Visualizing machine learning-based predictions of postpartum depression risk for lay audiences

OBJECTIVES

To determine if different formats for conveying machine learning (ML)-derived postpartum depression risks impact patient classification of recommended actions (primary outcome) and intention to seek care, perceived risk, trust, and preferences (secondary outcomes).

MATERIALS AND METHODS

We recruited English-speaking females of childbearing age (18-45 years) using an online survey platform. We created 2 exposure variables (presentation format and risk severity), each with 4 levels, manipulated within-subject. Presentation formats consisted of text only, numeric only, gradient number line, and segmented number line. For each format viewed, participants answered questions regarding each outcome.

RESULTS

Five hundred four participants (mean age 31 years) completed the survey. For the risk classification question, performance was high (93%) with no significant differences between presentation formats. There were main effects of risk level (all P < .001) such that participants perceived higher risk, were more likely to agree to treatment, and more trusting in their obstetrics team as the risk level increased, but we found inconsistencies in which presentation format corresponded to the highest perceived risk, trust, or behavioral intention. The gradient number line was the most preferred format (43%).

DISCUSSION AND CONCLUSION

All formats resulted high accuracy related to the classification outcome (primary), but there were nuanced differences in risk perceptions, behavioral intentions, and trust. Investigators should choose health data visualizations based on the primary goal they want lay audiences to accomplish with the ML risk score.

Stress and DNA Methylation of Blood Leukocytes among Pregnant Latina Women

Latinas experience physical and psychological stressors in pregnancy leading to increased morbidity and higher risk for adverse birth outcomes. Epigenetic changes, including DNA methylation (DNAm), have been proposed as markers to create more refined risk stratification, yet few of these studies have examined these changes in Latinas. We conducted a secondary analysis of stored blood leukocytes of Latina women (n = 58) enrolled in a larger National Institutes of Health funded R01 project (2011-2016). We examined DNAm on eight candidate stress genes to compare physically and psychologically stressed participants to healthy (low stress) participants. We found unique CpGs that were differentially methylated in stressed women early- and mid-pregnancy compared to the healthy group, though none remained significant after FDR correction. Both physical and psychological stress were associated with hypomethylation at two consecutive CpG sites on NR3C1 in early pregnancy and one CpG site on NR3C1 in mid-pregnancy before adjustment. Stress was also associated with hypomethylation at two CpG sites on FKBP5 in early and mid-pregnancy but were no longer significant after FDR adjustment. Though we did not find statistically significant differences in DNAm during pregnancy between stressed and healthy women in this sample, signals were consistent with previous findings. Future work in larger samples should further examine the associations between stress and DNAm in pregnancy as this mechanism may explain underlying perinatal health inequities.

Women's perspectives on the use of artificial intelligence (AI)-based technologies in mental healthcare

This study aimed to evaluate women's attitudes towards artificial intelligence (AI)-based technologies used in mental health care. We conducted a cross-sectional, online survey of U.S. adults reporting female sex at birth focused on bioethical considerations for AI-based technologies in mental healthcare, stratifying by previous pregnancy. Survey respondents (n = 258) were open to AI-based technologies in mental healthcare but concerned about medical harm and inappropriate data sharing. They held clinicians, developers, healthcare systems, and the government responsible for harm. Most reported it was "very important" for them to understand AI output. More previously pregnant respondents reported being told AI played a small role in mental healthcare was "very important" versus those not previously pregnant (P = .03). We conclude that protections against harm, transparency around data use, preservation of the patient-clinician relationship, and patient comprehension of AI predictions may facilitate trust in AI-based technologies for mental healthcare among women.

The Power of Language in Hospital Care for Pregnant and Birthing People: A Vision for Change

Language is commonly defined as the principal method of human communication made up of words and conveyed by writing, speech, or nonverbal expression. In the context of clinical care, language has power and meaning and reflects priorities, beliefs, values, and culture. Stigmatizing language can communicate unintended meanings that perpetuate socially constructed power dynamics and result in bias. This bias may harm pregnant and birthing people by centering positions of power and privilege and by reflecting cultural priorities in the United States, including judgments of demographic and reproductive health characteristics. This commentary builds on relationship-centered care and reproductive justice frameworks to analyze the role and use of language in pregnancy and birth care in the United States, particularly regarding people with marginalized identities. We describe the use of language in written documentation, verbal communication, and behaviors associated with caring for pregnant people. We also present recommendations for change, including alternative language at the individual, clinician, hospital, health systems, and policy levels. We define birth as the emergence of a new individual from the body of its parent, no matter what intervention or pathology may be involved. Thus, we propose a cultural shift in hospital-based care for birthing people that centers the birthing person and reconceptualizes all births as physiologic events, approached with a spirit of care, partnership, and support.

The cationic region from HIV tat enhances the cell-surface expression of epitope/MHC class I complexes

The potential of genetic immunization has been acknowledged for almost a decade, but disappointing immunogenicity in humans has delayed its introduction into the clinical arena. To try to increase the potency of genetic immunization, we and others have evaluated 'translocatory' proteins, which are thought to exit living cells by an uncharacterized pathway, and enter neighboring cells in an energy-independent manner. Several laboratories, including our own, have begun to question these remarkable properties. Our previous studies showed that the ability of an epitope to induce major histocompatibility complex (MHC) class I restricted CD8(+) T cells was, indeed, enhanced by its being attached to the proposed translocatory sequence of the HIV-1 tat protein. However, we found little evidence that the increased immunogenicity resulted from transfer of the fusion peptide between living cells, and we proposed that it resulted instead from an increased epitope/MHC expression on the surface of transfected cells. Here, we directly test this hypothesis. We show that cells cotransfected with plasmids encoding an epitope, and the relevant MHC class I allele, can stimulate epitope-specific T cells, and that attachment of the epitope to a putative translocatory sequence - which we term herein an 'integral cationic region' (ICR) - leads to a marked increase in stimulatory activity. This elevated stimulatory capacity does not result from a nonspecific increase in MHC class I expression. We use a high-affinity T-cell receptor (TcR) specific for the epitope/MHC combination to quantitate directly the cell-surface expression of the immunogenic complex, and we show that the attachment of the tat ICR to an epitope results in a substantial enhancement of its cell-surface presentation. These data suggest an alternative explanation for the immune enhancement seen with ICRs.

Full-length proteins attached to the HIV tat protein transduction domain are neither transduced between cells, nor exhibit enhanced immunogenicity

Several proteins have been accorded the unusual ability to translocate across cell membranes in a receptor-independent and temperature-independent manner, and this activity has been mapped to a highly basic series of residues currently termed a 'protein transduction domain' (PTD). This translocatory attribute, if authentic, would be valuable for purposes of gene therapy and vaccination. We have evaluated the PTD from the human immunodeficiency virus type 1 (HIV) tat protein and we conclude that, when synthesized de novo, (1) the HIV tat PTD does not enhance the immunogenicity of a full-length protein to which it is tethered; and (2) the HIV tat PTD does not cause intercellular transfer of an attached marker protein, as judged by careful quantitative analyses. From our data, and from a review of published materials, we suggest that contrary to current dogma there is little evidence that these supposedly translocatory proteins can move between live cells. Furthermore, we suggest that PTDs do not act to enhance translocation, but instead merely to increase binding to the cell surface; in which case, the term 'protein transduction domain', and the related acronym, are misnomers which should be abandoned. Our conclusions explain why the most dramatic demonstrations of PTD efficacy have been obtained using fixed cells and/or denatured proteins, and have obvious implications for gene therapy and vaccination.

Vehicle maintenance and inspections

Your customers look to you to provide emergency medical assistance. They expect you to be there with the right equipment when they are in need. To meet those expectations, you must spend the time necessary to inspect and maintain your vehicles and the lifesaving equipment they carry. Documentation of inspections and the maintenance performed is also important. It shows the steps you have taken to ensure your equipment is in proper working order, and can help protect you should you ever need to prove that.

CD4(+) T cells induced by a DNA vaccine: immunological consequences of epitope-specific lysosomal targeting

Our previous studies have shown that targeting DNA vaccine-encoded major histocompatibility complex class I epitopes to the proteasome enhanced CD8(+) T-cell induction and protection against lymphocytic choriomeningitis virus (LCMV) challenge. Here, we expand these studies to evaluate CD4(+) T-cell responses induced by DNA immunization and describe a system for targeting proteins and minigenes to lysosomes. Full-length proteins can be targeted to the lysosomal compartment by covalent attachment to the 20-amino-acid C-terminal tail of lysosomal integral membrane protein-II (LIMP-II). Using minigenes encoding defined T-helper epitopes from lymphocytic choriomeningitis virus, we show that the CD4(+) T-cell response induced by the NP(309-328) epitope of LCMV was greatly enhanced by addition of the LIMP-II tail. However, the immunological consequence of lysosomal targeting is not invariably positive; the CD4(+) T-cell response induced by the GP(61-80) epitope was almost abolished when attached to the LIMP-II tail. We identify the mechanism which underlies this marked difference in outcome. The GP(61-80) epitope is highly susceptible to cleavage by cathepsin D, an aspartic endopeptidase found almost exclusively in lysosomes. We show, using mass spectrometry, that the GP(61-80) peptide is cleaved between residues F(74) and K(75) and that this destroys its ability to stimulate virus-specific CD4(+) T cells. Thus, the immunological result of lysosomal targeting varies, depending upon the primary sequence of the encoded antigen. We analyze the effects of CD4(+) T-cell priming on the virus-specific antibody and CD8(+) T-cell responses which are mounted after virus infection and show that neither response appears to be accelerated or enhanced. Finally, we evaluate the protective benefits of CD4(+) T-cell vaccination in the LCMV model system; in contrast to DNA vaccine-induced CD8(+) T cells, which can confer solid protection against LCMV challenge, DNA vaccine-mediated priming of CD4(+) T cells does not appear to enhance the vaccinee's ability to combat viral challenge.

Two overlapping subdominant epitopes identified by DNA immunization induce protective CD8(+) T-cell populations with differing cytolytic activities

Subdominant CD8(+) T-cell responses contribute to control of several viral infections and to vaccine-induced immunity. Here, using the lymphocytic choriomeningitis virus model, we demonstrate that subdominant epitopes can be more reliably identified by DNA immunization than by other methods, permitting the identification, in the virus nucleoprotein, of two overlapping subdominant epitopes: one presented by L(d) and the other presented by K(d). This subdominant sequence confers immunity as effective as that induced by the dominant epitope, against which >90% of the antiviral CD8(+) T cells are normally directed. We compare the kinetics of the dominant and subdominant responses after vaccination with those following subsequent viral infection. The dominant CD8(+) response expands more rapidly than the subdominant responses, but after virus infection is cleared, mice which had been immunized with the "dominant" vaccine have a pool of memory T cells focused almost entirely upon the dominant epitope. In contrast, after virus infection, mice which had been immunized with the "subdominant" vaccine retain both dominant and subdominant memory cells. During the acute phase of the immune response, the acquisition of cytokine responsiveness by subdominant CD8(+) T cells precedes their development of lytic activity. Furthermore, in both dominant and subdominant populations, lytic activity declines more rapidly than cytokine responsiveness. Thus, the lysis(low)-cytokine(competent) phenotype associated with most memory CD8(+) T cells appears to develop soon after antigen clearance. Finally, lytic activity differs among CD8(+) T-cell populations with different epitope specificities, suggesting that vaccines can be designed to selectively induce CD8(+) T cells with distinct functional attributes.

Coxsackievirus infection of the pancreas: evaluation of receptor expression, pathogenesis, and immunopathology

Coxsackievirus type B (CVB) infection of the pancreas induces a massive cellular infiltrate composed of natural killer cells, T cells, and macrophages and leads to the destruction of exocrine tissue. The physiological manifestations of pancreatic CVB infection are correlated with viral tropism; the virus infects acinar cells but spares the islets of Langerhans. Here we evaluate the mechanisms underlying pancreatic inflammation and destruction and identify the determinants of viral tropism. T-cell-mediated immunopathology has been invoked, along with direct virus-mediated cytopathicity, to explain certain aspects of CVB-induced pancreatic disease. However, we show here that in the pancreas, the extent of inflammation and tissue destruction appears unaltered in the absence of the cytolytic protein perforin; these findings exclude any requirement for perforin-mediated lysis by natural killer cells or cytotoxic T cells in CVB3-induced pancreatic damage. Furthermore, perforin-mediated cytotoxic T-cell activity does not contribute to the control of CVB infection in this organ. In addition, we demonstrate that the recently identified coxsackie-adenovirus receptor is expressed at high levels in acinar cells but is barely detectable in islets, which is consistent with its being a major determinant of virus tropism and, therefore, of disease. However, further studies using various cell lines of pancreatic origin reveal secondary determinants of virus tropism.

Quantitative and qualitative analyses of the immune responses induced by a multivalent minigene DNA vaccine

Vaccines containing minigenes - isolated antigenic epitopes encoded by short open reading frames - can, under certain circumstances, confer protective immunity upon the vaccinee. Here we evaluate the efficacy of the minigene vaccine approach using DNA immunization and find that, to be immunogenic, a minigene-encoded epitope requires a perfect "Kozak" translational initiation region. In addition, using intracellular cytokine staining, we show that immunization with a plasmid encoding a full-length protein induces epitope-specific CD8(+) T cells which are detectable directly ex vivo, and constitute approximately 2% of the vaccinee's splenic CD8(+) T cells. In contrast, such cells are undetectable directly ex vivo in recipients of a minigene vaccine. Nevertheless, the minigene plasmid does induce a low number of epitope-specific CD8(+) T cells, which can be amplified to detectable levels by in vivo stimulation. Indeed, 4 days after in vivo stimulation (by virus infection), all vaccinated mice - regardless of whether they had been vaccinated with the minigene or with the full-length gene - had similar numbers of epitope-specific CD8(+) T cells. However, despite these strong responses at 4 days post-infection, recipients of the minigene vaccine showed no enhanced ability to limit virus replication and dissemination. We therefore observe a dichotomy; minigene vaccinees are not protected, despite the presence of strong virus-specific immune responses at 4 days post-challenge. We suggest that the protective benefits of vaccination exert themselves very soon - perhaps within minutes or hours - after virus challenge. If the vaccine-induced immune response is too low to achieve this early protective effect, virus-specific T cells will expand rapidly, but ineffectually, leading to the strong but non-protective response measured at 4 days post-infection. Thus, vaccine-induced immunity should be monitored very early in infection, since the extent to which these responses may later be amplified is largely irrelevant to the protection observed.

Virus-induced diabetes in a transgenic model: role of cross-reacting viruses and quantitation of effector T cells needed to cause disease

Virus-specific cytotoxic T lymphocytes (CTL) at frequencies of >1/1, 000 are sufficient to cause insulin-dependent diabetes mellitus (IDDM) in transgenic mice whose pancreatic beta cells express as "self" antigen a protein from a virus later used to initiate infection. The inability to generate sufficient effector CTL for other cross-reacting viruses that fail to cause IDDM could be mapped to point mutations in the CTL epitope or its COO(-) flanking region. These data indicate that IDDM and likely other autoimmune diseases are caused by a quantifiable number of T cells, that neither standard epidemiologic markers nor molecular analysis with nucleic acid probes reliably distinguishes between viruses that do or do not cause diabetes, and that a single-amino-acid change flanking a CTL epitope can interfere with antigen presentation and development of autoimmune disease in vivo.

The role of B lymphocytes in coxsackievirus B3 infection

Coxsackieviruses are important human pathogens, frequently causing myocarditis, pancreatitis, and a variety of less severe diseases. B lymphocytes appear central to the interaction between these viruses and their mammalian hosts, because agammaglobulinemic humans, genetically incapable of antibody production, are susceptible to chronic infections by coxsackieviruses and related enteroviruses, such as poliovirus and echovirus. However, recent studies show that Type B coxsackievirus (CVB) infects B lymphocytes soon after infection, suggesting the possibility that these cells may play some role in virus dissemination and/or that the virus may be able to modulate the host immune response. We analyzed the role of B lymphocytes in CVB infection and confirmed that CVB infects B lymphocytes, and extended these findings to show that this is a productive infection involving approximately 1 to 10% of the cells; however, infectious center assays show that other splenocytes are infected at approximately the same frequency. Virus is readily detectable by in situ hybridization in the spleen of immunocompetent mice but is difficult to detect in mice deficient in B cells (BcKO mice), consistent with much of the splenic signal being the result of B cell infection. Surprisingly, given the extent of their infection, B cells express barely detectable levels of the murine coxsackievirus-adenovirus receptor (mCAR), suggesting that another means of cell entry may be used. We found no evidence of B cell depletion following CVB infection, indicating that this is not the explanation for the transient immunosuppression previously reported. Virus replication and dissemination are slightly delayed in BcKO mice, consistent with B cells' playing a role as an important early target of infection and/or a means to distribute the virus to many tissues. In addition, we show that BcKO mice recapitulate a central feature of human agammaglobulinemia: CVB establishes chronic infection in a variety of organs (heart, liver, brain, kidney, lung, pancreas, spleen). In most of these tissues the viral titers remain high (10(5)-10(8) plaque forming units (pfu) per gram of tissue) for the life of the mouse, and in several there is severe pathology, particularly severe myocardial fibrosis with ventricular dilation, reminiscent of the dilated cardiomyopathy seen in humans with chronic enteroviral myocarditis. Transfer of B and/or T cells from non-immune mice had no discernible effect, whereas equivalent transfers from immune mice often resulted in transient or permanent disappearance of detectable CVB.

Coxsackievirus B3-induced myocarditis: perforin exacerbates disease, but plays no detectable role in virus clearance

Viral myocarditis is remarkably common, being detected in approximately 1% of unselected asymptomatic individuals. Many cases are attributable to enteroviral infection, and in particular to coxsackievirus B3. The underlying pathogenesis is controversial, but most studies admit the important immunopathological role of infiltrating CD8+ (cytotoxic) T lymphocytes (CTLs). We have previously shown that CTLs play conflicting roles in coxsackievirus B (CVB) myocarditis; they assist in controlling virus replication, but also are instrumental in causing the extensive inflammatory disease, which often results in severe myocardial scarring. A role for perforin, the major CTL cytolytic protein, in CVB myocarditis has been suggested, but never proven. In the present study we use perforin knockout (PKO) mice to show that perforin plays a major role in CVB infection; in broad terms, perforin is important in immunopathology, but not in CVB clearance. For example, PKO mice are better able to withstand a normally lethal dose of CVB (100% survival of PKO mice compared with 90% death in +/+ littermates). In addition, PKO mice given a nonlethal dose of CVB develop only a mild myocarditis, whereas their perforin+ littermates have extensive myocardial lesions. The myocarditis in PKO mice resolves more quickly, and these mice show minimal histological sequelae; in contrast, late in disease the perforin+ mice develop severe myocardial fibrosis. PKO mice, despite lacking this major CTL effector function, can control the infection and eradicate the virus; growth kinetics and peak CVB titers are indistinguishable in PKO and perforin+ mice. Therefore, the immunopathological and antiviral effects of CTLs can be uncoupled by ablation of perforin; this offers a promising target for therapy of myocarditis. Furthermore, we evaluate the possible roles of apoptosis, and of chemokine expression, in CVB infection. In perforin+ mice, apoptotic cells are detected within the inflammatory infiltrate, whereas in their PKO counterparts, apoptotic myocyte nuclei are seen. Chemokine expression in both PKO and perforin+ mice precedes and parallels the course of myocarditis. Several chemokines are detectable earlier in PKO mice than in perforin+ mice, but PKO mice show reduced peak levels, and chemokine expression decays sooner. In particular, MIP-1alpha expression is barely detectable at any time point in PKO mice, but it is readily identified in perforin+ animals, peaking just before the time of maximal myocarditis; this is particularly interesting, given that MIP-1alpha knockout mice are resistant to CVB myocarditis, but remain able to control viral infection. Thus, the chemokine pathway offers a second route of intervention to diminish myocarditis and its sequelae, while permitting the host to eradicate the virus.

DNA immunization with minigenes: low frequency of memory cytotoxic T lymphocytes and inefficient antiviral protection are rectified by ubiquitination

Our previous studies have shown that isolated cytotoxic T lymphocyte (CTL), B-cell, and T-helper epitopes, for which we coined the term minigenes, can be effective vaccines; when expressed from recombinant vaccinia viruses, these short immunogenic sequences confer protection against a variety of viruses and bacteria. In addition, we have previously demonstrated the utility of DNA immunization using plasmids encoding full-length viral proteins. Here we combine the two approaches and evaluate the effectiveness of minigenes in DNA immunization. We find that DNA immunization with isolated minigenes primes virus-specific memory CTL responses which, 4 days following virus challenge, appear similar in magnitude to those induced by vaccines known to be protective. Surprisingly, this vigorous CTL response fails to confer protection against a normally lethal virus challenge, although the CTL appear fully functional because, along with their high lytic activity, they are similar in affinity and cytokine secretion to CTL induced by virus infection. However this DNA immunization with isolated minigenes results in a low CTL precursor frequency; only 1 in approximately 40,000 T cells is epitope specific. In contrast, a plasmid encoding the same minigene sequences covalently attached to the cellular protein ubiquitin induces protective immunity and a sixfold-higher frequency of CTL precursors. Thus, we show that the most commonly employed criterion to evaluate CTL responses-the presence of lytic activity following secondary stimulation-does not invariably correlate with protection; instead, the better correlate of protection is the CTL precursor frequency. Recent observations indicate that certain effector functions are active in memory CTL and do not require prolonged stimulation. We suggest that these early effector functions of CTL, immediately following infection, are critical in controlling virus dissemination and in determining the outcome of the infection. Finally, we show that improved performance of the ubiquitinated minigenes most probably requires polyubiquitination of the fusion protein, suggesting that the enhancement results from more effective delivery of the minigene to the proteasome.

Use of cross-linked bovine pericardium as a disc replacement in the rabbit temporomandibular joint

Chemically purified bovine pericardial membrane, highly cross-linked with monomeric glutaraldehyde, was implanted in the TMJ area in ten adult rabbits in place of partially excised discs. Sham controls had the temporal aspect of the TMJ scraped with a scalpel. After 3 months, the whole TMJ area was processed for morphologic and scanning electron micrographic (SEM) evaluation. Evidence was found of the disintegration of the implanted pericardium, which was extensively infiltrated by inflammatory cells. The collagen matrix of the membrane was distended and dissociated, showing signs of degradation by its stainability with trichrome stain. It was concluded that cross-linking of a collagenous structure such as pericardium does not prevent it from being resorbed. It is suggested that the cytotoxic residues of polymeric glutaraldehyde within the implant may render the implant more resistant to cellular degradation.