Machine learning models for predicting blood pressure phenotypes by combining multiple polygenic risk scores.

We construct non-linear machine learning (ML) prediction models for systolic and diastolic blood pressure (SBP, DBP) using demographic and clinical variables and polygenic risk scores (PRSs). We developed a two-model ensemble, consisting of a baseline model, where prediction is based on demographic and clinical variables only, and a genetic model, where we also include PRSs. We evaluate the use of a linear versus a non-linear model at both the baseline and the genetic model levels and assess the improvement in performance when incorporating multiple PRSs. We report the ensemble model's performance as percentage variance explained (PVE) on a held-out test dataset. A non-linear baseline model improved the PVEs from 28.1 to 30.1% (SBP) and 14.3% to 17.4% (DBP) compared with a linear baseline model. Including seven PRSs in the genetic model computed based on the largest available GWAS of SBP/DBP improved the genetic model PVE from 4.8 to 5.1% (SBP) and 4.7 to 5% (DBP) compared to using a single PRS. Adding additional 14 PRSs computed based on two independent GWASs further increased the genetic model PVE to 6.3% (SBP) and 5.7% (DBP). PVE differed across self-reported race/ethnicity groups, with primarily all non-White groups benefitting from the inclusion of additional PRSs. In summary, non-linear ML models improves BP prediction in models incorporating diverse populations.

Impact of Telemedicine on Access to Care for Rural Transgender and Gender-Diverse Youth.

OBJECTIVE: To explore the impact of telemedicine on access to gender-affirming care for rural transgender and gender diverse youth. STUDY DESIGN: A retrospective analysis of data drawn from the electronic medical records of a clinic that provides approximately 10 000 adolescent and young adult visits per year and serves patients seeking gender health care. The no-show rate was examined as a proxy for access to care due to anticipated challenges with recruiting a representative sample of a historically marginalized population. Logistic regression with generalized estimating equations was conducted to model the association between the odds of a no-show visit and covariates of interest. RESULTS: Telemedicine visits, rural home address, gender health visits, longer travel time, and being younger than 18 years old were associated with lower odds of a no-show in univariate models (n = 17 928 visits). In the adjusted model, the OR of no-shows for gender health visits was 0.56 (95% CI 0.42-0.74), adjusting for rurality, telemedicine, age (< or >18 years), and travel time to the clinic. CONCLUSIONS: In this study, telemedicine was associated with reduced no-shows overall, and especially for rural, transgender and gender diverse youth, and patients who hold both identities. Although the no-show rate does not fully capture barriers to access, these findings provide insight into how this vulnerable population may benefit from expanded access to telemedicine for rural individuals whose communities may lack providers with the skills to serve this population.

Heart rate patterns predicting cerebral palsy in preterm infants.

BACKGROUND: Heart rate (HR) patterns can inform on central nervous system dysfunction. We previously used highly comparative time series analysis (HCTSA) to identify HR patterns predicting mortality among patients in the neonatal intensive care unit (NICU) and now use this methodology to discover patterns predicting cerebral palsy (CP) in preterm infants. METHOD: We studied NICU patients <37 weeks' gestation with archived every-2-s HR data throughout the NICU stay and with or without later diagnosis of CP (n = 57 CP and 1119 no CP). We performed HCTSA of >2000 HR metrics and identified 24 metrics analyzed on HR data from two 7-day periods: week 1 and 37 weeks' postmenstrual age (week 1, week 37). Multivariate modeling was used to optimize a parsimonious prediction model. RESULTS: Week 1 HR metrics with maximum AUC for CP prediction reflected low variability, including "RobustSD" (AUC 0.826; 0.772-0.870). At week 37, high values of a novel HR metric, "LongSD3," the cubed value of the difference in HR values 100 s apart, were added to week 1 HR metrics for CP prediction. A combined birthweight + early and late HR model had AUC 0.853 (0.805-0.892). CONCLUSIONS: Using HCTSA, we discovered novel HR metrics and created a parsimonious model for CP prediction in preterm NICU patients. IMPACT: We discovered new heart rate characteristics predicting CP in preterm infants. Using every-2-s HR from two 7-day periods and highly comparative time series analysis, we found a measure of low variability HR week 1 after birth and a pattern of recurrent acceleration in HR at term corrected age that predicted CP. Combined clinical and early and late HR features had AUC 0.853 for CP prediction.

Reproductive Coercion Among Women With Disabilities: An Analysis of Pregnancy Risk Assessment Monitoring Systems Data.

BACKGROUND/AIMS: Reproductive coercion (RC) is a widespread yet understudied type of intimate partner violence that is associated with numerous negative outcomes. Women with disabilities may be at an increased risk of RC; however, little research has been conducted among this population. Using population-based data, we sought to examine the prevalence of RC in postpartum women with disabilities. METHODS: This is a secondary analysis of a cross-sectional survey, the Pregnancy Risk Assessment Monitoring System, a nationally representative survey conducted by the Centers for Disease Control and Prevention in partnership with participating states. These analyses include 3,117 respondents who had information on both disability status and experiences of RC. RESULTS: Approximately 1.9% of respondents reported experiencing RC (95% CI [1.3, 2.4]). When stratified by disability status, approximately 1.7% of respondents without a disability reported RC whereas 6.2% of respondents with at least one disability reported RC ( p < 0.001). In univariable logistic models, disability, age, education, relationship status, income, and race were all significantly associated with RC. CONCLUSIONS: Our findings highlight the need for healthcare providers working with women with disabilities to screen for RC and potentially uncover intimate partner violence and prevent its negative health consequences. All states participating in Pregnancy Risk Assessment Monitoring System data collection are urged to incorporate measures of RC and disability status to better address this significant issue.

The Impact of Physician Race and Sex on Patient Ranking of Physician Competence and Perception of Leadership Ability.

Background Biases affect patient perceptions of their physician and influence the physician-patient relationship. While racial disparities in care and inequities in the healthcare workforce are well-documented, the impact of physician race on patient perceptions remains unclear. We aimed to investigate the association of physician race and sex on patient perceptions during simulated preoperative encounters. Methods Three hundred patients recruited consecutively in the Preanesthesia Evaluation and Testing Center viewed pictures of 4 anesthesiologists (black male, white male, black female, white female) in random order while listening to a set of paired audio recordings describing general anesthesia. Participants ranked each anesthesiologist on confidence, intelligence, and likelihood of choosing the anesthesiologist to care for their family member, and chose the one anesthesiologist most like a leader. Results Compared to white anesthesiologists, black anesthesiologists had greater odds of being ranked more confident (OR, 1.45; 95% CI, 1.10 to 1.89; P=0.008) and being considered a leader (OR, 2.06; 95% CI, 1.50 to 2.84; P<0.0001). Among white participants, black anesthesiologists had greater odds of being ranked more intelligent (OR, 2.08; 95% CI, 1.54 to 2.81; P<0.0001) and were more likely to be chosen to care for a family member (OR, 2.26; 95% CI, 1.66 to 3.08; P<0.0001). Female anesthesiologists had greater odds of being ranked more intelligent (OR, 1.36; 95% CI, 1.08 to 1.71; P=0.009) and were more likely to be chosen to care for a family member (OR, 1.58; 95% CI, 1.27 to 1.97; P<0.001) compared with male anesthesiologists. Conclusions Contrary to our hypothesis, patients ranked black physicians more highly on multiple competence and leadership quality metrics. Our data likely highlight the role social desirability bias may play in studies of racial disparities within medicine.

Intimate partner violence during the perinatal period by disability status: Findings from a United States population-based analysis.

AIMS: The aim of the current study was to compare the prevalence of intimate partner violence (IPV) during the perinatal period among respondents with self-reported disability compared with those without a disability. DESIGN: We conducted a secondary analysis of nationally representative data from the Pregnancy Risk Assessment Monitoring System data from 24 participating United States between 2018 and 2020. METHODS: A cross-sectional sample of 43,837 respondents provided data on disability, including difficulty in vision, hearing, ambulation, cognition, communication and self-care. The exposure was perinatal IPV, defined as experiencing abuse by a current or ex-partner in the year before or during pregnancy. Regression models were used to calculated odds of IPV by disability status while accounting for relevant sociodemographic characteristics. RESULTS: Respondents who self-reported disabilities experienced IPV at a higher rate than those without disabilities, both before and during pregnancy. In fully adjusted models, respondents with disabilities had about 2.6 times the odds of experiencing IPV before pregnancy, and about 2.5 times the odds of experiencing IPV during pregnancy, compared with those without disabilities. CONCLUSION: Respondents with disabilities experienced IPV at higher rates than the general population, and thus are at increased risk for adverse maternal, neonatal and infant health outcomes. IMPACT: Perinatal IPV is a significant issue globally, and our findings suggest perinatal IPV is particularly salient for persons with disability. Findings highlight the need to screen women with disabilities for IPV during the perinatal period as well as the importance of providing them appropriate, accessible information, resources and referrals.

Depressive symptoms during the perinatal period by disability status: Findings from the United States Pregnancy Risk Assessment Monitoring System.

AIMS: The aim of the current study was to compare the prevalence of depressive symptoms during the perinatal period among respondents with a disability as compared to those without a disability. DESIGN: We conducted a secondary analysis of nationally representative data from the Pregnancy Risk Assessment Monitoring System data from 24 participating United States between 2018 and 2020. METHODS: A cross-sectional sample of 37,989 respondents provided data on disability, including difficulty in vision, hearing, ambulation, cognition, communication and self-care. The outcome of interest was perinatal depressive symptoms, defined as experiencing depressive symptoms during the antenatal period or postpartum period. Regression models were used to calculate odds of depressive symptoms during these two time periods by disability status while controlling for relevant sociodemographic characteristics and depressive symptoms prior to pregnancy. RESULTS: Respondents with disabilities experienced a higher prevalence of depressive symptoms in both the antenatal period and postpartum period as compared to those without disabilities. In fully adjusted models, respondents with disabilities had 2.4 times the odds of experiencing depressive symptoms during pregnancy and 2.1 times the odds of experiencing postpartum depressive symptoms as compared to respondents without disabilities. CONCLUSION: Respondents with disabilities experience a higher prevalence of depressive symptoms throughout the perinatal period thereby increasing the risk for adverse maternal, neonatal and infant health outcomes. IMPACT: Perinatal depression is a significant public health issue globally, and our findings suggest that persons with disability are at an increased risk for depressive symptoms both during pregnancy and in the postpartum period. Our findings represent a call to action to improve clinical and supportive services for women with disabilities during the perinatal period to improve their mental health and the consequent health of their offspring. PATIENT OR PUBLIC CONTRIBUTION: We thank our Community Advisory Board members who have been instrumental in the conception of this study.

Machine learning models for blood pressure phenotypes combining multiple polygenic risk scores.

We construct non-linear machine learning (ML) prediction models for systolic and diastolic blood pressure (SBP, DBP) using demographic and clinical variables and polygenic risk scores (PRSs). We developed a two-model ensemble, consisting of a baseline model, where prediction is based on demographic and clinical variables only, and a genetic model, where we also include PRSs. We evaluate the use of a linear versus a non-linear model at both the baseline and the genetic model levels and assess the improvement in performance when incorporating multiple PRSs. We report the ensemble model's performance as percentage variance explained (PVE) on a held-out test dataset. A non-linear baseline model improved the PVEs from 28.1% to 30.1% (SBP) and 14.3% to 17.4% (DBP) compared with a linear baseline model. Including seven PRSs in the genetic model computed based on the largest available GWAS of SBP/DBP improved the genetic model PVE from 4.8% to 5.1% (SBP) and 4.7% to 5% (DBP) compared to using a single PRS. Adding additional 14 PRSs computed based on two independent GWASs further increased the genetic model PVE to 6.3% (SBP) and 5.7% (DBP). PVE differed across self-reported race/ethnicity groups, with primarily all non-White groups benefitting from the inclusion of additional PRSs.

Heart Rate and Pulse Oximetry Dynamics in the First Week after Birth in Neonatal Intensive Care Unit Patients and the Risk of Cerebral Palsy.

OBJECTIVE: Infants in the neonatal intensive care unit (NICU) are at high risk of adverse neuromotor outcomes. Atypical patterns of heart rate (HR) and pulse oximetry (SpO2) may serve as biomarkers for risk assessment for cerebral palsy (CP). The purpose of this study was to determine whether atypical HR and SpO2 patterns in NICU patients add to clinical variables predicting later diagnosis of CP. STUDY DESIGN: This was a retrospective study including patients admitted to a level IV NICU from 2009 to 2017 with archived cardiorespiratory data in the first 7 days from birth to follow-up at >2 years of age. The mean, standard deviation (SD), skewness, kurtosis and cross-correlation of HR and SpO2 were calculated. Three predictive models were developed using least absolute shrinkage and selection operator regression (clinical, cardiorespiratory and combined model), and their performance for predicting CP was evaluated. RESULTS: Seventy infants with CP and 1,733 controls met inclusion criteria for a 3.8% population prevalence. Area under the receiver operating characteristic curve for CP prediction was 0.7524 for the clinical model, 0.7419 for the vital sign model, and 0.7725 for the combined model. Variables included in the combined model were lower maternal age, outborn delivery, lower 5-minute Apgar's score, lower SD of HR, and more negative skewness of HR. CONCLUSION: In this study including NICU patients of all gestational ages, HR but not SpO2 patterns added to clinical variables to predict the eventual diagnosis of CP. Identification of risk of CP within the first few days of life could result in improved therapy resource allocation and risk stratification in clinical trials of new therapeutics. KEY POINTS: · SD and skewness of HR have some added predictive value of later diagnosis of CP.. · SpO2 measures do not add to CP prediction.. · Combining clinical variables with early HR measures may improve the prediction of later CP..

Repressed Ang 1-7 in COVID-19 Is Inversely Associated with Inflammation and Coagulation.

The coronavirus SARS-CoV-2 infects host cells by binding to the angiotensin-converting enzyme 2 (ACE2) receptor, which belongs to an anti-inflammatory, anti-thrombotic counter-regulatory arm of the renin-angiotensin system (RAS). ACE2 dysfunction and RAS dysregulation has been explored as a driving force in acute respiratory distress syndrome (ARDS), but data from COVID-19 patients has been inconsistent and inconclusive. We sought to identify disruptions of the classical (ACE)/angiotensin (Ang) II/Ang II type-1 receptor (AT1R) and the counter-regulatory ACE2/Ang 1-7/Mas Receptor (MasR) pathways in patients with COVID-19 and correlate these with severity of infection and markers of inflammation and coagulation. Ang II and Ang 1-7 levels in plasma were measured by enzyme-linked immunosorbent assay (ELISA) for 230 patients, 166 of whom were SARS-CoV-2+. Ang 1-7 was repressed in COVID-19 patients compared to that in SARS-CoV-2 negative outpatient controls. Since the control cohort was less sick than the SARS-CoV-2+ group, this association between decreased Ang 1-7 and COVID-19 cannot be attributed to COVID-19 specifically as opposed to critical illness more generally. Multivariable logistic regression analyses demonstrated that every 10-pg/mL increase in plasma Ang 1-7 was associated with a 3% reduction in the odds of hospitalization (adjusted odds ratio [AOR] 0.97, confidence interval [CI] 0.95 to 0.99) and a 3% reduction in odds of requiring oxygen supplementation (AOR 0.97, CI 0.95 to 0.99) and/or ventilation (AOR 0.97, CI 0.94 to 0.99). Ang 1-7 was also inversely associated with pro-inflammatory cytokines and d-dimer in this patient cohort, suggesting that reduced activity in this protective counter-regulatory arm of the RAS contributes to the hyper-immune response and diffuse coagulation activation documented in COVID-19. IMPORTANCE Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) causes a unique disease, COVID-19, which ranges in severity from asymptomatic to causing severe respiratory failure and death. Viral transmission throughout the world continues at a high rate despite the development and widespread use of effective vaccines. For those patients who contract COVID-19 and become severely ill, few therapeutic options have been shown to provide benefits and mortality rates are high. Additionally, the pathophysiology underlying COVID-19 disease presentation, progression, and severity is incompletely understood. The significance of our research is in confirming the role of renin-angiotensin system dysfunction in COVID-19 pathogenesis in a large cohort of patients with diverse disease severity and outcomes. Additionally, to our knowledge, this is the first study to pair angiotensin peptide levels with inflammatory and thrombotic markers. These data support the role of ongoing clinical trials examining renin-angiotensin system-targeted therapeutics for the treatment of COVID-19.

Non-linear machine learning models incorporating SNPs and PRS improve polygenic prediction in diverse human populations.

Polygenic risk scores (PRS) are commonly used to quantify the inherited susceptibility for a trait, yet they fail to account for non-linear and interaction effects between single nucleotide polymorphisms (SNPs). We address this via a machine learning approach, validated in nine complex phenotypes in a multi-ancestry population. We use an ensemble method of SNP selection followed by gradient boosted trees (XGBoost) to allow for non-linearities and interaction effects. We compare our results to the standard, linear PRS model developed using PRSice, LDpred2, and lassosum2. Combining a PRS as a feature in an XGBoost model results in a relative increase in the percentage variance explained compared to the standard linear PRS model by 22% for height, 27% for HDL cholesterol, 43% for body mass index, 50% for sleep duration, 58% for systolic blood pressure, 64% for total cholesterol, 66% for triglycerides, 77% for LDL cholesterol, and 100% for diastolic blood pressure. Multi-ancestry trained models perform similarly to specific racial/ethnic group trained models and are consistently superior to the standard linear PRS models. This work demonstrates an effective method to account for non-linearities and interaction effects in genetics-based prediction models.

Association of Intradialytic Hypotension and Ultrafiltration with AKI-D Outcomes in the Outpatient Dialysis Setting.

Identifying modifiable predictors of outcomes for cases of acute kidney injury requiring hemodialysis (AKI-D) will allow better care of patients with AKI-D. All patients with AKI-D discharged to University of Virginia (UVA) outpatient HD units between 1 January 2017 to 31 December 2019 (n = 273) were followed- for up to six months. Dialysis-related parameters were measured during the first 4 weeks of outpatient HD to test the hypothesis that modifiable factors during dialysis are associated with AKI-D outcomes of recovery, End Stage Kidney Disease (ESKD), or death. Patients were 42% female, 67% Caucasian, with mean age 62.8 ± 15.4 years. Median number of dialysis sessions was 11 (6-15), lasting 3.6 ± 0.6 h. At 90 days after starting outpatient HD, 45% recovered, 45% were declared ESKD and 9.9% died, with no significant changes noted between three and six months. Patients who recovered, died or were declared ESKD experienced an average of 9, 10 and 16 intradialytic hypotensive (IDH) episodes, respectively. More frequent IDH episodes were associated with increased risk of ESKD (p = 0.01). A one liter increment in net ultrafiltration was associated with 54% increased ratio of ESKD (p = 0.048). Optimizing dialysis prescription to decrease frequency of IDH episodes and minimize UF, and close monitoring of outpatient dialysis for patients with AKI-D, are crucial and may improve outcomes for these patients.

Attempts to conceive and the COVID-19 pandemic: data from the Apple Women's Health Study.

BACKGROUND: Previous studies have suggested that emergent events may affect pregnancy planning decisions. However, few have investigated the effect of factors related to the COVID-19 pandemic on pregnancy planning, measured by attempting conception, and how attempting conception status may differ by individual-level factors, such as social status or educational level. OBJECTIVE: This study aimed to examine the effects of factors related to the COVID-19 pandemic, until March 2021, on attempting conception status and to assess the effect measure modification by educational level and subjective social status. STUDY DESIGN: We conducted a longitudinal analysis within a subgroup of 21,616 participants in the Apple Women's Health Study who enrolled from November 2019 to March 2021, who met the inclusion criteria, and who responded to the monthly status menstrual update question on attempting conception status (yes or no). Participants reporting hysterectomy, pregnancy, lactation, or menopause were excluded. We used generalized estimating equation methodology to fit logistic regression models that estimate odds ratios and 95% confidence intervals for the association between the proportion of participants attempting conception and the month of response (compared with a prepandemic reference month of February 2020) while accounting for longitudinal correlation and adjusting for age, race and ethnicity, and marital status. We stratified the analysis by social status and educational level. RESULTS: We observed a trend of reduced odds of attempting conception, with an 18% reduction in the odds of attempting conception in August 2020 and October 2020 compared with the prepandemic month of February 2020 (August odds ratio: 0.82 [95% confidence interval, 0.70-0.97]; October odds ratio: 0.82 [95% confidence interval, 0.69-0.97). The participants with lower educational level (no college education) experienced a sustained reduction in the odds of attempting to conceive from June 2020 to March 2021 compared with February 2020, with up to a 24% reduction in the odds of attempting to conceive in October 2020 (odds ratio, 0.76; 95% confidence interval, 0.59-0.96). Among participants that were college educated, we observed an initial reduction in the odds of attempting to conceive starting in July 2020 (odds ratio 0.73; 95% confidence interval, 0.54-0.99) that returned near prepandemic odds. Moreover, we observed a reduction in the odds of attempting to conceive among those with low subjective social status, with a decline in the odds of attempting to conceive beginning in July 2020 (odds ratio, 0.83; 95% confidence interval, 0.63-1.10) and continuing until March 2021 (odds ratio, 0.79; 95% confidence interval, 0.59-1.06), with the greatest reduction in odds in October 2020 (odds ratio, 0.67; 95% confidence interval, 0.50-0.91). CONCLUSION: Among women in the Apple Women's Health Study cohort, our findings suggested a reduction in the odds of attempting to conceive during the COVID-19 pandemic, until March 2021, particularly among women of lower educational level and lower perceived social status.

Design and methods of the Apple Women's Health Study: a digital longitudinal cohort study.

BACKGROUND: Prospective longitudinal cohorts assessing women's health and gynecologic conditions have historically been limited. OBJECTIVE: The Apple Women's Health Study was designed to gain a deeper understanding of the relationship among menstrual cycles, health, and behavior. This paper describes the design and methods of the ongoing Apple Women's Health Study and provides the demographic characteristics of the first 10,000 participants. STUDY DESIGN: This was a mobile-application-based longitudinal cohort study involving survey and sensor-based data. We collected the data from 10,000 participants who responded to the demographics survey on enrollment between November 14, 2019 and May 20, 2020. The participants were asked to complete a monthly follow-up through November 2020. The eligibility included installed Apple Research app on their iPhone with iOS version 13.2 or later, were living in the United States, being of age greater than 18 years (19 in Alabama and Nebraska, 21 years old in Puerto Rico), were comfortable in communicating in written and spoken English, were the sole user of an iCloud account or iPhone, and were willing to provide consent to participate in the study. RESULTS: The mean age at enrollment was 33.6 years old (±standard deviation, 10.3). The race and ethnicity was representative of the US population (69% White and Non-Hispanic [6910/10,000]), whereas 51% (5089/10,000) had a college education or above. The participant geographic distribution included all the US states and Puerto Rico. Seventy-two percent (7223/10,000) reported the use of an Apple Watch, and 24.4% (2438/10,000) consented to sensor-based data collection. For this cohort, 38% (3490/9238) did not respond to the Monthly Survey: Menstrual Update after enrollment. At the 6-month follow-up, there was a 35% (3099/8972) response rate to the Monthly Survey: Menstrual Update. 82.7% (8266/10,000) of the initial cohort and 95.1% (2948/3099) of the participants who responded to month 6 of the Monthly Survey: Menstrual Update tracked at least 1 menstrual cycle via HealthKit. The participants tracked their menstrual bleeding days for an average of 4.44 (25%-75%; range, 3-6) calendar months during the study period. Non-White participants were slightly more likely to drop out than White participants; those remaining at 6 months were otherwise similar in demographic characteristics to the original enrollment group. CONCLUSION: The first 10,000 participants of the Apple Women's Health Study were recruited via the Research app and were diverse in race and ethnicity, educational attainment, and economic status, despite all using an Apple iPhone. Future studies within this cohort incorporating this high-dimensional data may facilitate discovery in women's health in exposure outcome relationships and population-level trends among iPhone users. Retention efforts centered around education, communication, and engagement will be utilized to improve the survey response rates, such as the study update feature.

Statewide Analysis Reveals Period of Well-Child Visit Attendance for Earlier Diagnosis of Autism Spectrum Disorder.

OBJECTIVE: To explore the relationship between well-child visit (WCV) attendance during early childhood and age at autism spectrum disorder (ASD) diagnosis using data drawn from a statewide all-payer claims database. STUDY DESIGN: We used a correlational study design with longitudinal data drawn from the Virginia All-Payer Claims Database. All children born in 2011 with a diagnosis of ASD were included (n = 253). Survival analysis determined the impact of WCV attendance on ASD diagnosis at each American Academy of Pediatrics-recommended early childhood visit, and the 5-year visit. RESULTS: Survival analysis revealed a significant impact of WCV attendance at the 24-month, 3-, and 4-year visits on earlier ASD diagnosis. Children who attended the 24-month visit were diagnosed nearly 10 months earlier than those who did not. Overall, children with ASD attended fewer than 50% of visits during early childhood. CONCLUSIONS: Promoting consistent WCV attendance during early childhood is an actionable strategy for improving early identification of ASD. Further exploration is needed to determine barriers to visit attendance and the impact of patterns of early childhood WCV attendance on age of ASD diagnosis. Development and implementation of interventions to promote adherence to the American Academy of Pediatrics-recommended visits is needed.

Developing the Total Health Profile, a Generalizable Unified Set of Multimorbidity Risk Scores Derived From Machine Learning for Broad Patient Populations: Retrospective Cohort Study.

BACKGROUND: Multimorbidity clinical risk scores allow clinicians to quickly assess their patients' health for decision making, often for recommendation to care management programs. However, these scores are limited by several issues: existing multimorbidity scores (1) are generally limited to one data group (eg, diagnoses, labs) and may be missing vital information, (2) are usually limited to specific demographic groups (eg, age), and (3) do not formally provide any granularity in the form of more nuanced multimorbidity risk scores to direct clinician attention. OBJECTIVE: Using diagnosis, lab, prescription, procedure, and demographic data from electronic health records (EHRs), we developed a physiologically diverse and generalizable set of multimorbidity risk scores. METHODS: Using EHR data from a nationwide cohort of patients, we developed the total health profile, a set of six integrated risk scores reflecting five distinct organ systems and overall health. We selected the occurrence of an inpatient hospital visitation over a 2-year follow-up window, attributable to specific organ systems, as our risk endpoint. Using a physician-curated set of features, we trained six machine learning models on 794,294 patients to predict the calibrated probability of the aforementioned endpoint, producing risk scores for heart, lung, neuro, kidney, and digestive functions and a sixth score for combined risk. We evaluated the scores using a held-out test cohort of 198,574 patients. RESULTS: Study patients closely matched national census averages, with a median age of 41 years, a median income of $66,829, and racial averages by zip code of 73.8% White, 5.9% Asian, and 11.9% African American. All models were well calibrated and demonstrated strong performance with areas under the receiver operating curve (AUROCs) of 0.83 for the total health score (THS), 0.89 for heart, 0.86 for lung, 0.84 for neuro, 0.90 for kidney, and 0.83 for digestive functions. There was consistent performance of this scoring system across sexes, diverse patient ages, and zip code income levels. Each model learned to generate predictions by focusing on appropriate clinically relevant patient features, such as heart-related hospitalizations and chronic hypertension diagnosis for the heart model. The THS outperformed the other commonly used multimorbidity scoring systems, specifically the Charlson Comorbidity Index (CCI) and the Elixhauser Comorbidity Index (ECI) overall (AUROCs: THS=0.823, CCI=0.735, ECI=0.649) as well as for every age, sex, and income bracket. Performance improvements were most pronounced for middle-aged and lower-income subgroups. Ablation tests using only diagnosis, prescription, social determinants of health, and lab feature groups, while retaining procedure-related features, showed that the combination of feature groups has the best predictive performance, though only marginally better than the diagnosis-only model on at-risk groups. CONCLUSIONS: Massive retrospective EHR data sets have made it possible to use machine learning to build practical multimorbidity risk scores that are highly predictive, personalizable, intuitive to explain, and generalizable across diverse patient populations.

U-Shaped Association between Maternal Hemoglobin and Low Birth Weight in Rural Bangladesh.

Low birth weight (LBW) is associated with a higher risk of neonatal mortality and the development of adult-onset chronic disease. Understanding the ongoing contribution of maternal hemoglobin (Hgb) levels to the incidence of LBW in South Asia is crucial to achieve the World Health Assembly global nutrition target of a 30% reduction in LBW by 2025. We enrolled pregnant women from the rural Tangail District of Bangladesh in a Maternal Newborn Health Registry established under The Global Network for Women's and Children's Health Research. We measured the Hgb of pregnant women at enrollment and birth weights of all infants born after 20 weeks gestation. Using logistic regression to adjust for multiple potential confounders, we estimated the association between maternal Hgb and the risk of LBW. We obtained Hgb measurements and birth weights from 1,665 mother-child dyads between July 2019 and April 2020. Using trimester-specific cutoffs for anemia, 48.3% of the women were anemic and the mean (±SD) Hgb level was 10.6 (±1.24) g/dL. We identified a U-shaped relationship where the highest risk of LBW was seen at very low (< 7.0 g/dL, OR = 2.00, 95% CI = 0.43-7.01, P = 0.31) and high (> 13.0 g/dL, OR = 2.17, 95% CI = 1.01-4.38, P = 0.036) Hgb levels. The mechanisms underlying this U-shaped association may include decreased plasma expansion during pregnancy and/or iron dysregulation resulting in placental disease. Further research is needed to explain the observed U-shaped relationship, to guide iron supplementation in pregnancy and to minimize the risk of LBW outcomes.

Identifying Communities at Risk for COVID-19-Related Burden Across 500 US Cities and Within New York City: Unsupervised Learning of the Coprevalence of Health Indicators.

BACKGROUND: Although it is well-known that older individuals with certain comorbidities are at the highest risk for complications related to COVID-19 including hospitalization and death, we lack tools to identify communities at the highest risk with fine-grained spatial resolution. Information collected at a county level obscures local risk and complex interactions between clinical comorbidities, the built environment, population factors, and other social determinants of health. OBJECTIVE: This study aims to develop a COVID-19 community risk score that summarizes complex disease prevalence together with age and sex, and compares the score to different social determinants of health indicators and built environment measures derived from satellite images using deep learning. METHODS: We developed a robust COVID-19 community risk score (COVID-19 risk score) that summarizes the complex disease co-occurrences (using data for 2019) for individual census tracts with unsupervised learning, selected on the basis of their association with risk for COVID-19 complications such as death. We mapped the COVID-19 risk score to corresponding zip codes in New York City and associated the score with COVID-19-related death. We further modeled the variance of the COVID-19 risk score using satellite imagery and social determinants of health. RESULTS: Using 2019 chronic disease data, the COVID-19 risk score described 85% of the variation in the co-occurrence of 15 diseases and health behaviors that are risk factors for COVID-19 complications among ~28,000 census tract neighborhoods (median population size of tracts 4091). The COVID-19 risk score was associated with a 40% greater risk for COVID-19-related death across New York City (April and September 2020) for a 1 SD change in the score (risk ratio for 1 SD change in COVID-19 risk score 1.4; P<.001) at the zip code level. Satellite imagery coupled with social determinants of health explain nearly 90% of the variance in the COVID-19 risk score in the United States in census tracts (r2=0.87). CONCLUSIONS: The COVID-19 risk score localizes risk at the census tract level and was able to predict COVID-19-related mortality in New York City. The built environment explained significant variations in the score, suggesting risk models could be enhanced with satellite imagery.

IL-13 is a driver of COVID-19 severity.

Immune dysregulation is characteristic of the more severe stages of SARS-CoV-2 infection. Understanding the mechanisms by which the immune system contributes to COVID-19 severity may open new avenues to treatment. Here, we report that elevated IL-13 was associated with the need for mechanical ventilation in 2 independent patient cohorts. In addition, patients who acquired COVID-19 while prescribed Dupilumab, a mAb that blocks IL-13 and IL-4 signaling, had less severe disease. In SARS-CoV-2-infected mice, IL-13 neutralization reduced death and disease severity without affecting viral load, demonstrating an immunopathogenic role for this cytokine. Following anti-IL-13 treatment in infected mice, hyaluronan synthase 1 (Has1) was the most downregulated gene, and accumulation of the hyaluronan (HA) polysaccharide was decreased in the lung. In patients with COVID-19, HA was increased in the lungs and plasma. Blockade of the HA receptor, CD44, reduced mortality in infected mice, supporting the importance of HA as a pathogenic mediator. Finally, HA was directly induced in the lungs of mice by administration of IL-13, indicating a new role for IL-13 in lung disease. Understanding the role of IL-13 and HA has important implications for therapy of COVID-19 and, potentially, other pulmonary diseases. IL-13 levels were elevated in patients with severe COVID-19. In a mouse model of the disease, IL-13 neutralization reduced the disease and decreased lung HA deposition. Administration of IL-13-induced HA in the lung. Blockade of the HA receptor CD44 prevented mortality, highlighting a potentially novel mechanism for IL-13-mediated HA synthesis in pulmonary pathology.