular filtration rate, and C-reactive protein show the most significant associations. Genetic analyses replicate the possible associations between age gap and health-related outcomes and further identify CST3 as an essential gene for biological aging, which is highly expressed in the brain and is associated with immune and metabolic traits. Our study profiles the landscape of biological aging and provides insights into the preventive strategies and therapeutic targets for aging.

Explainable variational autoencoder (E-VAE) model using genome-wide SNPs to predict dementia

OBJECTIVE

Alzheimer's disease (AD) and AD related dementias (ADRD) are complex multifactorial neurodegenerative diseases. The associations between genetic variants obtained from genome wide association studies (GWAS) are the most widely available and well documented variants associated with ADRD. Application of deep learning methods to analyze large scale GWAS data may be a powerful approach to elucidate the biological mechanisms in ADRD compared to penalized regression models that may lead to over-fitting.

METHODS

We developed a deep learning frame work explainable variational autoencoder (E-VAE) classifier model using genotype (GWAS SNPs = 5474) data from 2714 study participants in the Health and Retirement Study (HRS) to classify ADRD. We validated the generalizability of this model among 234 participants in the Religious Orders Study and Memory and Aging Project (ROSMAP). Utilizing a linear decoder approach we have extracted the weights associated with latent features for biological interpretation.

RESULTS

We obtained a predictive accuracy of 0.71 (95 % CI [0.59, 0.84]) with an AUC of 0.69 in the HRS test dataset and got an accuracy of 0.62 (95 % CI [0.56, 0.68]) with an AUC of 0.63 in the ROSMAP dataset.

CONCLUSION

This is the first study showing the generalizability of a deep learning prediction model for dementia using genetic variants in an independent cohort. The latent features identified using E-VAE can help us understand the biology of AD/ ADRD and better characterize disease status.

Cardiovascular disease risk models and dementia or cognitive decline: a systematic review

Background

Health cognitive promotion and protection is a critical topic. With the world's aging population and rising life expectancy, there will be many people living with highly age-related dementia illnesses. Cardiovascular disease (CVD) and dementia share the same risk factors, such as unhealthy lifestyles and metabolic factors. These recognized risks associated with CVD and dementia frequently co-occur. CVD risk models may have a close association with dementia and cognitive decline. So, this systematic review aimed to determine whether CVD risk models were connected with dementia or cognitive decline and compare the predictive ability of various models.

Methods

PubMed, Web of Science, PsychINFO, Embase, Cochrane Library, CNKI, Sinomed, and WanFang were searched from 1 January 2014 until 16 February 2023. Only CVD risk models were included. We used the Newcastle-Ottawa scale (NOS) for the quality assessment of included cohort studies and the Agency for Healthcare Research and Quality (AHRQ) for cross-sectional studies. The Preferred Reporting Items for Systematic Reviews and Meta-analysis (PRISMA) statement's guidelines were followed in this systematic study.

Results

In all, 9,718 references were screened, of which 22 articles were included. A total of 15 CVD risk models were summarized. Except for the Cardiovascular Health in Ambulatory Care Research Team (CANHEART) health index, the other 14 CVD risk models were associated with dementia and cognitive decline. In comparison, different CVD risk models and domain-specific cognitive function correlation variation depended on cohort characteristics, risk models, cognitive function tests, and study designs. Moreover, it needed to be clarified when comparing the predicting performance of different CVD risk models.

Conclusion

It is significant for public health to improve disease risk prediction and prevention and mitigate the potential adverse effects of the heart on the brain. More cohort studies are warranted to prove the correlation between CVD risk models and cognitive function. Moreover, further studies are encouraged to compare the efficacy of CVD risk models in predicting cognitive disorders.

Current challenges and potential solutions to the use of digital health technologies in evidence generation: a narrative review

Digital health is a field that aims to improve patient care through the use of technology, such as telemedicine, mobile health, electronic health records, and artificial intelligence. The aim of this review is to examine the challenges and potential solutions for the implementation and evaluation of digital health technologies. Digital tools are used across the world in different settings. In Australia, the Digital Health Translation and Implementation Program (DHTI) emphasizes the importance of involving stakeholders and addressing infrastructure and training issues for healthcare workers. The WHO's Global Task Force on Digital Health for TB aims to address tuberculosis through digital health innovations. Digital tools are also used in mental health care, but their effectiveness must be evaluated during development. Oncology supportive care uses digital tools for cancer patient intervention and surveillance, but evaluating their effectiveness can be challenging. In the COVID and post-COVID era, digital health solutions must be evaluated based on their technological maturity and size of deployment, as well as the quality of data they provide. To safely and effectively use digital healthcare technology, it is essential to prioritize evaluation using complex systems and evidence-based medical frameworks. To address the challenges of digital health implementation, it is important to prioritize ethical research addressing issues of user consent and addressing socioeconomic disparities in access and effectiveness. It is also important to consider the impact of digital health on health outcomes and the cost-effectiveness of service delivery.

Artificial intelligence in psychiatry research, diagnosis, and therapy

Psychiatric disorders are now responsible for the largest proportion of the global burden of disease, and even more challenges have been seen during the COVID-19 pandemic. Artificial intelligence (AI) is commonly used to facilitate the early detection of disease, understand disease progression, and discover new treatments in the fields of both physical and mental health. The present review provides a broad overview of AI methodology and its applications in data acquisition and processing, feature extraction and characterization, psychiatric disorder classification, potential biomarker detection, real-time monitoring, and interventions in psychiatric disorders. We also comprehensively summarize AI applications with regard to the early warning, diagnosis, prognosis, and treatment of specific psychiatric disorders, including depression, schizophrenia, autism spectrum disorder, attention-deficit/hyperactivity disorder, addiction, sleep disorders, and Alzheimer's disease. The advantages and disadvantages of AI in psychiatry are clarified. We foresee a new wave of research opportunities to facilitate and improve AI technology and its long-term implications in psychiatry during and after the COVID-19 era.

Dementia prediction in the general population using clinically accessible variables: a proof-of-concept study using machine learning. The AGES-Reykjavik study

BACKGROUND

Early identification of dementia is crucial for prompt intervention for high-risk individuals in the general population. External validation studies on prognostic models for dementia have highlighted the need for updated models. The use of machine learning in dementia prediction is in its infancy and may improve predictive performance. The current study aimed to explore the difference in performance of machine learning algorithms compared to traditional statistical techniques, such as logistic and Cox regression, for prediction of all-cause dementia. Our secondary aim was to assess the feasibility of only using clinically accessible predictors rather than MRI predictors.

METHODS

Data are from 4,793 participants in the population-based AGES-Reykjavik Study without dementia or mild cognitive impairment at baseline (mean age: 76 years, % female: 59%). Cognitive, biometric, and MRI assessments (total: 59 variables) were collected at baseline, with follow-up of incident dementia diagnoses for a maximum of 12 years. Machine learning algorithms included elastic net regression, random forest, support vector machine, and elastic net Cox regression. Traditional statistical methods for comparison were logistic and Cox regression. Model 1 was fit using all variables and model 2 was after feature selection using the Boruta package. A third model explored performance when leaving out neuroimaging markers (clinically accessible model). Ten-fold cross-validation, repeated ten times, was implemented during training. Upsampling was used to account for imbalanced data. Tuning parameters were optimized for recalibration automatically using the caret package in R.

RESULTS

19% of participants developed all-cause dementia. Machine learning algorithms were comparable in performance to logistic regression in all three models. However, a slight added performance was observed in the elastic net Cox regression in the third model (c = 0.78, 95% CI: 0.78-0.78) compared to the traditional Cox regression (c = 0.75, 95% CI: 0.74-0.77).

CONCLUSIONS

Supervised machine learning only showed added benefit when using survival techniques. Removing MRI markers did not significantly worsen our model's performance. Further, we presented the use of a nomogram using machine learning methods, showing transportability for the use of machine learning models in clinical practice. External validation is needed to assess the use of this model in other populations. Identifying high-risk individuals will amplify prevention efforts and selection for clinical trials.

Developing and validating a nomogram for cognitive impairment in the older people based on the NHANES

Objective

To use the United States National Health and Nutrition Examination Study (NHANES) to develop and validate a risk-prediction nomogram for cognitive impairment in people aged over 60 years.

Methods

A total of 2,802 participants (aged ≥ 60 years) from NHANES were analyzed. The least absolute shrinkage and selection operator (LASSO) regression model and multivariable logistic regression analysis were used for variable selection and model development. ROC-AUC, calibration curve, and decision curve analysis (DCA) were used to evaluate the nomogram's performance.

Results

The nomogram included five predictors, namely sex, moderate activity, taste problem, age, and education. It demonstrated satisfying discrimination with a AUC of 0.744 (95% confidence interval, 0.696-0.791). The nomogram was well-calibrated according to the calibration curve. The DCA demonstrated that the nomogram was clinically useful.

Conclusion

The risk-prediction nomogram for cognitive impairment in people aged over 60 years was effective. All predictors included in this nomogram can be easily accessed from its' user.

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

Importance

While the Australian National University-Alzheimer Disease Risk Index (ANU-ADRI), Cardiovascular Risk Factors, Aging, and Dementia (CAIDE), and Lifestyle for Brain Health (LIBRA) dementia risk tools have been widely used, a large body of new evidence has emerged since their publication. Recently, Cognitive Health and Dementia Risk Index (CogDrisk) and CogDrisk for Alzheimer disease (CogDrisk-AD) risk tools have been developed for the assessment of dementia and AD risk, respectively, using contemporary evidence; comparison of the relative performance of these risk tools is limited.

Objective

To evaluate the performance of CogDrisk, ANU-ADRI, CAIDE, LIBRA, and modified LIBRA (LIBRA with age and sex estimates from ANU-ADRI) in estimating dementia and AD risks (with CogDrisk-AD and ANU-ADRI).

Design, Setting, and Participants

This population-based cohort study obtained data from the Rush Memory and Aging Project (MAP), the Cardiovascular Health Study Cognition Study (CHS-CS), and the Health and Retirement Study-Aging, Demographics and Memory Study (HRS-ADAMS). Participants who were free of dementia at baseline were included. The factors were component variables in the risk tools that included self-reported baseline demographics, medical risk factors, and lifestyle habits. The study was conducted between November 2021 and March 2023, and statistical analysis was performed from January to June 2023.

Main outcomes and measures

Risk scores were calculated based on available factors in each of these cohorts. Area under the receiver operating characteristic curve (AUC) was calculated to measure the performance of each risk score. Multiple imputation was used to assess whether missing data may have affected estimates for dementia risk.

Results

Among the 6107 participants in 3 validation cohorts included for this study, 2184 participants without dementia at baseline were available from MAP (mean [SD] age, 80.0 [7.6] years; 1606 [73.5%] female), 548 participants without dementia at baseline were available from HRS-ADAMS (mean [SD] age, 79.5 [6.3] years; 288 [52.5%] female), and 3375 participants without dementia at baseline were available from CHS-CS (mean [SD] age, 74.8 [4.9] years; 1994 [59.1%] female). In all 3 cohorts, a similar AUC for dementia was obtained using CogDrisk, ANU-ADRI, and modified LIBRA (MAP cohort: CogDrisk AUC, 0.65 [95% CI, 0.61-0.69]; ANU-ADRI AUC, 0.65 [95% CI, 0.61-0.69]; modified LIBRA AUC, 0.65 [95% CI, 0.61-0.69]; HRS-ADAMS cohort: CogDrisk AUC, 0.75 [95% CI, 0.71-0.79]; ANU-ADRI AUC, 0.74 [95% CI, 0.70-0.78]; modified LIBRA AUC, 0.75 [95% CI, 0.71-0.79]; CHS-CS cohort: CogDrisk AUC, 0.70 [95% CI, 0.67-0.72]; ANU-ADRI AUC, 0.69 [95% CI, 0.66-0.72]; modified LIBRA AUC, 0.70 [95% CI, 0.68-0.73]). The CAIDE and LIBRA also provided similar but lower AUCs than the 3 aforementioned tools (eg, MAP cohort: CAIDE AUC, 0.50 [95% CI, 0.46-0.54]; LIBRA AUC, 0.53 [95% CI, 0.48-0.57]). The performance of CogDrisk-AD and ANU-ADRI in estimating AD risks was also similar.

Conclusions and relevance

CogDrisk and CogDrisk-AD performed similarly to ANU-ADRI in estimating dementia and AD risks. These results suggest that CogDrisk and CogDrisk-AD, with a greater range of modifiable risk factors compared with other risk tools in this study, may be more informative for risk reduction.

Phenomic Studies on Diseases: Potential and Challenges

The rapid development of such research field as multi-omics and artificial intelligence (AI) has made it possible to acquire and analyze the multi-dimensional big data of human phenomes. Increasing evidence has indicated that phenomics can provide a revolutionary strategy and approach for discovering new risk factors, diagnostic biomarkers and precision therapies of diseases, which holds profound advantages over conventional approaches for realizing precision medicine: first, the big data of patients' phenomes can provide remarkably richer information than that of the genomes; second, phenomic studies on diseases may expose the correlations among cross-scale and multi-dimensional phenomic parameters as well as the mechanisms underlying the correlations; and third, phenomics-based studies are big data-driven studies, which can significantly enhance the possibility and efficiency for generating novel discoveries. However, phenomic studies on human diseases are still in early developmental stage, which are facing multiple major challenges and tasks: first, there is significant deficiency in analytical and modeling approaches for analyzing the multi-dimensional data of human phenomes; second, it is crucial to establish universal standards for acquirement and management of phenomic data of patients; third, new methods and devices for acquirement of phenomic data of patients under clinical settings should be developed; fourth, it is of significance to establish the regulatory and ethical guidelines for phenomic studies on diseases; and fifth, it is important to develop effective international cooperation. It is expected that phenomic studies on diseases would profoundly and comprehensively enhance our capacity in prevention, diagnosis and treatment of diseases.

Harnessing the potential of machine learning and artificial intelligence for dementia research

Progress in dementia research has been limited, with substantial gaps in our knowledge of targets for prevention, mechanisms for disease progression, and disease-modifying treatments. The growing availability of multimodal data sets opens possibilities for the application of machine learning and artificial intelligence (AI) to help answer key questions in the field. We provide an overview of the state of the science, highlighting current challenges and opportunities for utilisation of AI approaches to move the field forward in the areas of genetics, experimental medicine, drug discovery and trials optimisation, imaging, and prevention. Machine learning methods can enhance results of genetic studies, help determine biological effects and facilitate the identification of drug targets based on genetic and transcriptomic information. The use of unsupervised learning for understanding disease mechanisms for drug discovery is promising, while analysis of multimodal data sets to characterise and quantify disease severity and subtype are also beginning to contribute to optimisation of clinical trial recruitment. Data-driven experimental medicine is needed to analyse data across modalities and develop novel algorithms to translate insights from animal models to human disease biology. AI methods in neuroimaging outperform traditional approaches for diagnostic classification, and although challenges around validation and translation remain, there is optimism for their meaningful integration to clinical practice in the near future. AI-based models can also clarify our understanding of the causality and commonality of dementia risk factors, informing and improving risk prediction models along with the development of preventative interventions. The complexity and heterogeneity of dementia requires an alternative approach beyond traditional design and analytical approaches. Although not yet widely used in dementia research, machine learning and AI have the potential to unlock current challenges and advance precision dementia medicine.

Reconceptualization of the Erlangen Score for the Assessment of Dementia Risk: The ERlangen Score

BACKGROUND

The established Erlangen Score (ES) for the interpretation of cerebrospinal fluid (CSF) biomarkers in the diagnostics of Alzheimer's disease (AD) uses markers of amyloidopathy and tauopathy, equally weighted to form an easy-interpretable ordinal scale. However, these biomarkers are not equally predictive for AD.

OBJECTIVE

The higher weighting of the Aβ42/Aβ40 ratio, as a reconceptualized ERlangen Score (ERS), was tested for advantages in diagnostic performance.

METHODS

Non-demented subjects (N = 154) with a mean follow up of 5 years were assigned to a group ranging from 0 to 4 in ES or ERS. Psychometric trajectories and dementia risk were assessed.

RESULTS

The distribution of subjects between ES and ERS among the groups differed considerably, as grouping allocated 32 subjects to ES group 2, but only 2 to ERS group 2. The discriminative accuracy between the ES (AUC 73.2%, 95% CI [64.2, 82.2]) and ERS (AUC 72.0%, 95% CI [63.1, 81.0]) for dementia risk showed no significant difference. Without consideration of the Aβ42/Aβ40 ratio in ES grouping, the optimal cut-off of the ES shifted to ≥2.

CONCLUSIONS

The ERS showed advantages over the ES in test interpretation with comparable overall test performance, as fewer cases were allocated to the intermediate risk group. The established cut-off of ≥2 can be maintained for the ERS, whereas it must be adjusted for the ES when determining the Aβ42/Aβ40 ratio.

Validation of the CogDrisk Instrument as Predictive of Dementia in Four General Community-Dwelling Populations

BACKGROUND

Lack of external validation of dementia risk tools is a major limitation for generalizability and translatability of prediction scores in clinical practice and research.

OBJECTIVES

We aimed to validate a new dementia prediction risk tool called CogDrisk and a version, CogDrisk-AD for predicting Alzheimer's disease (AD) using cohort studies.

DESIGN, SETTING, PARTICIPANTS AND MEASUREMENTS

Four cohort studies were identified that included majority of the dementia risk factors from the CogDrisk tool. Participants who were free of dementia at baseline were included. The predictors were component variables in the CogDrisk tool that include self-reported demographics, medical risk factors and lifestyle habits. Risk scores for Any Dementia and AD were computed and Area Under the Curve (AUC) was assessed. To examine modifiable risk factors for dementia, the CogDrisk tool was tested by excluding age and sex estimates from the model.

RESULTS

The performance of the tool varied between studies. The overall AUC and 95% CI for predicting dementia was 0.77 (0.57, 0.97) for the Swedish National study on Aging and Care in Kungsholmen, 0.76 (0.70, 0.83) for the Health and Retirement Study - Aging, Demographics and Memory Study, 0.70 (0.67,0.72) for the Cardiovascular Health Study Cognition Study, and 0.66 (0.62,0.70) for the Rush Memory and Aging Project.

CONCLUSIONS

The CogDrisk and CogDrisk-AD performed well in the four studies. Overall, this tool can be used to assess individualized risk factors of dementia and AD in various population settings.

Application of Grey Relational Analysis to Predict Dementia Tendency by Cognitive Function, Sleep Disturbances, and Health Conditions of Diabetic Patients

Background: The number of elderly diabetic patients has been increasing recently, and these patients have a higher morbidity of dementia than those without diabetes. Diabetes is associated with an increased risk for the development of dementia in elderly individuals, which is a serious health problem. Objectives: The primary aim was to examine whether diabetes is a risk factor for dementia among elderly individuals. The secondary aim was to apply grey theory to integrate the results and how they relate to cognitive impairments in elderly diabetic patients and to predict which participants are at high risk of developing dementia. Methods: Two hundred and twenty patients aged 50 years or older who were diagnosed with diabetes mellitus were recruited. Information on demographics, disease characteristics, activities of daily living, Mini Mental State Examination, sleep quality, depressive symptoms, and health-related quality of life was collected via questionnaires. The grey relational analysis approach was applied to evaluate the relationship between the results and health outcomes. Results: A total of 13.6% of participants had cognitive disturbances, of whom 1.4% had severe cognitive dysfunction. However, with regard to sleep disorders, 56.4% had sleep disturbances of varying degrees from light to severe. Further investigation is needed to address this problem. A higher prevalence of sleep disturbances among diabetic patients translates to a higher degree of depressive symptoms and a worse physical and mental health-related quality of life. Furthermore, based on the grey relational analysis, the grey relation coefficient varies from 0.6217~0.7540. Among the subjects, Participant 101 had the highest value, suggesting a need for immediate medical care. In this study, we observed that 20% of the total participants, for whom the grey relation coefficient was 0.6730, needed further and immediate medical care.