Data encoding for healthcare data democratization and information leakage prevention.

The lack of data democratization and information leakage from trained models hinder the development and acceptance of robust deep learning-based healthcare solutions. This paper argues that irreversible data encoding can provide an effective solution to achieve data democratization without violating the privacy constraints imposed on healthcare data and clinical models. An ideal encoding framework transforms the data into a new space where it is imperceptible to a manual or computational inspection. However, encoded data should preserve the semantics of the original data such that deep learning models can be trained effectively. This paper hypothesizes the characteristics of the desired encoding framework and then exploits random projections and random quantum encoding to realize this framework for dense and longitudinal or time-series data. Experimental evaluation highlights that models trained on encoded time-series data effectively uphold the information bottleneck principle and hence, exhibit lesser information leakage from trained models.

Self-Aware SGD: Reliable Incremental Adaptation Framework For Clinical AI Models.

Healthcare is dynamic as demographics, diseases, and therapeutics constantly evolve. This dynamic nature induces inevitable distribution shifts in populations targeted by clinical AI models, often rendering them ineffective. Incremental learning provides an effective method of adapting deployed clinical models to accommodate these contemporary distribution shifts. However, since incremental learning involves modifying a deployed or in-use model, it can be considered unreliable as any adverse modification due to maliciously compromised or incorrectly labelled data can make the model unsuitable for the targeted application. This paper introduces self-aware stochastic gradient descent (SGD), an incremental deep learning algorithm that utilises a contextual bandit-like sanity check to only allow reliable modifications to a model. The contextual bandit analyses incremental gradient updates to isolate and filter unreliable gradients. This behaviour allows self-aware SGD to balance incremental training and integrity of a deployed model. Experimental evaluations on the Oxford University Hospital datasets highlight that self-aware SGD can provide reliable incremental updates for overcoming distribution shifts in challenging conditions induced by label noise.

Development and validation of early warning score systems for COVID-19 patients.

COVID-19 is a major, urgent, and ongoing threat to global health. Globally more than 24 million have been infected and the disease has claimed more than a million lives as of November 2020. Predicting which patients will need respiratory support is important to guiding individual patient treatment and also to ensuring sufficient resources are available. The ability of six common Early Warning Scores (EWS) to identify respiratory deterioration defined as the need for advanced respiratory support (high-flow nasal oxygen, continuous positive airways pressure, non-invasive ventilation, intubation) within a prediction window of 24 h is evaluated. It is shown that these scores perform sub-optimally at this specific task. Therefore, an alternative EWS based on the Gradient Boosting Trees (GBT) algorithm is developed that is able to predict deterioration within the next 24 h with high AUROC 94% and an accuracy, sensitivity, and specificity of 70%, 96%, 70%, respectively. The GBT model outperformed the best EWS (LDTEWS:NEWS), increasing the AUROC by 14%. Our GBT model makes the prediction based on the current and baseline measures of routinely available vital signs and blood tests.

Examining the link between facial affect recognition and violent offending: A comparison between web-recruited volunteers with histories of violent and non-violent offending.

BACKGROUND: The link between facial affect recognition and criminal justice involvement has been extensively researched, yet there are virtually no data on the capacity for facial affect recognition in post-incar+cerated individuals, and the results of many studies are limited due to a narrow focus on psychopathy rather than offence category. AIMS: To test the first hypothesis that individuals reporting a history of a violent offence would show a deficit in facial affect recognition and the second hypothesis that the violent offender's deficit would be exclusive to recognition of negative expressions, not affecting positive or neutral expressions. METHOD: Post-incarcerated individuals (N = 298) were recruited online through Qualtrics and completed questionnaires assessing their criminal justice background and demographics. They completed measures of facial affect recognition, anxiety and depression, and components of aggression. RESULTS: A logistic regression, including sex, ethnicity, age and years of education and depression/anxiety scores, indicated that committing a violent offence was independently associated with lower facial affect recognition scores as well as male gender and a trait-based propensity towards physical aggression, but no other co-variable. These data provided no evidence that this deficit was specific to negative emotions. CONCLUSIONS AND IMPLICATIONS FOR FUTURE RESEARCH/PRACTICE: Our study is one of the first to examine facial affect recognition in a post-incarcerated sample. It suggests that deficits in facial affect recognition, already well documented among violent prisoners, persist. While acknowledging that these may be relatively fixed characteristics, this study also suggests that, for these people, nothing happening during their imprisonment was touching this. Improving capacity in facial affect recognition should be considered as a target of intervention for violent offenders, developing or revising in-prison programmes as required.

Dynamic linkage of COVID-19 test results between Public Health England's Second Generation Surveillance System and UK Biobank.

UK Biobank (UKB) is an international health resource enabling research into the genetic and lifestyle determinants of common diseases of middle and older age. It comprises 500 000 participants. Public Health England's Second Generation Surveillance System is a centralized microbiology database covering English clinical diagnostics laboratories that provides national surveillance of legally notifiable infections, bacterial isolations and antimicrobial resistance. We previously developed secure, pseudonymized, individual-level linkage of these systems. In this study, we implemented rapid dynamic linkage, which allows us to provide a regular feed of new COVID-19 (SARS-CoV-2) test results to UKB to facilitate rapid and urgent research into the epidemiological and human genetic risk factors for severe infection in the cohort. Here, we have characterized the first 1352 cases of COVID-19 in UKB participants, of whom 895 met our working definition of severe COVID-19 as inpatients hospitalized on or after 16 March 2020. We found that the incidence of severe COVID-19 among UKB cases was 27.4 % lower than the general population in England, although this difference varied significantly by age and sex. The total number of UKB cases could be estimated as 0.6 % of the publicly announced number of cases in England. We considered how increasing case numbers will affect the power of genome-wide association studies. This new dynamic linkage system has further potential to facilitate the investigation of other infections and the prospective collection of microbiological cultures to create a microbiological biobank (bugbank) for studying the interaction of environment, human and microbial genetics on infection in the UKB cohort.

Characteristics Associated with a History of Physical and Sexual Abuse in a Community Corrections Sample.

Background: Most predictors of future criminal justice involvement are gender neutral. However, recent research has stressed the importance of physical and sexual abuse as a precursor of incarceration for women. Purpose: The aim of the present study was to assess the influence of a history of physical and sexual abuse on mental health, substance use, and criminal justice history for men and women under community corrections supervision. Methods: A sample of 613 (203 women and 410 men) participants completed structured clinical interviews and questionnaires assessing demographics, mental health and abuse history (physical vs. sexual), substance use, and criminal justice involvement. Results: Results of multivariate analyses indicated that for men, physical abuse was linked to White race, a higher number of arrests, history of a person offense, family problems, and suicidality; while sexual abuse was linked to White race, family problems, suicidality, and antisocial personality disorder. For women, physical abuse was only associated with meeting criteria for an anxiety or depressive disorder; while sexual abuse was linked to reporting a history of a substance offense, meeting criteria for an anxiety or depressive disorder, and increased suicidality. Substance use was not associated with any form of abuse in either gender. Conclusions: In general, abuse was associated with worse mental health and more severe criminal justice involvement. Women reported much greater rates of abuse and our results provide some support for the idea that a history of abuse may be an important precursor to criminal justice involvement for individuals under community corrections supervision.

Pleomorphic Invasive Lobular Carcinoma of the Male Breast: Two Case Reports with Opposite Hormone Receptor Status.

Male breast cancer represents <1% of breast cancers, with invasive lobular carcinoma in male breast being extremely rare. Only four cases of male breast with pleomorphic lobular carcinoma (PLC) are reported in literature. Here, we report two additional cases. The first case was a 63-year old male presenting with a left neck mass, progressive facial numbness, bilateral cervical lymphadenopathy, and brain metastasis. Neck mass and left breast biopsies confirmed left breast PLC with metastasis. The second case was a 79-year old male with a left breast mass; biopsy and subsequent mastectomy showed PLC. Awareness of this entity is important for rendering an accurate diagnosis, especially in the setting of metastases.

More than "just a joke": the prejudice-releasing function of sexist humor.

The results of two experiments supported the hypothesis that, for sexist men, exposure to sexist humor can promote the behavioral release of prejudice against women. Experiment 1 demonstrated that hostile sexism predicted the amount of money participants were willing to donate to a women's organization after reading sexist jokes but not after reading nonhumorous sexist statements or neutral jokes. Experiment 2 showed that hostile sexism predicted the amount of money participants cut from the budget of a women's organization relative to four other student organizations upon exposure to sexist comedy skits but not neutral comedy skits. A perceived local norm of approval of funding cuts for the women's organization mediated the relationship between hostile sexism and discrimination against the women's organization.