Development and validation of prediction models for gestational diabetes treatment modality using supervised machine learning: a population-based cohort study.

BACKGROUND: Gestational diabetes (GDM) is prevalent and benefits from timely and effective treatment, given the short window to impact glycemic control. Clinicians face major barriers to choosing effectively among treatment modalities [medical nutrition therapy (MNT) with or without pharmacologic treatment (antidiabetic oral agents and/or insulin)]. We investigated whether clinical data at varied stages of pregnancy can predict GDM treatment modality. METHODS: Among a population-based cohort of 30,474 pregnancies with GDM delivered at Kaiser Permanente Northern California in 2007-2017, we selected those in 2007-2016 as the discovery set and 2017 as the temporal/future validation set. Potential predictors were extracted from electronic health records at different timepoints (levels 1-4): (1) 1-year preconception to the last menstrual period, (2) the last menstrual period to GDM diagnosis, (3) at GDM diagnosis, and (4) 1 week after GDM diagnosis. We compared transparent and ensemble machine learning prediction methods, including least absolute shrinkage and selection operator (LASSO) regression and super learner, containing classification and regression tree, LASSO regression, random forest, and extreme gradient boosting algorithms, to predict risks for pharmacologic treatment beyond MNT. RESULTS: The super learner using levels 1-4 predictors had higher predictability [tenfold cross-validated C-statistic in discovery/validation set: 0.934 (95% CI: 0.931-0.936)/0.815 (0.800-0.829)], compared to levels 1, 1-2, and 1-3 (discovery/validation set C-statistic: 0.683-0.869/0.634-0.754). A simpler, more interpretable model, including timing of GDM diagnosis, diagnostic fasting glucose value, and the status and frequency of glycemic control at fasting during one-week post diagnosis, was developed using tenfold cross-validated logistic regression based on super learner-selected predictors. This model compared to the super learner had only a modest reduction in predictability [discovery/validation set C-statistic: 0.825 (0.820-0.830)/0.798 (95% CI: 0.783-0.813)]. CONCLUSIONS: Clinical data demonstrated reasonably high predictability for GDM treatment modality at the time of GDM diagnosis and high predictability at 1-week post GDM diagnosis. These population-based, clinically oriented models may support algorithm-based risk-stratification for treatment modality, inform timely treatment, and catalyze more effective management of GDM.

Who is most at risk of dying if infected with SARS-CoV-2? A mortality risk factor analysis using machine learning of patients with COVID-19 over time: a large population-based cohort study in Mexico.

OBJECTIVE: COVID-19 would kill fewer people if health programmes can predict who is at higher risk of mortality because resources can be targeted to protect those people from infection. We predict mortality in a very large population in Mexico with machine learning using demographic variables and pre-existing conditions. DESIGN: Cohort study. SETTING: March 2020 to November 2021 in Mexico, nationally represented. PARTICIPANTS: 1.4 million laboratory-confirmed patients with COVID-19 in Mexico at or over 20 years of age. PRIMARY AND SECONDARY OUTCOME MEASURES: Analysis is performed on data from March 2020 to November 2021 and over three phases: (1) from March to October in 2020, (2) from November 2020 to March 2021 and (3) from April to November 2021. We predict mortality using an ensemble machine learning method, super learner, and independently estimate the adjusted mortality relative risk of each pre-existing condition using targeted maximum likelihood estimation. RESULTS: Super learner fit has a high predictive performance (C-statistic: 0.907), where age is the most predictive factor for mortality. After adjusting for demographic factors, renal disease, hypertension, diabetes and obesity are the most impactful pre-existing conditions. Phase analysis shows that the adjusted mortality risk decreased over time while relative risk increased for each pre-existing condition. CONCLUSIONS: While age is the most important predictor of mortality, younger individuals with hypertension, diabetes and obesity are at comparable mortality risk as individuals who are 20 years older without any of the three conditions. Our model can be continuously updated to identify individuals who should most be protected against infection as the pandemic evolves.

Associations of household food insufficiency with childhood depression and anxiety: a nationwide cross-sectional study in the USA.

OBJECTIVE: Household food insufficiency (HFIS) is a major public health threat to children. Children may be particularly vulnerable to HFIS as a psychological stressor due to their rapid growth and accelerated behavioural and cognitive states, whereas data focusing on HFIS and childhood mental disorders are as-yet sparse. We aimed to examine the associations of HFIS with depression and anxiety in US children. DESIGN: Cross-sectional study. SETTING: The 2016-2018 National Survey of Children's Health, a nationally-representative study. PARTICIPANTS: Primary caregivers of 102 341 children in the USA. PRIMARY AND SECONDARY OUTCOME MEASURES: Physician diagnosed depression and anxiety were assessed by questionnaires administered to primary caregivers of 102 341 children. Multivariable logistic regression models estimated adjusted OR (aOR) for current depression or anxiety associated with HFIS measured through a validated single-item instrument. RESULTS: Among children aged 3-17 years, 3.2% and 7.4% had parent-reported physician-diagnosed current depression and anxiety, respectively. Compared with children without HFIS, children with HFIS had approximately twofold higher weighted prevalence of anxiety or depression. After adjusting for covariates, children with versus without HFIS had a 1.53-fold (95% CI 1.15 to 2.03) and 1.48-fold (95% CI 1.20 to 1.82) increased odds of current depression and anxiety, respectively. Associations were slightly more pronounced among girls (aOR (95% CI): depression 1.69 (1.16 to 2.48); anxiety 1.78 (1.33 to 2.38)) than boys (1.42 (0.98 to 2.08); 1.32 (1.00 to 1.73); both P-for-interaction <0.01). The associations did not vary by children's age or race/ethnicity. CONCLUSIONS: HFIS was independently associated with depression and anxiety among US children. Girls presented slightly greater vulnerability to HFIS in terms of impaired mental health. Children identified as food-insufficient may warrant mental health assessment and possible intervention. Assessment of HFIS among children with impaired mental health is also warranted. Our findings also highlight the importance of promptly addressing HFIS with referral to appropriate resources and inform its potential to alleviate childhood mental health issues.

Tools for interpretation of wastewater SARS-CoV-2 temporal and spatial trends demonstrated with data collected in the San Francisco Bay Area.

Wastewater surveillance for severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) RNA can be integrated with COVID-19 case data to inform timely pandemic response. However, more research is needed to apply and develop systematic methods to interpret the true SARS-CoV-2 signal from noise introduced in wastewater samples (e.g., from sewer conditions, sampling and extraction methods, etc.). In this study, raw wastewater was collected weekly from five sewersheds and one residential facility. The concentrations of SARS-CoV-2 in wastewater samples were compared to geocoded COVID-19 clinical testing data. SARS-CoV-2 was reliably detected (95% positivity) in frozen wastewater samples when reported daily new COVID-19 cases were 2.4 or more per 100,000 people. To adjust for variation in sample fecal content, four normalization biomarkers were evaluated: crAssphage, pepper mild mottle virus, Bacteroides ribosomal RNA (rRNA), and human 18S rRNA. Of these, crAssphage displayed the least spatial and temporal variability. Both unnormalized SARS-CoV-2 RNA signal and signal normalized to crAssphage had positive and significant correlation with clinical testing data (Kendall's Tau-b (τ)=0.43 and 0.38, respectively), but no normalization biomarker strengthened the correlation with clinical testing data. Locational dependencies and the date associated with testing data impacted the lead time of wastewater for clinical trends, and no lead time was observed when the sample collection date (versus the result date) was used for both wastewater and clinical testing data. This study supports that trends in wastewater surveillance data reflect trends in COVID-19 disease occurrence and presents tools that could be applied to make wastewater signal more interpretable and comparable across studies.

Protocol for audit of current Filipino practice in rehabilitation of stroke inpatients.

BACKGROUND: Stroke is one of the leading medical conditions in the Philippines. Over 500,000 Filipinos suffer from stroke annually. Provision of evidence-based medical and rehabilitation management for stroke patients has been a challenge due to existing environmental, social, and local health system issues. Thus, existing western guidelines on stroke rehabilitation were contextualized to draft recommendations relevant to the local Philippine setting. Prior to fully implementing the guidelines, an audit of current practice needs to be undertaken, thus the purpose of this audit protocol. METHODS: A clinical audit of current practices in stroke rehabilitation in the Philippines will be undertaken. A consensus list of data items to be captured was identified by the audit team during a 2-day meeting in 2012. These items, including patient demographics, type of stroke, time to referral for rehabilitation management, length of hospital stay, and other relevant descriptors of stroke management were included as part of the audit. Hospitals in the Philippines will be recruited to take part in the audit activity. Recruitment will be via the registry of the Philippine Academy of Rehabilitation Medicine, where 90% of physiatrists (medical doctors specialized in rehabilitation medicine) are active members and are affiliated with various hospitals in the Philippines. Data collectors will be identified and trained in the audit process. A pilot audit will be conducted to test the feasibility of the audit protocol, and refinements to the protocol will be undertaken as necessary. The comprehensive audit process will take place for a period of 3 months. Data will be encoded using MS Excel(®). Data will be reported as means and percentages as appropriate. Subgroup analysis will be undertaken to look into differences and variability of stroke patient descriptors and rehabilitation activities. CONCLUSION: This audit study is an ambitious project, but given the "need" to conduct the audit to identify "gaps" in current practice, and the value it can bring to serve as a platform for implementation of evidence-based stroke management in the Philippines to achieve best patient and health outcomes, the audit team is more than ready to take up the challenge.