Investigation of Super Learner Methodology on HIV-1 Small Sample: Application on Jaguar Trial Data

Predictors of Covid-19 level of concern among older adults from the health and retirement study

The purpose of this longitudinal study is to construct a prediction model for Covid-19 level of concern using established Covid-19 socio-demographic, lifestyle and health risk characteristics and to examine specific contributions of obesity-related cardiometabolic health characteristics as predictors of Covid-19 level of concern among a representative sample of U.S. older adults. We performed secondary analyses of existing data on 2872 2006–2020 Health and Retirement Study participants and examined 19 characteristics in relation to the outcome of interest using logistic regression and machine learning algorithms. In mixed-effects ordinal logistic regression models, a history of diabetes, stroke as well as 1–2 cardiometabolic risk factors and/or chronic conditions were associated with greater Covid-19 level of concern, after controlling for confounders. Female sex, birth cohort, minority race, Hispanic ethnicity and total wealth as well as depressive symptoms were associated with higher level of Covid-19 concern, and education was associated with lower level of Covid-19 concern in fully adjusted mixed-effects ordinal logistic regression models. The selected socio-demographic, lifestyle and health characteristics accounted for < 70% of the variability in Covid-19 level of concern based on machine learning algorithms. Independent risk factors for Covid-19 level of concern among U.S. older adults include socio-demographic characteristics and depressive symptoms. Advanced research is needed to identify relevant predictors and elucidate underlying mechanisms of observed relationships.

Development and Validation of a Machine Learning Model to Estimate Bacterial Sepsis Among Immunocompromised Recipients of Stem Cell Transplant

IMPORTANCE

Sepsis disproportionately affects recipients of allogeneic hematopoietic cell transplant (allo-HCT), and timely detection is crucial. However, the atypical presentation of sepsis within this population makes detection challenging, and existing clinical sepsis tools have limited prognostic value among this high-risk population.

OBJECTIVE

To develop a full risk factor (demographic, transplant, clinical, and laboratory factors) and clinical factor-specific automated bacterial sepsis decision support tool for recipients of allo-HCT with potential bloodstream infections (PBIs).

DESIGN, SETTING, AND PARTICIPANTS

This prognostic study used data from adult recipients of allo-HCT transplanted at the Fred Hutchinson Cancer Research Center, Seattle, Washington, between June 2010 and June 2019 randomly divided into 70% modeling and 30% validation data sets. Tools were developed using the area under the curve (AUC) optimized SuperLearner, and their performance was compared with existing clinical sepsis tools: National Early Warning Score (NEWS), quick Sequential Organ Failure Assessment (qSOFA), and Systemic Inflammatory Response Syndrome (SIRS), using the validation data set. Data were analyzed between January and October of 2020.

MAIN OUTCOMES AND MEASURES

The primary outcome was high-sepsis risk bacteremia (culture confirmed gram-negative species, Staphylococcus aureus, or Streptococcus spp bacteremia), and the secondary outcomes were 10- and 28-day mortality. Tool discrimination and calibration were examined using accuracy metrics and expected vs observed probabilities.

RESULTS

Between June 2010 and June 2019, 1943 recipients of allo-HCT received their first transplant, and 1594 recipients (median [interquartile range] age at transplant, 54 [43-63] years; 911 [57.2%] men; 1242 individuals [77.9%] identifying as White) experienced at least 1 PBI. Of 8131 observed PBIs, 238 (2.9%) were high-sepsis risk bacteremia. Compared with high-sepsis risk bacteremia, the full decision support tool had the highest AUC (0.85; 95% CI, 0.81-0.89), followed by the clinical factor-specific tool (0.72; 95% CI, 0.66-0.78). SIRS had the highest AUC of existing tools (0.64; 95% CI, 0.57-0.71). The full decision support tool had the highest AUCs for PBIs identified in inpatient (0.82; 95% CI, 0.76-0.89) and outpatient (0.82; 95% CI, 0.75-0.89) settings and for 10-day (0.85; 95% CI, 0.79-0.91) and 28-day (0.80; 95% CI, 0.75-0.84) mortality.

CONCLUSIONS AND RELEVANCE

These findings suggest that compared with existing tools and the clinical factor-specific tool, the full decision support tool had superior prognostic accuracy for the primary (high-sepsis risk bacteremia) and secondary (short-term mortality) outcomes in inpatient and outpatient settings. If used at the time of culture collection, the full decision support tool may inform more timely sepsis detection among recipients of allo-HCT.

Machine learning without borders? An adaptable tool to optimize mortality prediction in diverse clinical settings

BACKGROUND

Mortality prediction aids clinical decision making and is necessary for quality improvement initiatives. Validated metrics rely on prespecified variables and often require advanced diagnostics, which are unfeasible in resource-constrained contexts. We hypothesize that machine learning will generate superior mortality prediction in both high-income and low- and middle-income country cohorts.

METHODS

SuperLearner, an ensemble machine-learning algorithm, was applied to data from three prospective trauma cohorts: a highest-activation cohort in the United States, a high-volume center cohort in South Africa (SA), and a multicenter registry in Cameroon. Cross-validation was used to assess model discrimination of discharge mortality by site using receiver operating characteristic curves. SuperLearner discrimination was compared with standard scoring methods. Clinical variables driving SuperLearner prediction at each site were evaluated.

RESULTS

Data from 28,212 injured patients were used to generate prediction. Discharge mortality was 17%, 1.3%, and 1.7% among US, SA, and Cameroonian cohorts. SuperLearner delivered superior prediction of discharge mortality in the United States (area under the curve [AUC], 94-97%) and vastly superior prediction in Cameroon (AUC, 90-94%) compared with conventional scoring algorithms. It provided similar prediction to standard scores in the SA cohort (AUC, 90-95%). Context-specific variables (partial thromboplastin time in the United States and hospital distance in Cameroon) were prime drivers of predicted mortality in their respective cohorts, whereas severe brain injury predicted mortality across sites.

CONCLUSIONS

Machine learning provides excellent discrimination of injury mortality in diverse settings. Unlike traditional scores, data-adaptive methods are well suited to optimizing precise site-specific prediction regardless of diagnostic capabilities or data set inclusion allowing for individualized decision making and expanded access to quality improvement programming.

LEVEL OF EVIDENCE

Prognostic and therapeutic, level II and III.

Scoring methods for building genotypic scores: an application to didanosine resistance in a large derivation set

Background

Several attempts have been made to determine HIV-1 resistance from genotype resistance testing. We compare scoring methods for building weighted genotyping scores and commonly used systems to determine whether the virus of a HIV-infected patient is resistant.

Methods and Principal Findings

Three statistical methods (linear discriminant analysis, support vector machine and logistic regression) are used to determine the weight of mutations involved in HIV resistance. We compared these weighted scores with known interpretation systems (ANRS, REGA and Stanford HIV-db) to classify patients as resistant or not. Our methodology is illustrated on the Forum for Collaborative HIV Research didanosine database (N = 1453). The database was divided into four samples according to the country of enrolment (France, USA/Canada, Italy and Spain/UK/Switzerland). The total sample and the four country-based samples allow external validation (one sample is used to estimate a score and the other samples are used to validate it). We used the observed precision to compare the performance of newly derived scores with other interpretation systems. Our results show that newly derived scores performed better than or similar to existing interpretation systems, even with external validation sets. No difference was found between the three methods investigated. Our analysis identified four new mutations associated with didanosine resistance: D123S, Q207K, H208Y and K223Q.

Conclusions

We explored the potential of three statistical methods to construct weighted scores for didanosine resistance. Our proposed scores performed at least as well as already existing interpretation systems and previously unrecognized didanosine-resistance associated mutations were identified. This approach could be used for building scores of genotypic resistance to other antiretroviral drugs.