A primer on the use of machine learning to distil knowledge from data in biological psychiatry.

Applications of machine learning in the biomedical sciences are growing rapidly. This growth has been spurred by diverse cross-institutional and interdisciplinary collaborations, public availability of large datasets, an increase in the accessibility of analytic routines, and the availability of powerful computing resources. With this increased access and exposure to machine learning comes a responsibility for education and a deeper understanding of its bases and bounds, borne equally by data scientists seeking to ply their analytic wares in medical research and by biomedical scientists seeking to harness such methods to glean knowledge from data. This article provides an accessible and critical review of machine learning for a biomedically informed audience, as well as its applications in psychiatry. The review covers definitions and expositions of commonly used machine learning methods, and historical trends of their use in psychiatry. We also provide a set of standards, namely Guidelines for REporting Machine Learning Investigations in Neuropsychiatry (GREMLIN), for designing and reporting studies that use machine learning as a primary data-analysis approach. Lastly, we propose the establishment of the Machine Learning in Psychiatry (MLPsych) Consortium, enumerate its objectives, and identify areas of opportunity for future applications of machine learning in biological psychiatry. This review serves as a cautiously optimistic primer on machine learning for those on the precipice as they prepare to dive into the field, either as methodological practitioners or well-informed consumers.

Anti-PF4 antibodies associated with disease severity in COVID-19.

Severe COVID-19 is characterized by a prothrombotic state associated with thrombocytopenia, with microvascular thrombosis being almost invariably present in the lung and other organs at postmortem examination. We evaluated the presence of antibodies to platelet factor 4 (PF4)-polyanion complexes using a clinically validated immunoassay in 100 hospitalized patients with COVID-19 with moderate or severe disease (World Health Organization score, 4 to 10), 25 patients with acute COVID-19 visiting the emergency department, and 65 convalescent individuals. Anti-PF4 antibodies were detected in 95 of 100 hospitalized patients with COVID-19 (95.0%) irrespective of prior heparin treatment, with a mean optical density value of 0.871 ± 0.405 SD (range, 0.177 to 2.706). In contrast, patients hospitalized for severe acute respiratory disease unrelated to COVID-19 had markedly lower levels of the antibodies. In a high proportion of patients with COVID-19, levels of all three immunoglobulin (Ig) isotypes tested (IgG, IgM, and IgA) were simultaneously elevated. Antibody levels were higher in male than in female patients and higher in African Americans and Hispanics than in White patients. Anti-PF4 antibody levels were correlated with the maximum disease severity score and with significant reductions in circulating platelet counts during hospitalization. In individuals convalescent from COVID-19, the antibody levels returned to near-normal values. Sera from patients with COVID-19 induced higher levels of platelet activation than did sera from healthy blood donors, but the results were not correlated with the levels of anti-PF4 antibodies. These results demonstrate that the vast majority of patients with severe COVID-19 develop anti-PF4 antibodies, which may play a role in the clinical complications of COVID-19.

Impact of variable sampling on estimates of HIV-1 reservoir formation dates.

Timing of HIV-1 reservoir formation is important for informing HIV cure efforts. It is unclear how much of the variability seen in dating reservoir formation is due to sampling and gene-specific differences. We used a Bayesian extension of root to tip regression (bayroot) to re-estimate formation date distributions in participants from Swedish and South African cohorts, and assessed the impact of variable timing, frequency, and depth of sampling on these estimates. Significant shifts in formation date distributions were only observed with use of faster-evolving genes, while timing, frequency, and depth of sampling had minor or no significant effect on estimates.

Association of oral microbiome with oral human papillomavirus infection: a population study of the National Health and Nutrition Examination Survey, 2009-2012.

BACKGROUND: Oral human papillomavirus(HPV) infection and the oral microbiome are associated with oropharyngeal cancer. However, population-based data on the association of oral microbiome with oral HPV infection are limited. METHOD: We performed a cross-sectional analysis of 5,496 participants aged 20-59 in National Health and Nutrition Examination Surveys(NHANES):2009-2012. The association between either oral microbiome alpha diversity or beta diversity and oral HPV infection was assessed using multivariable logistic regression or principal coordinate analyses(PCoA) and multivariate analysis of variance(PERMANOVA). RESULTS: For alpha diversity, we found a lower number of observed Amplicon sequence variants(ASVs) (adjusted odds ratio[aOR] = 0.996; 95%CI = 0.992-0.999) and reduced Faith's Phylogenetic Diversity(aOR = 0.95; 95%CI = 0.90-0.99) associated with high-risk oral HPV infection in the overall population. This trend was observed in males for both high-risk and any oral HPV infection. Beta diversity showed differentiation of oral microbiome community by high-risk oral HPV infection as measured by Bray-Curtis dissimilarity (R2 = 0.054%; P = .029) and unweighted UniFrac distance (R2 = 0.046%; P = .045) among the overall population, and associations were driven by males. CONCLUSIONS: Both oral microbiome alpha diversity(within-sample richness and phylogenetic diversity) and beta diversity(heterogeneous dispersion of oral microbiome community) are associated with HPV infection. Longitudinal studies are needed to characterize the role of the microbiome in the natural history of oral HPV infection.

Incidence of Health Facility Switching and Associations With HIV Viral Rebound Among Persons on Antiretroviral Therapy in Uganda: A Population-based Study.

BACKGROUND: A substantial proportion of persons on antiretroviral therapy (ART) considered lost to follow-up have actually transferred their human immunodeficiency virus (HIV) care to other facilities. However, the relationship between facility switching and virologic outcomes, including viral rebound, is poorly understood. METHODS: We used data from 40 communities (2015-2020) in the Rakai Community Cohort Study to estimate incidence of facility switching and viral rebound. Persons aged 15-49 years with serologically confirmed HIV who self-reported ART use and contributed ≥1 follow-up visit were included. Facility switching and virologic outcomes were assessed between 2 consecutive study visits (ie, index and follow-up visits, interval of approximately 18 months). Those who reported different HIV treatment facilities between index and follow-up study visits were classified as having switched facilities. Virologic outcomes included viral rebound among individuals initially suppressed (<200 copies/mL). Multivariable Poisson regression was used to estimate associations between facility switching and viral rebound. RESULTS: Overall, 2257 persons who self-reported ART use (median age, 35 years; 65% female, 92% initially suppressed) contributed 3335 visit-pairs and 5959 person-years to the analysis. Facility switching was common (4.8 per 100 person-years; 95% confidence interval [CI], 4.2-5.5) and most pronounced in persons aged <30 years and fishing community residents. Among persons suppressed at their index visit (n = 2076), incidence of viral rebound was more than twice as high in persons who switched facilities (adjusted incidence rate ratio = 2.27; 95% CI, 1.16-4.45). CONCLUSIONS: Facility switching was common and associated with viral rebound among persons initially suppressed. Investments in more agile, person-centered models for mobile clients are needed to address system inefficiencies and bottlenecks that can disrupt HIV care continuity.

Association of Baseline MRI PI-RADS Score With Prostate Cancer Active Surveillance Early Biopsy Reclassification: Data From the Michigan Urological Surgery Improvement Collaborative (MUSIC).

PURPOSE: The purpose of our study was to evaluate the association of baseline MRI PI-RADS score with biopsy reclassification in a multicenter active surveillance cohort. METHODS: We identified men in the Michigan Urological Surgery Improvement Collaborative registry (46 hospital based/academic/private practice urology groups) with NCCN low risk and favorable intermediate risk prostate cancer who underwent MRI within 6 months before or after initial biopsy and enrolled in active surveillance (AS) from June 2016 to January 2021. The primary objective was to determine the association of baseline MRI PI-RADS score (≥4 lesion) with reclassification to high-grade prostate cancer (≥grade group 3) on surveillance biopsy. Multivariable Cox proportional hazards regression models were constructed and adjusted for pathologic, MRI, and clinical/biopsy factors, with landmark time of 6 months from diagnostic biopsy. We included an interaction term between PI-RADS score and NCCN group in the Cox model. RESULTS: Total of 1491 men were included with median age 64 years (interquartile range (IQ): 59-69) with median follow-up 11.0 months (IQ: 6.0-23.0) after landmark. Baseline PI-RADS ≥ 4 lesion was associated with an increased hazard of biopsy reclassification (Hazard ratio (HR): 2.3 (95% confidence interval (CI): 1.6-3.2), P < .001), along with grade group 2 vs 1 (HR: 2.5 (95%CI: 1.7-3.7), P < .001), and increasing age (per 10 years) (HR: 1.8 (95%CI: 1.4-2.4), P < .001). The interaction between NCCN risk group with MRI findings was not significant (P = .7). CONCLUSION: In this multicenter cohort study of real-world data, baseline MRI PI-RADS score was significantly associated with early biopsy reclassification in men undergoing active surveillance with NCCN low or favorable intermediate risk prostate cancer.

Distinct evolutionary lineages of Schistocephalus parasites infecting co-occurring sculpin and stickleback fishes in Alaska.

Sculpins (coastrange and slimy) and sticklebacks (ninespine and threespine) are widely distributed fishes cohabiting 2 south-central Alaskan lakes (Aleknagik and Iliamna), and all these species are parasitized by cryptic diphyllobothriidean cestodes in the genus Schistocephalus. The goal of this investigation was to test for host-specific parasitic relationships between sculpins and sticklebacks based upon morphological traits (segment counts) and sequence variation across the NADH1 gene. A total of 446 plerocercoids was examined. Large, significant differences in mean segment counts were found between cestodes in sculpin (mean = 112; standard deviation [s.d.] = 15) and stickleback (mean = 86; s.d. = 9) hosts within and between lakes. Nucleotide sequence divergence between parasites from sculpin and stickleback hosts was 20.5%, and Bayesian phylogenetic analysis recovered 2 well-supported clades of cestodes reflecting intermediate host family (i.e. sculpin, Cottidae vs stickleback, Gasterosteidae). Our findings point to the presence of a distinct lineage of cryptic Schistocephalus in sculpins from Aleknagik and Iliamna lakes that warrants further investigation to determine appropriate evolutionary and taxonomic recognition.

Trends in Adolescent Human Papillomavirus Vaccination and Parental Hesitancy in the United States.

BACKGROUND: Human papillomavirus (HPV) vaccination coverage remains suboptimal in the United States, underscoring the importance of monitoring trends in vaccine hesitancy. METHODS: Cross-sectional data from the 2011-2020 National Immunization Survey-Teen were used to assess trends in HPV vaccination initiation among 13-17-year-olds, parental intent to initiate vaccination, and primary reasons for parental hesitancy. RESULTS: Among all sex and race and ethnicity groups, the prevalence of HPV vaccination initiation increased over time, but parental intent to vaccinate against HPV for unvaccinated teens remained consistently low (≤45%). Among hesitant parents, "safety concerns" increased in nearly all demographic groups, with the greatest increases observed for non-Hispanic white female and male teens and no change for non-Hispanic black female teens. In 2019-2020, parents of unvaccinated non-Hispanic white teens were least likely to intend on vaccinating their teens, and the most common reason for hesitancy varied by sex and race and ethnicity (eg, "safety concerns" for white teens and "not necessary" for black female teens). CONCLUSIONS: Although HPV vaccination initiation increased over time, a substantial fraction of parents remain hesitant, and trends in their reason varied by sex and race and ethnicity. Health campaigns and clinicians should address vaccine safety and necessity.

Adolescent vaccination against human papillomavirus (HPV) is a critical tool for cancer prevention. We analyzed trends in HPV vaccination initiation among adolescents aged 13­17 years and trends in parental hesitancy to initiate HPV vaccination for their teen, using data from a national survey in the United States. Between 2011­2012 and 2019­2020, adolescent HPV vaccination initiation increased over time for both female teens (from 53.4% to 75.2%) and male teens (from 14.5% to 71.5%). However, the majority of parents/guardians of unvaccinated teens did not intend to vaccinate their teen against HPV (ie, were vaccine hesitant), and this was consistent over time in all sex and race and ethnicity groups. Among hesitant parents, the proportion reporting safety concerns as their main reason for being hesitant increased over time in nearly all demographic groups, with the greatest increases in this reasoning observed for white teens. In 2019­2020, parents of unvaccinated white teens were most likely to be vaccine hesitant. The most common reason for being vaccine hesitant also differed by sex and race and ethnicity. Although HPV vaccination has been shown to be safe and effective, HPV vaccination coverage remains suboptimal, and a substantial fraction of parents/guardians continue to be hesitant to adolescent HPV vaccination.

Age Patterns of HSV-2 Incidence and Prevalence in Two Ugandan Communities: A Catalytic Incidence Model Applied to Population-Based Seroprevalence Data.

BACKGROUND: Herpes simplex virus type 2 (HSV-2) is an incurable sexually transmitted infection associated with increased risk of acquiring and transmitting human immunodeficiency virus (HIV). HSV-2 is highly prevalent in sub-Saharan Africa, but population-level estimates of incidence are sparse. METHODS: We measured HSV-2 prevalence from cross-sectional serological data among adults aged 18-49 years in 2 south-central Uganda communities (fishing, inland). We identified risk factors for seropositivity, then inferred age patterns of HSV-2 with a Bayesian catalytic model. RESULTS: HSV-2 prevalence was 53.6% (n = 975/1819; 95% confidence interval, 51.3%-55.9%). Prevalence increased with age, was higher in the fishing community, and among women, reaching 93.6% (95% credible interval, 90.2%-96.6%) by age 49 years. Factors associated with HSV-2 seropositivity included more lifetime sexual partners, HIV positive status, and lower education. HSV-2 incidence peakied at age 18 years for women and 19-20 years for men. HIV prevalence was up to 10-fold higher in HSV-2-positive individuals. CONCLUSIONS: HSV-2 prevalence and incidence were extremely high, with most infections occurring in late adolescence. Interventions against HSV-2, such as future vaccines or therapeutics, must target young populations. Remarkably higher HIV prevalence among HSV-2-positive individuals underscores this population as a priority for HIV prevention.

HIV-Positive Liver Transplant Does not Alter the Latent Viral Reservoir in Recipients With Antiretroviral Therapy-Suppressed HIV.

The latent viral reservoir (LVR) remains a major barrier to HIV-1 curative strategies. It is unknown whether receiving a liver transplant from a donor with HIV might lead to an increase in the LVR because the liver is a large lymphoid organ. We found no differences in intact provirus, defective provirus, or the ratio of intact to defective provirus between recipients with ART-suppressed HIV who received a liver from a donor with (n = 19) or without HIV (n = 10). All measures remained stable from baseline by 1 year posttransplant. These data demonstrate that the LVR is stable after liver transplantation in people with HIV. Clinical Trials Registration. NCT02602262 and NCT03734393.

Absence of pathogenic viruses in COVID-19 convalescent plasma.

BACKGROUND: It is important to maintain the safety of blood products by avoiding the transfusion of units with known and novel viral pathogens. It is unknown whether COVID-19 convalescent plasma (CCP) may contain pathogenic viruses (either newly acquired or reactivated) that are not routinely screened for by blood centers. METHODS: The DNA virome was characterized in potential CCP donors (n = 30) using viral genome specific PCR primers to identify DNA plasma virome members of the Herpesviridae [Epstein Barr Virus (EBV), cytomegalovirus (CMV), human herpesvirus 6A/B, human herpesvirus 7] and Anelloviridae [Torque teno viruses (TTV), Torque teno mini viruses (TTMV), and Torque teno midi viruses (TTMDV)] families. In addition, the RNA plasma virome was characterized using unbiased metagenomic sequencing. Sequencing was done on a HiSeq2500 using high output mode with a read length of 2X100 bp. The sequencing reads were taxonomically classified using Kraken2. CMV and EBV seroprevalence were evaluated using a chemiluminescent immunoassay. RESULTS: TTV and TTMDV were detected in 12 (40%) and 4 (13%) of the 30 study participants, respectively; TTMDV was always associated with infection with TTV. We did not observe TTMV DNAemia. Despite CMV and EBV seroprevalences of 33.3% and 93.3%, respectively, we did not detect Herpesviridae DNA among the study participants. Metagenomic sequencing did not reveal any human RNA viruses in CCP, including no evidence of circulating SARS-CoV-2. DISCUSSION: There was no evidence of pathogenic viruses, whether newly acquired or reactivated, in CCP despite the presence of non-pathogenic Anelloviridae. These results confirm the growing safety data supporting CCP.

SARS-CoV-2 seroprevalence among blood donors in Uganda: 2019-2022.

BACKGROUND: The true burden of COVID-19 in low- and middle-income countries remains poorly characterized, especially in Africa. Even prior to the availability of SARS-CoV-2 vaccines, countries in Africa had lower numbers of reported COVID-19 related hospitalizations and deaths than other regions globally. METHODS: Ugandan blood donors were evaluated between October 2019 and April 2022 for IgG antibodies to SARS-CoV-2 nucleocapsid (N), spike (S), and five variants of the S protein using multiplexed electrochemiluminescence immunoassays (MesoScale Diagnostics, Rockville, MD). Seropositivity for N and S was assigned using manufacturer-provided cutoffs and trends in seroprevalence were estimated by quarter. Statistically significant associations between N and S antibody seropositivity and donor characteristics in November-December 2021 were assessed by chi-square tests. RESULTS: A total of 5393 blood unit samples from donors were evaluated. N and S seropositivity increased throughout the pandemic to 82.6% in January-April 2022. Among seropositive individuals, N and S antibody levels increased ≥9-fold over the study period. In November-December 2021, seropositivity to N and S antibody was higher among repeat donors (61.3%) compared with new donors (55.1%; p = .043) and among donors from Kampala (capital city of Uganda) compared with rural regions (p = .007). Seropositivity to S antibody was significantly lower among HIV-seropositive individuals (58.8% vs. 84.9%; p = .009). CONCLUSIONS: Despite previously reported low numbers of COVID-19 cases and related deaths in Uganda, high SARS-CoV-2 seroprevalence and increasing antibody levels among blood donors indicated that the country experienced high levels of infection over the course of the pandemic.

National Landscape of Human Immunodeficiency Virus-Positive Deceased Organ Donors in the United States.

BACKGROUND: Organ transplantation from donors with human immunodeficiency virus (HIV) to recipients with HIV (HIV D+/R+) presents risks of donor-derived infections. Understanding clinical, immunologic, and virologic characteristics of HIV-positive donors is critical for safety. METHODS: We performed a prospective study of donors with HIV-positive and HIV false-positive (FP) test results within the HIV Organ Policy Equity (HOPE) Act in Action studies of HIV D+/R+ transplantation (ClinicalTrials.gov NCT02602262, NCT03500315, and NCT03734393). We compared clinical characteristics in HIV-positive versus FP donors. We measured CD4 T cells, HIV viral load (VL), drug resistance mutations (DRMs), coreceptor tropism, and serum antiretroviral therapy (ART) detection, using mass spectrometry in HIV-positive donors. RESULTS: Between March 2016 and March 2020, 92 donors (58 HIV positive, 34 FP), representing 98.9% of all US HOPE donors during this period, donated 177 organs (131 kidneys and 46 livers). Each year the number of donors increased. The prevalence of hepatitis B (16% vs 0%), syphilis (16% vs 0%), and cytomegalovirus (CMV; 91% vs 58%) was higher in HIV-positive versus FP donors; the prevalences of hepatitis C viremia were similar (2% vs 6%). Most HIV-positive donors (71%) had a known HIV diagnosis, of whom 90% were prescribed ART and 68% had a VL <400 copies/mL. The median CD4 T-cell count (interquartile range) was 194/µL (77-331/µL), and the median CD4 T-cell percentage was 27.0% (16.8%-36.1%). Major HIV DRMs were detected in 42%, including nonnucleoside reverse-transcriptase inhibitors (33%), integrase strand transfer inhibitors (4%), and multiclass (13%). Serum ART was detected in 46% and matched ART by history. CONCLUSION: The use of HIV-positive donor organs is increasing. HIV DRMs are common, yet resistance that would compromise integrase strand transfer inhibitor-based regimens is rare, which is reassuring regarding safety.

HOPE in action: A prospective multicenter pilot study of liver transplantation from donors with HIV to recipients with HIV.

Liver transplantation (LT) from donors-with-HIV to recipients-with-HIV (HIV D+/R+) is permitted under the HOPE Act. There are only three international single-case reports of HIV D+/R+ LT, each with limited follow-up. We performed a prospective multicenter pilot study comparing HIV D+/R+ to donors-without-HIV to recipients-with-HIV (HIV D-/R+) LT. We quantified patient survival, graft survival, rejection, serious adverse events (SAEs), human immunodeficiency virus (HIV) breakthrough, infections, and malignancies, using Cox and negative binomial regression with inverse probability of treatment weighting. Between March 2016-July 2019, there were 45 LTs (8 simultaneous liver-kidney) at 9 centers: 24 HIV D+/R+, 21 HIV D-/R+ (10 D- were false-positive). The median follow-up time was 23 months. Median recipient CD4 was 287 cells/µL with 100% on antiretroviral therapy; 56% were hepatitis C virus (HCV)-seropositive, 13% HCV-viremic. Weighted 1-year survival was 83.3% versus 100.0% in D+ versus D- groups (p = .04). There were no differences in one-year graft survival (96.0% vs. 100.0%), rejection (10.8% vs. 18.2%), HIV breakthrough (8% vs. 10%), or SAEs (all p > .05). HIV D+/R+ had more opportunistic infections, infectious hospitalizations, and cancer. In this multicenter pilot study of HIV D+/R+ LT, patient and graft survival were better than historical cohorts, however, a potential increase in infections and cancer merits further investigation.

Antibody attributes that predict the neutralization and effector function of polyclonal responses to SARS-CoV-2.

BACKGROUND: While antibodies can provide significant protection from SARS-CoV-2 infection and disease sequelae, the specific attributes of the humoral response that contribute to immunity are incompletely defined. METHODS: We employ machine learning to relate characteristics of the polyclonal antibody response raised by natural infection to diverse antibody effector functions and neutralization potency with the goal of generating both accurate predictions of each activity based on antibody response profiles as well as insights into antibody mechanisms of action. RESULTS: To this end, antibody-mediated phagocytosis, cytotoxicity, complement deposition, and neutralization were accurately predicted from biophysical antibody profiles in both discovery and validation cohorts. These models identified SARS-CoV-2-specific IgM as a key predictor of neutralization activity whose mechanistic relevance was supported experimentally by depletion. CONCLUSIONS: Validated models of how different aspects of the humoral response relate to antiviral antibody activities suggest desirable attributes to recapitulate by vaccination or other antibody-based interventions.

Differentiation of Individuals Previously Infected with and Vaccinated for SARS-CoV-2 in an Inner-City Emergency Department.

Emergency departments (EDs) can serve as surveillance sites for infectious diseases. The objective of this study was to determine the burden of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection and to monitor the prevalence of vaccination against coronavirus disease 2019 (COVID-19) among patients attending an urban ED in Baltimore City. Using 1,914 samples of known exposure status, we developed an algorithm to differentiate previously infected, vaccinated, and unexposed individuals using a combination of antibody assays. We applied this testing algorithm to 4,360 samples from ED patients obtained in the spring of 2020 and 2021. Using multinomial logistic regression, we determined factors associated with infection and vaccination. For the algorithm, sensitivity and specificity for identifying vaccinated individuals were 100% and 99%, respectively, and 84% and 100% for previously infected individuals. Among the ED subjects, seroprevalence to SARS-CoV-2 increased from 2% to 24% between April 2020 and March 2021. Vaccination prevalence rose to 11% by mid-March 2021. Marked differences in burden of disease and vaccination coverage were seen by sex, race, and ethnicity. Hispanic patients, though accounting for 7% of the study population, had the highest relative burden of disease (17% of total infections) but with similar vaccination rates. Women and white individuals were more likely to be vaccinated than men or Black individuals. Individuals previously infected with SARS-CoV-2 can often be differentiated from vaccinated individuals using a serologic testing algorithm. The utility of this algorithm can aid in monitoring SARS-CoV-2 exposure and vaccination uptake frequencies and can potentially reflect gender, race, and ethnic health disparities.

Daily Written Care Summaries for Families of Critically Ill Patients: A Randomized Controlled Trial.

OBJECTIVES: To determine the effect of daily written updates on the satisfaction and psychologic symptoms of families of ICU patients. DESIGN: Randomized controlled trial. SETTING: Single, urban academic medical center. SUBJECTS: Surrogates of nondecisional, critically ill adults with high risk of mortality ( n = 252) enrolled from June 2019 to January 2021. INTERVENTIONS: Usual communication with the medical team with or without written communication detailing the suspected cause and management approach of each ICU problem, updated each day. MEASUREMENTS AND MAIN RESULTS: Participants completed surveys at three time points during the ICU stay: enrollment ( n = 252), 1 week ( n = 229), and 2 weeks ( n = 109) after enrollment. Satisfaction with care was measured using the Critical Care Family Needs Inventory (CCFNI). The presence of anxiety, depression, and acute stress were assessed using the Hospital Anxiety and Depression Scale (HADS) and Impact of Events Scale Revised (IES-R). CCFNI, HADS, and IES-R scores were similar among participants assigned to the intervention group and control group upon enrollment and during the first week after enrollment ( p > 0.05). From enrollment to the second week after enrollment, there was an improvement in CCFNI and HADS scores among participants assigned to the intervention group versus the control group. At week 2, CCFNI scores were significantly lower among participants in the intervention group versus the control group, indicating greater satisfaction with care: 15.1 (95% CI, 14.2-16.0) versus 16.4, (95% CI, 15.5-17.3); p = 0.04. In addition, 2 weeks after enrollment, the odds of symptoms of anxiety, depression, and acute stress among participants assigned to the intervention versus control group were 0.16 (95% CI, 0.03-0.82; p = 0.03); 0.15 (95% CI, 0.01-1.87; p = 0.14); and 0.27 (95% CI, 0.06-1.27; p = 0.10), respectively. CONCLUSIONS: Written communication improved satisfaction and the emotional well-being of families of critically ill patients, supporting its use as a supplement to traditional communication approaches.

Learning sparse log-ratios for high-throughput sequencing data.

MOTIVATION: The automatic discovery of sparse biomarkers that are associated with an outcome of interest is a central goal of bioinformatics. In the context of high-throughput sequencing (HTS) data, and compositional data (CoDa) more generally, an important class of biomarkers are the log-ratios between the input variables. However, identifying predictive log-ratio biomarkers from HTS data is a combinatorial optimization problem, which is computationally challenging. Existing methods are slow to run and scale poorly with the dimension of the input, which has limited their application to low- and moderate-dimensional metagenomic datasets. RESULTS: Building on recent advances from the field of deep learning, we present CoDaCoRe, a novel learning algorithm that identifies sparse, interpretable and predictive log-ratio biomarkers. Our algorithm exploits a continuous relaxation to approximate the underlying combinatorial optimization problem. This relaxation can then be optimized efficiently using the modern ML toolbox, in particular, gradient descent. As a result, CoDaCoRe runs several orders of magnitude faster than competing methods, all while achieving state-of-the-art performance in terms of predictive accuracy and sparsity. We verify the outperformance of CoDaCoRe across a wide range of microbiome, metabolite and microRNA benchmark datasets, as well as a particularly high-dimensional dataset that is outright computationally intractable for existing sparse log-ratio selection methods. AVAILABILITY AND IMPLEMENTATION: The CoDaCoRe package is available at https://github.com/egr95/R-codacore. Code and instructions for reproducing our results are available at https://github.com/cunningham-lab/codacore. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

GraphDTA: predicting drug-target binding affinity with graph neural networks.

SUMMARY: The development of new drugs is costly, time consuming and often accompanied with safety issues. Drug repurposing can avoid the expensive and lengthy process of drug development by finding new uses for already approved drugs. In order to repurpose drugs effectively, it is useful to know which proteins are targeted by which drugs. Computational models that estimate the interaction strength of new drug-target pairs have the potential to expedite drug repurposing. Several models have been proposed for this task. However, these models represent the drugs as strings, which is not a natural way to represent molecules. We propose a new model called GraphDTA that represents drugs as graphs and uses graph neural networks to predict drug-target affinity. We show that graph neural networks not only predict drug-target affinity better than non-deep learning models, but also outperform competing deep learning methods. Our results confirm that deep learning models are appropriate for drug-target binding affinity prediction, and that representing drugs as graphs can lead to further improvements. AVAILABILITY OF IMPLEMENTATION: The proposed models are implemented in Python. Related data, pre-trained models and source code are publicly available at https://github.com/thinng/GraphDTA. All scripts and data needed to reproduce the post hoc statistical analysis are available from https://doi.org/10.5281/zenodo.3603523. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

A de novo approach to inferring within-host fitness effects during untreated HIV-1 infection.

In the absence of effective antiviral therapy, HIV-1 evolves in response to the within-host environment, of which the immune system is an important aspect. During the earliest stages of infection, this process of evolution is very rapid, driven by a small number of CTL escape mutations. As the infection progresses, immune escape variants evolve under reduced magnitudes of selection, while competition between an increasing number of polymorphic alleles (i.e., clonal interference) makes it difficult to quantify the magnitude of selection acting upon specific variant alleles. To tackle this complex problem, we developed a novel multi-locus inference method to evaluate the role of selection during the chronic stage of within-host infection. We applied this method to targeted sequence data from the p24 and gp41 regions of HIV-1 collected from 34 patients with long-term untreated HIV-1 infection. We identify a broad distribution of beneficial fitness effects during infection, with a small number of variants evolving under strong selection and very many variants evolving under weaker selection. The uniquely large number of infections analysed granted a previously unparalleled statistical power to identify loci at which selection could be inferred to act with statistical confidence. Our model makes no prior assumptions about the nature of alleles under selection, such that any synonymous or non-synonymous variant may be inferred to evolve under selection. However, the majority of variants inferred with confidence to be under selection were non-synonymous in nature, and in most cases were have previously been associated with either CTL escape in p24 or neutralising antibody escape in gp41. We also identified a putative new CTL escape site (residue 286 in gag), and a region of gp41 (including residues 644, 648, 655 in env) likely to be associated with immune escape. Sites inferred to be under selection in multiple hosts have high within-host and between-host diversity although not all sites with high between-host diversity were inferred to be under selection at the within-host level. Our identification of selection at sites associated with resistance to broadly neutralising antibodies (bNAbs) highlights the need to fully understand the role of selection in untreated individuals when designing bNAb based therapies.