A Conference (Missingness in Action) to Address Missingness in Data and AI in Health Care: Qualitative Thematic Analysis.

BACKGROUND: Missingness in health care data poses significant challenges in the development and implementation of artificial intelligence (AI) and machine learning solutions. Identifying and addressing these challenges is critical to ensuring the continued growth and accuracy of these models as well as their equitable and effective use in health care settings. OBJECTIVE: This study aims to explore the challenges, opportunities, and potential solutions related to missingness in health care data for AI applications through the conduct of a digital conference and thematic analysis of conference proceedings. METHODS: A digital conference was held in September 2022, attracting 861 registered participants, with 164 (19%) attending the live event. The conference featured presentations and panel discussions by experts in AI, machine learning, and health care. Transcripts of the event were analyzed using the stepwise framework of Braun and Clark to identify key themes related to missingness in health care data. RESULTS: Three principal themes-data quality and bias, human input in model development, and trust and privacy-emerged from the analysis. Topics included the accuracy of predictive models, lack of inclusion of underrepresented communities, partnership with physicians and other populations, challenges with sensitive health care data, and fostering trust with patients and the health care community. CONCLUSIONS: Addressing the challenges of data quality, human input, and trust is vital when devising and using machine learning algorithms in health care. Recommendations include expanding data collection efforts to reduce gaps and biases, involving medical professionals in the development and implementation of AI models, and developing clear ethical guidelines to safeguard patient privacy. Further research and ongoing discussions are needed to ensure these conclusions remain relevant as health care and AI continue to evolve.

A Stanford Conference on Social Media, Ethics, and COVID-19 Misinformation (INFODEMIC): Qualitative Thematic Analysis.

BACKGROUND: The COVID-19 pandemic continues to challenge the world's population, with approximately 266 million cases and 5 million deaths to date. COVID-19 misinformation and disinformation led to vaccine hesitancy among the public, particularly in vulnerable communities, which persists today. Social media companies are attempting to curb the ongoing spread of an overwhelming amount of COVID-19 misinformation on their platforms. In response to this problem, the authors hosted INFODEMIC: A Stanford Conference on Social Media and COVID-19 Misinformation (INFODEMIC) to develop best practices for social media companies to mitigate online misinformation and disinformation. OBJECTIVE: The primary aim of this study was to develop recommendations for social media companies to address the COVID-19 infodemic. We report the methods used to execute the INFODEMIC conference, conference attendee engagement and analytics, and a qualitative thematic analysis of the conference presentations. The primary study outcomes were the identified themes and corresponding recommendations. METHODS: Using a constructivist paradigm, we conducted a thematic analysis of the 6-hour conference transcript to develop best practice recommendations. The INFODEMIC conference was the study intervention, the conference speakers were the study participants, and transcripts of their presentations were the data for this study. We followed the 6-step framework for thematic analysis described by Braun and Clarke. We also used descriptive statistics to report measures of conference engagement including registrations, viewership, post-conference asynchronous participation, and conference evaluations. RESULTS: A total of 26 participants spoke at the virtual conference and represented a wide array of occupations, expertise, and countries of origin. From their remarks, we identified 18 response categories and 4 themes: trust, equity, social media practices, and interorganizational partnerships. From these, a total of 16 best practice recommendations were formulated for social media companies, health care organizations, and the general public. These recommendations focused on rebuilding trust in science and medicine among certain communities, redesigning social media platforms and algorithms to reduce the spread of misinformation, improving partnerships between key stakeholders, and educating the public to critically analyze online information. Of the 1090 conference registrants, 587 (53.9%) attended the live conference, and another 9996 individuals viewed or listened to the conference recordings asynchronously. Conference evaluations averaged 8.9 (best=10). CONCLUSIONS: Social media companies play a significant role in the COVID-19 infodemic and should adopt evidence-based measures to mitigate misinformation on their platforms.

A Deadly Infodemic: Social Media and the Power of COVID-19 Misinformation.

COVID-19 is currently the third leading cause of death in the United States, and unvaccinated people continue to die in high numbers. Vaccine hesitancy and vaccine refusal are fueled by COVID-19 misinformation and disinformation on social media platforms. This online COVID-19 infodemic has deadly consequences. In this editorial, the authors examine the roles that social media companies play in the COVID-19 infodemic and their obligations to end it. They describe how fake news about the virus developed on social media and acknowledge the initially muted response by the scientific community to counteract misinformation. The authors then challenge social media companies to better mitigate the COVID-19 infodemic, describing legal and ethical imperatives to do so. They close with recommendations for better partnerships with community influencers and implementation scientists, and they provide the next steps for all readers to consider. This guest editorial accompanies the Journal of Medical Internet Research special theme issue, "Social Media, Ethics, and COVID-19 Misinformation."

Randomized Control Trial of the 3Rs Health Knowledge Training Program for Persons with Intellectual Disabilities.

BACKGROUND: Persons with intellectual disabilities (ID) experience a wide range of health problems. Research is needed on teaching persons with intellectual disabilities about their health to promote self-advocacy. This study used a RCT to evaluate a health knowledge training program for adults with intellectual disabilities and verbal skills. METHODS: Participants were randomly assigned to training (n = 12) or no training control (n = 10) groups. Topics included key body organs, systems, functions, health maintenance and illnesses. Participants played a game answering questions (e.g. 'What does the heart do?'). Instruction involved visuals (e.g., PowerPoint slides), cueing, modelling and feedback. The control group received pre-, post- and follow-up tests as the training group. RESULTS: The training group had significantly higher overall post-test and follow-up health knowledge test scores than the control group. CONCLUSIONS: Health knowledge training is one step in promoting health self-advocacy and better health in persons with intellectual disabilities.

Can you ask? We just did! Assessing sexual function and concerns in patients presenting for initial gynecologic oncology consultation.

OBJECTIVES: To describe patterns of response to, and assess sexual function and activity elicited by, a self-administered assessment incorporated into a new patient intake form for gynecologic oncology consultation. METHODS: A cross-sectional study of patients presenting to a single urban academic medical center between January 2010 and September 2012. New patients completed a self-administered intake form, including six brief sexual activity and function items. These items, along with abstracted medical record data, were descriptively analyzed. Logistic regression was used to assess the association between sexual activity and function and disease status, adjusting for age. RESULTS: Median age was 50 years (range 18-91, N=499); more than half had a final diagnosis of cancer. Most patients completed all sex-related items on the intake form; 98% answered at least one. Among patients who were sexually active in the prior 12 months (57% with cancer, 64% with benign disease), 52% indicated on the intake form having, during that period, a sexual problem lasting several months or more. Of these, 15% had physician documentation of the sexual problem. Eighteen women were referred for care. Providers reported no patient complaints about the inclusion of sexual items on the intake form. CONCLUSIONS: Nearly all new patients presenting for gynecologic oncology consultation answered self-administered items to assess sexual activity and function. Further study is needed to determine the role of pre-treatment identification of sexual function concerns in improving sexual outcomes associated with cancer diagnosis and treatment.

Interaction of insulin and PPAR-α genes in Alzheimer's disease: the Epistasis Project.

Altered glucose metabolism has been described in Alzheimer's disease (AD). We re-investigated the interaction of the insulin (INS) and the peroxisome proliferator-activated receptor alpha (PPARA) genes in AD risk in the Epistasis Project, including 1,757 AD cases and 6,294 controls. Allele frequencies of both SNPs (PPARA L162V, INS intron 0 A/T) differed between Northern Europeans and Northern Spanish. The PPARA 162LL genotype increased AD risk in Northern Europeans (p = 0.04), but not in Northern Spanish (p = 0.2). There was no association of the INS intron 0 TT genotype with AD. We observed an interaction on AD risk between PPARA 162LL and INS intron 0 TT genotypes in Northern Europeans (Synergy factor 2.5, p = 0.016), but not in Northern Spanish. We suggest that dysregulation of glucose metabolism contributes to the development of AD and might be due in part to genetic variations in INS and PPARA and their interaction especially in Northern Europeans.

The dopamine ß-hydroxylase -1021C/T polymorphism is associated with the risk of Alzheimer's disease in the Epistasis Project.

BACKGROUND: The loss of noradrenergic neurones of the locus coeruleus is a major feature of Alzheimer's disease (AD). Dopamine ß-hydroxylase (DBH) catalyses the conversion of dopamine to noradrenaline. Interactions have been reported between the low-activity -1021T allele (rs1611115) of DBH and polymorphisms of the pro-inflammatory cytokine genes, IL1A and IL6, contributing to the risk of AD. We therefore examined the associations with AD of the DBH -1021T allele and of the above interactions in the Epistasis Project, with 1757 cases of AD and 6294 elderly controls. METHODS: We genotyped eight single nucleotide polymorphisms (SNPs) in the three genes, DBH, IL1A and IL6. We used logistic regression models and synergy factor analysis to examine potential interactions and associations with AD. RESULTS: We found that the presence of the -1021T allele was associated with AD: odds ratio = 1.2 (95% confidence interval: 1.06-1.4, p = 0.005). This association was nearly restricted to men < 75 years old: odds ratio = 2.2 (1.4-3.3, 0.0004). We also found an interaction between the presence of DBH -1021T and the -889TT genotype (rs1800587) of IL1A: synergy factor = 1.9 (1.2-3.1, 0.005). All these results were consistent between North Europe and North Spain. CONCLUSIONS: Extensive, previous evidence (reviewed here) indicates an important role for noradrenaline in the control of inflammation in the brain. Thus, the -1021T allele with presumed low activity may be associated with misregulation of inflammation, which could contribute to the onset of AD. We suggest that such misregulation is the predominant mechanism of the association we report here.

The sex-specific associations of the aromatase gene with Alzheimer's disease and its interaction with IL10 in the Epistasis Project.

Epistasis between interleukin-10 (IL10) and aromatase gene polymorphisms has previously been reported to modify the risk of Alzheimer's disease (AD). However, although the main effects of aromatase variants suggest a sex-specific effect in AD, there has been insufficient power to detect sex-specific epistasis between these genes to date. Here we used the cohort of 1757 AD patients and 6294 controls in the Epistasis Project. We replicated the previously reported main effects of aromatase polymorphisms in AD risk in women, for example, adjusted odds ratio of disease for rs1065778 GG=1.22 (95% confidence interval: 1.01-1.48, P=0.03). We also confirmed a reported epistatic interaction between IL10 rs1800896 and aromatase (CYP19A1) rs1062033, again only in women: adjusted synergy factor=1.94 (1.16-3.25, 0.01). Aromatase, a rate-limiting enzyme in the synthesis of estrogens, is expressed in AD-relevant brain regions ,and is downregulated during the disease. IL-10 is an anti-inflammatory cytokine. Given that estrogens have neuroprotective and anti-inflammatory activities and regulate microglial cytokine production, epistasis is biologically plausible. Diminishing serum estrogen in postmenopausal women, coupled with suboptimal brain estrogen synthesis, may contribute to the inflammatory state, that is a pathological hallmark of AD.

Associations between traffic-related black carbon exposure and attention in a prospective birth cohort of urban children.

BACKGROUND: Ambient air pollution may have neurotoxic effects in children. Data examining associations between traffic-related air pollution and attention domains remain sparse. OBJECTIVES: We examined associations between black carbon (BC), a marker of traffic particles, and attention measures ascertained at 7-14 years of age among 174 children in a birth cohort based in the Boston, Massachusetts, area. METHODS: We estimated BC levels using a validated spatial-temporal land-use regression model based on residence during children's lifetime. Children completed the Conner's Continuous Performance Test (CPT) measuring omission errors, commission errors, and hit reaction time (HRT), with higher scores indicating increased errors or slower reaction time. Multivariable-adjusted linear regression analyses were used to examine associations between BC and each attention outcome. RESULTS: Children were primarily Hispanic (56%) and Caucasian (41%); 53% were boys. We found a positive association between higher BC levels with increased commission errors and slower HRT, adjusting for child IQ, age, sex, blood lead level, maternal education, pre- and postnatal tobacco smoke exposure, and community-level social stress. Notably, the association was weaker, though still positive, for the highest BC quartile relative to the middle two quartiles. Sex-stratified analysis demonstrated statistically significant associations between BC and both commission errors and HRT in boys, but BC was not significantly associated with any of the CPT outcomes in girls. CONCLUSIONS: In this population of urban children, we found associations between BC exposure and higher commission errors and slower reaction time. These associations were overall more apparent in boys than girls.

Discovery by the Epistasis Project of an epistatic interaction between the GSTM3 gene and the HHEX/IDE/KIF11 locus in the risk of Alzheimer's disease.

Despite recent discoveries in the genetics of sporadic Alzheimer's disease, there remains substantial "hidden heritability." It is thought that some of this missing heritability may be because of gene-gene, i.e., epistatic, interactions. We examined potential epistasis between 110 candidate polymorphisms in 1757 cases of Alzheimer's disease and 6294 control subjects of the Epistasis Project, divided between a discovery and a replication dataset. We found an epistatic interaction, between rs7483 in GSTM3 and rs1111875 in the HHEX/IDE/KIF11 gene cluster, with a closely similar, significant result in both datasets. The synergy factor (SF) in the combined dataset was 1.79, 95% confidence interval [CI], 1.35-2.36; p = 0.00004. Consistent interaction was also found in 7 out of the 8 additional subsets that we examined post hoc: i.e., it was shown in both North Europe and North Spain, in both men and women, in both those with and without the ε4 allele of apolipoprotein E, and in people older than 75 years (SF, 2.27; 95% CI, 1.60-3.20; p < 0.00001), but not in those younger than 75 years (SF, 1.06; 95% CI, 0.59-1.91; p = 0.84). The association with Alzheimer's disease was purely epistatic with neither polymorphism showing an independent effect: odds ratio, 1.0; p ≥ 0.7. Indeed, each factor was associated with protection in the absence of the other factor, but with risk in its presence. In conclusion, this epistatic interaction showed a high degree of consistency when stratifying by sex, the ε4 allele of apolipoprotein E genotype, and geographic region.

A modified fetal heart rate tracing interpretation system for prediction of cesarean section.

OBJECTIVE: To investigate whether a modified version of the 2008 National Institute of Child Health and Human Development (NICHD) interpretation system upon admission decreases cesarean delivery risk. METHODS: This retrospective cohort study ascribed a modified category to the first 30 min of fetal heart rate (FHR) tracings in labor. Category I was divided into two subsets (Ia and Ib) by the presence of accelerations. Category II was divided into four subsets (IIa-IId) based on baseline FHR, variability, response to stimulation and decelerations. Log-binomial regression was used to calculate risk ratios (RR) and 95% confidence intervals (CI). RESULTS: A category was ascribed to 910 women. Most FHR tracings were Category Ia (65.8%), Ib (7.7%), IIb (11.8%) and IId (14.0%). Category Ib tracings (fewer than two accelerations) were 2.26 (95% CI: 1.13-4.52) times more likely to result in cesarean delivery for abnormal FHR tracing than Category Ia tracings. A similar increase in risk was seen when comparing Category IIb and Category IId with Category Ia. CONCLUSION: Application of a modified version of the 2008 NICHD FHR interpretation system to the initial 30 min of labor can identify women at increased risk of cesarean delivery for abnormal FHR tracing.

Management of abnormal serum markers in the absence of aneuploidy or neural tube defects.

OBJECTIVE: Few guidelines address the management of pregnancies complicated by abnormal maternal serum analytes (MSAs) in the absence of aneuploidy or neural tube defects (NTDs). Our objective was to gather preliminary data regarding current opinions and management strategies among perinatologists in the US. METHODS: This survey of Maternal Fetal Medicine (MFM) physicians and fellows used a secure electronic web-based data capture tool. RESULTS: A total of 545 potential participants were contacted, and 136 (25%) responded. The majority were experienced academic physicians with robust practices. Nearly all (97.7%) respondents reported a belief in an association between abnormal MSAs and adverse pregnancy outcomes other than aneuploidy or NTDs. Plasma protein A (PAPP-A) and α-fetoprotein (AFP) were most often chosen as markers demonstrating a strong association with adverse outcomes. Most (86.9%) respondents acknowledged that abnormal MSAs influenced their counseling approach, and the majority (80.1%) offered additional ultrasound examinations. Nearly half started at 28 weeks and almost one-third at 32 weeks. Respondents acknowledging a relevant protocol in their hospital or practice were more likely to offer additional antenatal testing (p = 0.01). CONCLUSIONS: Although most perinatologists were in agreement regarding the association of MSAs with adverse pregnancy outcomes, a lack of consensus exists regarding management strategies.

Interactions between PPAR-α and inflammation-related cytokine genes on the development of Alzheimer's disease, observed by the Epistasis Project.

OBJECTIVE: Neuroinflammation contributes to the pathogenesis of sporadic Alzheimer's disease (AD). Variations in genes relevant to inflammation may be candidate genes for AD risk. Whole-genome association studies have identified relevant new and known genes. Their combined effects do not explain 100% of the risk, genetic interactions may contribute. We investigated whether genes involved in inflammation, i.e. PPAR-α, interleukins (IL) IL- 1α, IL-1ß, IL-6, and IL-10 may interact to increase AD risk. METHODS: The Epistasis Project identifies interactions that affect the risk of AD. Genotyping of single nucleotide polymorphisms (SNPs) in PPARA, IL1A, IL1B, IL6 and IL10 was performed. Possible associations were analyzed by fitting logistic regression models with AD as outcome, controlling for centre, age, sex and presence of apolipoprotein ε4 allele (APOEε4). Adjusted synergy factors were derived from interaction terms (p<0.05 two-sided). RESULTS: We observed four significant interactions between different SNPs in PPARA and in interleukins IL1A, IL1B, IL10 that may affect AD risk. There were no significant interactions between PPARA and IL6. CONCLUSIONS: In addition to an association of the PPARA L162V polymorphism with the AD risk, we observed four significant interactions between SNPs in PPARA and SNPs in IL1A, IL1B and IL10 affecting AD risk. We prove that gene-gene interactions explain part of the heritability of AD and are to be considered when assessing the genetic risk. Necessary replications will require between 1450 and 2950 of both cases and controls, depending on the prevalence of the SNP, to have 80% power to detect the observed synergy factors.

Atypical fibroxanthoma and malignant fibrous histiocytoma.

BACKGROUND: Atypical fibroxanthoma (AFX) and malignant fibrous histiocytoma (MFH) are soft-tissue tumours with variable aggressiveness. There is considerable debate about the relationship between these lesions, as histological and immunochemical differentiation is difficult. METHODS: Current opinions and evidence for diagnostic differences between AFX and MFH were reviewed. Consecutive cases of AFX and MFH were identified from our non-melanoma skin cancer (NMSC) database 1996-2007 for the Central Region of New Zealand. RESULTS: Of the 50,411 NMSC lesions excised surgically from 26,138 patients, there were 101 AFX and 15 MFH cases. Three MFH cases were originally diagnosed as AFX. AFX and MFH share similar patient demographics, size and location and histological and immunohistochemical features. Most diagnostic biopsies of AFX were not followed by formal excision. Incomplete excision occurred in a large proportion of patients with AFX, which often did not proceed to re-excision, resulting in local recurrence. Cases of MFH generally underwent definitive treatment including re-excision if incompletely excised, and postoperative adjuvant radiotherapy. CONCLUSIONS: The failure to treat AFX adequately may have resulted from the lack of appreciation of its aggressiveness. Contrary to the literature, we found few clinical differences between AFX and MFH. AFX and MFH also share similar histologic features and there are no immunohistochemical markers that reliably distinguish them. AFX is best considered a distinct entity with MFH, now reclassified as an undifferentiated pleomorphic sarcoma.

Vaginal cuff dehiscence after different modes of hysterectomy.

OBJECTIVE: To update the incidence of vaginal cuff dehiscence after different modes of hysterectomy and to describe surgical and patient characteristics of dehiscence complications. METHODS: This was an observational cohort study at a large academic hospital. All women who underwent hysterectomy and dehiscence repair between January 2006 and December 2009 were identified. Data from this study period were analyzed separately and in combination with our preliminary study (January 2000 to December 2005) for a 10-year analysis (January 2000 to December 2009). The primary outcome was incidence of vaginal cuff dehiscence after total laparoscopic hysterectomy compared with abdominal, vaginal, and laparoscopically assisted vaginal hysterectomy (LAVH). RESULTS: Between 2006 and 2009, the overall incidence of dehiscence was 0.39% (95% confidence interval [CI] 0.21-0.56). The incidence after total laparoscopic hysterectomy was 0.75% (95% CI 0.09-1.4), which was the highest among all modes of hysterectomy (LAVH was 0.46% [95% CI 0.0-1.10]; total abdominal hysterectomy was 0.38% [95% CI 0.16-0.61]; and total vaginal hysterectomy was 0.11%, [95% CI 0.0-0.32]). This incidence was appreciably lower than previously reported (4.93% in 2007 publication, 2.76% readjusted calculation). The 10-year cumulative incidence of dehiscence after all modes of hysterectomy was 0.24% (95% CI 0.15-0.33) and 1.35% (95% CI 0.72-2.3) among total laparoscopic hysterectomies. During the 10-year study period, total laparoscopic hysterectomy-related dehiscence was significantly increased compared with other modes of hysterectomy, with a risk ratio of dehiscence after total laparoscopic hysterectomy of 9.1 (95% CI 4.1-20.3) compared with total abdominal hysterectomy, risk ratio of 17.2 (95% CI 3.9-75.9) compared with total vaginal hysterectomy, and risk ratio of 4.9 (95% CI 1.1-21.5) compared with LAVH. CONCLUSION: Our updated 1.35% incidence of dehiscence after total laparoscopic hysterectomy is much lower than previously reported. LEVEL OF EVIDENCE: II.