Emulate randomized clinical trials using heterogeneous treatment effect estimation for personalized treatments: Methodology review and benchmark.

Big data and (deep) machine learning have been ambitious tools in digital medicine, but these tools focus mainly on association. Intervention in medicine is about the causal effects. The average treatment effect has long been studied as a measure of causal effect, assuming that all populations have the same effect size. However, no "one-size-fits-all" treatment seems to work in some complex diseases. Treatment effects may vary by patient. Estimating heterogeneous treatment effects (HTE) may have a high impact on developing personalized treatment. Lots of advanced machine learning models for estimating HTE have emerged in recent years, but there has been limited translational research into the real-world healthcare domain. To fill the gap, we reviewed and compared eleven recent HTE estimation methodologies, including meta-learner, representation learning models, and tree-based models. We performed a comprehensive benchmark experiment based on nationwide healthcare claim data with application to Alzheimer's disease drug repurposing. We provided some challenges and opportunities in HTE estimation analysis in the healthcare domain to close the gap between innovative HTE models and deployment to real-world healthcare problems.

Enoxaparin may be associated with lower rates of mortality than unfractionated heparin in neurocritical and surgical patients.

Unfractionated heparin (UFH) and low molecular weight heparin (LMWH) are often administered to prevent venous thromboembolism (VTE) in critically ill patients. However, the preferred prophylactic agent (UFH or LMWH) is not known. We compared the all-cause mortality rate in patients receiving UFH to LMWH for VTE prophylaxis. We conducted a retrospective propensity score adjusted analysis of patients admitted to neuro-critical, surgical, or medical intensive care units. Patients were included if they were screened with venous duplex ultrasonography or computed tomography angiography for detection of VTE. The primary outcome was all-cause mortality. Secondary outcomes included the prevalence of VTE, deep vein thrombosis (DVT), pulmonary embolism (PE), and hospital length of stay (LOS). Initially 2228 patients in the cohort were included for analysis, 1836 (82%) patients received UFH, and 392 (18%) patients received enoxaparin. After propensity score matching, a well-balanced cohort of 618 patients remained in the study (309 patients receiving UFH; 309 patients receiving enoxaparin). The use of UFH for VTE prophylaxis in ICU patients was associated with similar rates of all-cause mortality compared with enoxaparin [RR 0.73; 95% CI 0.43-1.24, p = 0.310]. There were no differences in the prevalence of DVT, prevalence of PE or hospital LOS between the two groups, DVT [RR 0.93; 95% CI 0.56-1.53, p = 0.889], PE [RR 1.50; 95% CI 0.78-2.90, p = 0.296] and LOS [9 ± 9 days vs 9 ± 8; p = 0.857]. A trend toward mortality benefit was observed in NICU [RR 0.37; 95% CI 0.13-1.07, p = 0.062] and surgical patients [RR 0.43; 95% CI 0.17-1.02, p = 0.075] favoring the enoxaparin group. The use of UFH for VTE prophylaxis in ICU patients was associated with similar rates of VTE, all-cause mortality and LOS compared to enoxaparin. In subgroup analysis, neuro-critical and surgical patients who received UFH had a higher rate of mortality than those who received enoxaparin.

Identifying predictors of the tooth loss phenotype in a large periodontitis patient cohort using a machine learning approach.

OBJECTIVES: This study aimed to identify predictors associated with the tooth loss phenotype in a large periodontitis patient cohort in the university setting. METHODS: Information on periodontitis patients and nineteen factors identified at the initial visit was extracted from electronic health records. The primary outcome is tooth loss phenotype (presence or absence of tooth loss). Prediction models were built on significant factors (single or combinatory) selected by the RuleFit algorithm, and these factors were further adopted by regression models. Model performance was evaluated by Area Under the Receiver Operating Characteristic Curve (AUROC) and Area Under the Precision-Recall Curve (AUPRC). Associations between predictors and the tooth loss phenotype were also evaluated by classical statistical approaches to validate the performance of machine learning models. RESULTS: In total, 7840 patients were included. The machine learning model predicting the tooth loss phenotype achieved AUROC of 0.71 and AUPRC of 0.66. Age, periodontal diagnosis, number of missing teeth at baseline, furcation involvement, and tooth mobility were associated with the tooth loss phenotype in both machine learning and classical statistical models. CONCLUSIONS: The rule-based machine learning approach improves model explainability compared to classical statistical methods. However, the model's generalizability needs to be further validated by external datasets. CLINICAL SIGNIFICANCE: Predictors identified by the current machine learning approach using the RuleFit algorithm had clinically relevant thresholds in predicting the tooth loss phenotype in a large and diverse periodontitis patient cohort. The results of this study will assist clinicians in performing risk assessment for periodontitis at the initial visit.

An interpretable framework to identify responsive subgroups from clinical trials regarding treatment effects: Application to treatment of intracerebral hemorrhage.

Randomized Clinical trials (RCT) suffer from a high failure rate which could be caused by heterogeneous responses to treatment. Despite many models being developed to estimate heterogeneous treatment effects (HTE), there remains a lack of interpretable methods to identify responsive subgroups. This work aims to develop a framework to identify subgroups based on treatment effects that prioritize model interpretability. The proposed framework leverages an ensemble uplift tree method to generate descriptive decision rules that separate samples given estimated responses to the treatment. Subsequently, we select a complementary set of these decision rules and rank them using a sparse linear model. To address the trial's limited sample size problem, we proposed a data augmentation strategy by borrowing control patients from external studies and generating synthetic data. We apply the proposed framework to a failed randomized clinical trial for investigating an intracerebral hemorrhage therapy plan. The Qini-scores show that the proposed data augmentation strategy plan can boost the model's performance and the framework achieves greater interpretability by selecting complementary descriptive rules without compromising estimation quality. Our model derives clinically meaningful subgroups. Specifically, we find those patients with Diastolic Blood Pressure≥70 mm hg and Systolic Blood Pressure<215 mm hg benefit more from intensive blood pressure reduction therapy. The proposed interpretable HTE analysis framework offers a promising potential for extracting meaningful insight from RCTs with neutral treatment effects. By identifying responsive subgroups, our framework can contribute to developing personalized treatment strategies for patients more efficiently.

Do vaccinations influence the development of Alzheimer disease?

A growing literature supports a protective association between vaccines targeting an array of pathogens (e.g., influenza, pneumococcus, herpes zoster) and the risk of Alzheimer disease (AD). This article discusses the potential underlying mechanisms for this apparent protective effect of immunizations against infectious pathogens on the risk of AD; explores the basic and pharmacoepidemiologic evidence for this association, with particular attention paid to important methodological variations among the epidemiologic studies; and reviews the remaining uncertainties regarding the effects of anti-pathogen vaccines on Alzheimer disease and all-cause dementia, with recommendations for future directions to address those uncertainties.

Inferring Personalized Treatment Effect of Antihypertensives on Alzheimer's Disease Using Deep Learning.

Alzheimer's disease (AD) is one of the leading causes of death in the United States, especially among the elderly. Recent studies have shown how hypertension is related to cognitive decline in elderly patients, which in turn leads to increased mortality as well as morbidity. There have been various studies that have looked at the effect of antihypertensive drugs in reducing cognitive decline, and their results have proved inconclusive. However, most of these studies assume the treatment effect is similar for all patients, thus considering only the average treatment effects of antihypertensive drugs. In this paper, we assume that the effect of antihypertensives on the onset of AD depends on patient characteristics. We develop a deep learning method called LASSO-Dragonnet to estimate the individualized treatment effects of each patient. We considered six antihypertensive drugs, and each of the six models considered one of the drugs as the treatment and the remaining as control. Our studies showed that although many antihypertensives have a positive impact in delaying AD onset on average, the impact varies from individual to individual, depending on their various characteristics. We also analyzed the importance of various covariates in such an estimation. Our results showed that the individualized treatment effects of each patient could be estimated accurately using a deep learning method, and that the importance of various covariates could be determined.

Identifying predictors of tooth loss using a rule-based machine learning approach: A retrospective study at university-setting clinics.

BACKGROUND: This study aimed to identify predictors associated with tooth loss in a large periodontitis patient cohort in the university setting using the machine learning approach. METHODS: Information on periodontitis patients and 18 factors identified at the initial visit was extracted from electronic health records. A two-step machine learning pipeline was proposed to develop the tooth loss prediction model. The primary outcome is tooth loss count. The prediction model was built on significant factors (single or combination) selected by the RuleFit algorithm, and these factors were further adopted by the count regression model. Model performance was evaluated by root-mean-squared error (RMSE). Associations between predictors and tooth loss were also assessed by a classical statistical approach to validate the performance of the machine learning model. RESULTS: In total, 7840 patients were included. The machine learning model predicting tooth loss count achieved RMSE of 2.71. Age, smoking, frequency of brushing, frequency of flossing, periodontal diagnosis, bleeding on probing percentage, number of missing teeth at baseline, and tooth mobility were associated with tooth loss in both machine learning and classical statistical models. CONCLUSION: The two-step machine learning pipeline is feasible to predict tooth loss in periodontitis patients. Compared to classical statistical methods, this rule-based machine learning approach improves model explainability. However, the model's generalizability needs to be further validated by external datasets.

Characterizing Treatment Non-responders vs. Responders in Completed Alzheimer's Disease Clinical Trials.

Alzheimer's disease (AD) patients have varying responses to AD drugs and there may be no single treatment for all AD patients. Trial after trial shows that identifying non-responsive and responsive subgroups and their corresponding moderators will provide better insights into subject selection and interpretation in future clinical trials. We aim to extensively investigate pre-treatment features that moderate treatment effect of Galantamine, Bapineuzumab, and Semagacestat from completed trial data. We obtained individual-level patient data from ten randomized clinical trials. Six Galantamine trials and two Bapineuzumab trials were from Yale University Open Data Access Project and two Semagacestat trials were from the Center for Global Clinical Research Data. We included a total of 10,948 subjects. The trials were conducted worldwide from 2001 to 2012. We estimated treatment effect using causal forest modeling on each trial. Finally, we identified important pre-treatment features that determine treatment efficacy and identified responsive or nonresponsive subgroups. As a result, patient's pre-treatment conditions that determined the treatment efficacy of Galantamine differed by dementia stages, but we consistently observed that non-responders in Galantamine trials had lower BMI (25 vs 28, P < .001) and increased ages (74 vs 68, P < .001). Responders in Bapineuzumab and Semagacestat trials had lower Aß42 levels (6.41 vs 6.53 pg/ml, P < .001) and smaller whole brain volumes (983.13 vs 1052.78 ml, P < .001). 6 'positive' treatment trials had subsets of patients who had, in fact, not responded. 4 "negative" treatment trials had subsets of patients who had, in fact, responded. This study suggests that analyzing heterogeneity in treatment effects in "positive" or "negative" trials may be a very powerful tool for identifying distinct subgroups that are responsive to treatments, which may significantly benefit future clinical trial design and interpretation.

The Impact of Routine Vaccinations on Alzheimer's Disease Risk in Persons 65 Years and Older: A Claims-Based Cohort Study using Propensity Score Matching.

BACKGROUND: Accumulating evidence suggests that adult vaccinations can reduce the risk of developing Alzheimer's disease (AD) and Alzheimer's disease related dementias. OBJECTIVE: To compare the risk for developing AD between adults with and without prior vaccination against tetanus and diphtheria, with or without pertussis (Tdap/Td); herpes zoster (HZ); or pneumococcus. METHODS: A retrospective cohort study was performed using Optum's de-identified Clinformatics® Data Mart Database. Included patients were free of dementia during a 2-year look-back period and were≥65 years old by the start of the 8-year follow-up period. We compared two similar cohorts identified using propensity score matching (PSM), one vaccinated and another unvaccinated, with Tdap/Td, HZ, or pneumococcal vaccines. We calculated the relative risk (RR) and absolute risk reduction (ARR) for developing AD. RESULTS: For the Tdap/Td vaccine, 7.2% (n = 8,370) of vaccinated patients and 10.2% (n = 11,857) of unvaccinated patients developed AD during follow-up; the RR was 0.70 (95% CI, 0.68-0.72) and ARR was 0.03 (95% CI, 0.02-0.03). For the HZ vaccine, 8.1% (n = 16,106) of vaccinated patients and 10.7% (n = 21,417) of unvaccinated patients developed AD during follow-up; the RR was 0.75 (95% CI, 0.73-0.76) and ARR was 0.02 (95% CI, 0.02-0.02). For the pneumococcal vaccine, 7.92% (n = 20,583) of vaccinated patients and 10.9% (n = 28,558) of unvaccinated patients developed AD during follow-up; the RR was 0.73 (95% CI, 0.71-0.74) and ARR was 0.02 (95% CI, 0.02-0.03). CONCLUSION: Several vaccinations, including Tdap/Td, HZ, and pneumococcal, are associated with a reduced risk for developing AD.

Risk of Alzheimer's Disease Following Influenza Vaccination: A Claims-Based Cohort Study Using Propensity Score Matching.

BACKGROUND: Prior studies have found a reduced risk of dementia of any etiology following influenza vaccination in selected populations, including veterans and patients with serious chronic health conditions. However, the effect of influenza vaccination on Alzheimer's disease (AD) risk in a general cohort of older US adults has not been characterized. OBJECTIVE: To compare the risk of incident AD between patients with and without prior influenza vaccination in a large US claims database. METHODS: Deidentified claims data spanning September 1, 2009 through August 31, 2019 were used. Eligible patients were free of dementia during the 6-year look-back period and≥65 years old by the start of follow-up. Propensity-score matching (PSM) was used to create flu-vaccinated and flu-unvaccinated cohorts with similar baseline demographics, medication usage, and comorbidities. Relative risk (RR) and absolute risk reduction (ARR) were estimated to assess the effect of influenza vaccination on AD risk during the 4-year follow-up. RESULTS: From the unmatched sample of eligible patients (n = 2,356,479), PSM produced a sample of 935,887 flu-vaccinated-unvaccinated matched pairs. The matched sample was 73.7 (SD, 8.7) years of age and 56.9% female, with median follow-up of 46 (IQR, 29-48) months; 5.1% (n = 47,889) of the flu-vaccinated patients and 8.5% (n = 79,630) of the flu-unvaccinated patients developed AD during follow-up. The RR was 0.60 (95% CI, 0.59-0.61) and ARR was 0.034 (95% CI, 0.033-0.035), corresponding to a number needed to treat of 29.4. CONCLUSION: This study demonstrates that influenza vaccination is associated with reduced AD risk in a nationwide sample of US adults aged 65 and older.

Artificial intelligence in clinical and translational science: Successes, challenges and opportunities.

Artificial intelligence (AI) is transforming many domains, including finance, agriculture, defense, and biomedicine. In this paper, we focus on the role of AI in clinical and translational research (CTR), including preclinical research (T1), clinical research (T2), clinical implementation (T3), and public (or population) health (T4). Given the rapid evolution of AI in CTR, we present three complementary perspectives: (1) scoping literature review, (2) survey, and (3) analysis of federally funded projects. For each CTR phase, we addressed challenges, successes, failures, and opportunities for AI. We surveyed Clinical and Translational Science Award (CTSA) hubs regarding AI projects at their institutions. Nineteen of 63 CTSA hubs (30%) responded to the survey. The most common funding source (48.5%) was the federal government. The most common translational phase was T2 (clinical research, 40.2%). Clinicians were the intended users in 44.6% of projects and researchers in 32.3% of projects. The most common computational approaches were supervised machine learning (38.6%) and deep learning (34.2%). The number of projects steadily increased from 2012 to 2020. Finally, we analyzed 2604 AI projects at CTSA hubs using the National Institutes of Health Research Portfolio Online Reporting Tools (RePORTER) database for 2011-2019. We mapped available abstracts to medical subject headings and found that nervous system (16.3%) and mental disorders (16.2) were the most common topics addressed. From a computational perspective, big data (32.3%) and deep learning (30.0%) were most common. This work represents a snapshot in time of the role of AI in the CTSA program.