Foresight-a generative pretrained transformer for modelling of patient timelines using electronic health records: a retrospective modelling study.

BACKGROUND: An electronic health record (EHR) holds detailed longitudinal information about a patient's health status and general clinical history, a large portion of which is stored as unstructured, free text. Existing approaches to model a patient's trajectory focus mostly on structured data and a subset of single-domain outcomes. This study aims to evaluate the effectiveness of Foresight, a generative transformer in temporal modelling of patient data, integrating both free text and structured formats, to predict a diverse array of future medical outcomes, such as disorders, substances (eg, to do with medicines, allergies, or poisonings), procedures, and findings (eg, relating to observations, judgements, or assessments). METHODS: Foresight is a novel transformer-based pipeline that uses named entity recognition and linking tools to convert EHR document text into structured, coded concepts, followed by providing probabilistic forecasts for future medical events, such as disorders, substances, procedures, and findings. The Foresight pipeline has four main components: (1) CogStack (data retrieval and preprocessing); (2) the Medical Concept Annotation Toolkit (structuring of the free-text information from EHRs); (3) Foresight Core (deep-learning model for biomedical concept modelling); and (4) the Foresight web application. We processed the entire free-text portion from three different hospital datasets (King's College Hospital [KCH], South London and Maudsley [SLaM], and the US Medical Information Mart for Intensive Care III [MIMIC-III]), resulting in information from 811 336 patients and covering both physical and mental health institutions. We measured the performance of models using custom metrics derived from precision and recall. FINDINGS: Foresight achieved a precision@10 (ie, of 10 forecasted candidates, at least one is correct) of 0·68 (SD 0·0027) for the KCH dataset, 0·76 (0·0032) for the SLaM dataset, and 0·88 (0·0018) for the MIMIC-III dataset, for forecasting the next new disorder in a patient timeline. Foresight also achieved a precision@10 value of 0·80 (0·0013) for the KCH dataset, 0·81 (0·0026) for the SLaM dataset, and 0·91 (0·0011) for the MIMIC-III dataset, for forecasting the next new biomedical concept. In addition, Foresight was validated on 34 synthetic patient timelines by five clinicians and achieved a relevancy of 33 (97% [95% CI 91-100]) of 34 for the top forecasted candidate disorder. As a generative model, Foresight can forecast follow-on biomedical concepts for as many steps as required. INTERPRETATION: Foresight is a general-purpose model for biomedical concept modelling that can be used for real-world risk forecasting, virtual trials, and clinical research to study the progression of disorders, to simulate interventions and counterfactuals, and for educational purposes. FUNDING: National Health Service Artificial Intelligence Laboratory, National Institute for Health and Care Research Biomedical Research Centre, and Health Data Research UK.

CYCLOIDEA paralogs function partially redundantly to specify dorsal flower development in Mimulus lewisii.

PREMISE: Duplicated genes (paralogs) are abundant in plant genomes, and their retention may influence the function of genetic programs and contribute to evolutionary novelty. How gene duplication affects genetic modules and what forces contribute to paralog retention are outstanding questions. The CYCLOIDEA(CYC)-dependent flower symmetry program is a model for understanding the evolution of gene duplication, providing multiple examples of paralog partitioning and novelty. However, a novel CYC gene lineage duplication event near the origin of higher core Lamiales (HCL) has received little attention. METHODS: To understand the evolutionary fate of duplicated HCL CYC2 genes, we determined the effects on flower symmetry by suppressing MlCYC2A and MlCYC2B expression using RNA interference (RNAi). We determined the phenotypic effects on flower symmetry in single- and double-silenced backgrounds and coupled our functional analyses with expression surveys of MlCYC2A, MlCYC2B, and a putative downstream RADIALIS (MlRAD5) ortholog. RESULTS: MlCYC2A and MlCYC2B jointly contribute to bilateral flower symmetry. MlCYC2B exhibits a clear dorsal flower identity function and may additionally function in carpel development. MlCYC2A functions in establishing dorsal petal shape. Further, our results suggest an MlCYC2A-MlCYC2B regulatory interaction, which may affect pathway homeostasis. CONCLUSIONS: Our results suggest that CYC paralogs specific to higher core Lamiales may be selectively retained for their joint contribution to bilateral flower symmetry, similar to the independently derived CYC paralogs in the Lamiales model for bilateral flower symmetry research, Antirrhinum majus (snapdragon).

Artificial intelligence and machine learning for clinical pharmacology.

Artificial intelligence (AI) will impact many aspects of clinical pharmacology, including drug discovery and development, clinical trials, personalized medicine, pharmacogenomics, pharmacovigilance and clinical toxicology. The rapid progress of AI in healthcare means clinical pharmacologists should have an understanding of AI and its implementation in clinical practice. As with any new therapy or health technology, it is imperative that AI tools are subject to robust and stringent evaluation to ensure that they enhance clinical practice in a safe and equitable manner. This review serves as an introduction to AI for the clinical pharmacologist, highlighting current applications, aspects of model development and issues surrounding evaluation and deployment. The aim of this article is to empower clinical pharmacologists to embrace and lead on the safe and effective use of AI within healthcare.

Heightened generalized conditioned fear and avoidance in women and underlying psychological processes.

Heightened generalization of conditioned fear and avoidance to safe stimuli resembling threat is a key feature of pathological anxiety and might contribute to the increased prevalence of anxiety-related disorders among women. Though animal studies have documented over-generalized fear in female versus male rodents, analogous work in humans is sparse, and no studies to date have examined gender differences in generalized avoidance. We addressed this gap by testing 170 self-identified women (n = 85) and men (n = 85) using a video game-based task assessing generalized Pavlovian fear (perceived threat, fear-potentiated startle) and generalized instrumental avoidance. Instrumental measures of generalization reflected maladaptive avoidance by virtue of being unnecessary to secure safety and incurring a cost of losing the game in which the task is embedded. Women displayed increases in both Pavlovian generalization of perceived threat and maladaptive generalized avoidance. Additionally, decreased motivation to win the game among women mediated the effect of gender on generalized avoidance, and generalized perceived risk and tendencies toward experiential avoidance positively predicted generalized avoidance in women but not men. Overall, findings implicate the undue spread of fear and avoidance to safe stimuli resembling danger among women as a candidate mechanism for differential rates of clinical anxiety across the genders.

The impact of artificial intelligence on clinical education: perceptions of postgraduate trainee doctors in London (UK) and recommendations for trainers.

BACKGROUND: Artificial intelligence (AI) technologies are increasingly used in clinical practice. Although there is robust evidence that AI innovations can improve patient care, reduce clinicians' workload and increase efficiency, their impact on medical training and education remains unclear. METHODS: A survey of trainee doctors' perceived impact of AI technologies on clinical training and education was conducted at UK NHS postgraduate centers in London between October and December 2020. Impact assessment mirrored domains in training curricula such as 'clinical judgement', 'practical skills' and 'research and quality improvement skills'. Significance between Likert-type data was analysed using Fisher's exact test. Response variations between clinical specialities were analysed using k-modes clustering. Free-text responses were analysed by thematic analysis. RESULTS: Two hundred ten doctors responded to the survey (response rate 72%). The majority (58%) perceived an overall positive impact of AI technologies on their training and education. Respondents agreed that AI would reduce clinical workload (62%) and improve research and audit training (68%). Trainees were skeptical that it would improve clinical judgement (46% agree, p = 0.12) and practical skills training (32% agree, p < 0.01). The majority reported insufficient AI training in their current curricula (92%), and supported having more formal AI training (81%). CONCLUSIONS: Trainee doctors have an overall positive perception of AI technologies' impact on clinical training. There is optimism that it will improve 'research and quality improvement' skills and facilitate 'curriculum mapping'. There is skepticism that it may reduce educational opportunities to develop 'clinical judgement' and 'practical skills'. Medical educators should be mindful that these domains are protected as AI develops. We recommend that 'Applied AI' topics are formalized in curricula and digital technologies leveraged to deliver clinical education.

Beta-blocker therapy in patients with COPD: a systematic literature review and meta-analysis with multiple treatment comparison.

BACKGROUND: Beta-blockers are associated with reduced mortality in patients with cardiovascular disease but are often under prescribed in those with concomitant COPD, due to concerns regarding respiratory side-effects. We investigated the effects of beta-blockers on outcomes in patients with COPD and explored within-class differences between different agents. METHODS: We searched the Cochrane Central Register of Controlled Trials, Embase, Cumulative Index to Nursing and Allied Health Literature (CINAHL) and Medline for observational studies and randomized controlled trials (RCTs) investigating the effects of beta-blocker exposure versus no exposure or placebo, in patients with COPD, with and without cardiovascular indications. A meta-analysis was performed to assess the association of beta-blocker therapy with acute exacerbations of COPD (AECOPD), and a network meta-analysis was conducted to investigate the effects of individual beta-blockers on FEV1. Mortality, all-cause hospitalization, and quality of life outcomes were narratively synthesized. RESULTS: We included 23 observational studies and 14 RCTs. In pooled observational data, beta-blocker therapy was associated with an overall reduced risk of AECOPD versus no therapy (HR 0.77, 95%CI 0.70 to 0.85). Among individual beta-blockers, only propranolol was associated with a relative reduction in FEV1 versus placebo, among 199 patients evaluated in RCTs. Narrative syntheses on mortality, all-cause hospitalization and quality of life outcomes indicated a high degree of heterogeneity in study design and patient characteristics but suggested no detrimental effects of beta-blocker therapy on these outcomes. CONCLUSION: The class effect of beta-blockers remains generally positive in patients with COPD. Reduced rates of AECOPD, mortality, and improved quality of life were identified in observational studies, while propranolol was the only agent associated with a deterioration of lung function in RCTs.

Device Utilization Ratios in Infection Prevention: Process or Outcome Measure?

PURPOSE OF REVIEW: The purpose of this review is to describe the role of device utilization as a component of surveillance for healthcare-associated infections and describe its potential role as a measurement of healthcare quality. RECENT FINDINGS: Device utilization, while primarily a process-based measure in the prevention of device-associated infections can also serve as an important outcome in the evaluation of an infection prevention program. Device utilization can be an important and resource-efficient measurement when coupled with measurements of risk-adjusted infection rates. The measurement of the device utilization ratio can provide insight into the risk of device-associated harms, including non-infectious harms, which would not be captured with currently used infection-based surveillance metrics. Further study and validation of standardized, risk-adjusted device utilization measurements is an important area for future exploration.

Are Routine Blood Group and Save Samples Needed for Laparoscopic Day Case Surgery?

BACKGROUND: Common day case laparoscopic procedures are usually safe, with low rates of bleeding complications. At our trust, most patients undergo pre-operative group and save (G&S) for these procedures, at a cost of £18.39 per sample excluding laboratory staffing costs. Our aim was to assess if routine G&S is indicated. METHODS: We performed a retrospective review of all patients who underwent laparoscopic cholecystectomy (LC), laparoscopic inguinal hernia repair (LIH) and diagnostic laparoscopy (DL) in our institution between April 2012 and March 2014. Patients were identified using hospital coding records. Transfusion department records were reviewed to see which patients had undergone pre-operative G&S or cross-match, and peri-operative transfusion. RESULTS: Five hundred and thirty-two procedures were performed in 2 years: 293 LC, 123 LIH and 116 DL. G&S was performed in 256 (87 %; LC), 67 (54 %; LIH) and 88 (76 %; DL), respectively. Zero patients were transfused for bleeding complications. One patient was transfused following diagnostic laparoscopy to optimise pre-existing anaemia. The total cost of G&S over the study period was £7558. CONCLUSION: Blood transfusion rates for bleeding complications following laparoscopic day case surgery are 0 % in our unit. G&S samples for these procedures cost £7558 over 2 years. Abandoning pre-operative G&S for these patients appears to be clinically indicated and would lead to substantial financial savings.

Antibiotic Utilization and Opportunities for Stewardship Among Hospitalized Patients With Influenza Respiratory Tract Infection.

OBJECTIVE: Hospitalized influenza patients are often treated with antibiotics empirically while awaiting final diagnosis. The goal of this study was to describe the inappropriate continuation of antibiotics for influenza respiratory tract infections (RTIs). DESIGN: We retrospectively studied adults admitted to our institution over 2 respiratory flu seasons with positive influenza RTIs. Inappropriate antibiotic duration (IAD) was defined as antibiotic use for >24 hours after a positive influenza test in patients presenting with <72 hours of RTI symptoms and with no other indications of bacterial infection. RESULTS: During the study period, 322 patients included in this study were admitted for influenza RTI. Respiratory cultures were ordered for 50 of these patients (15.5%) and 71 patients (22%) had a positive chest x-ray, but antibiotics were prescribed to 211 patients (65.5%) on admission. Antibiotics were inappropriately continued in 73 patients (34.5%). Patients receiving IAD had a longer length of stay (LOS) (median, 6 days; range, 4-9 days) compared with those whose antibiotics were discontinued appropriately (median, 5 days; range, 3-8 days) and those who were not treated with antibiotics (median, 4 days; range, 3-6 days; P<.001). However, mortality was similar among these 3 groups: 3 patients (4.1%) from the IAD cohort died; 6 patients (4.3%) from the group with an appropriate antibiotic duration died; and 2 patients [1.8%] from the group given no antibiotics died (P=.510). The 30-day readmission rates were similar as well: 9 patients (12.3%) from the IAD group were readmitted within 30 days; 21 patients (15.2%) from the group with appropriate antibiotic duration were readmitted; and 11 patients (9.9%) from the group given no antibiotics were readmitted (P=.455). Total hospital costs were greater in patients treated with IAD ($10,645; range, $6,485-$18,035) compared with the group treated with appropriate antibiotic duration ($7,479; range, $4,866-$12,922) and the group given no antibiotics $5,961 (range, $4,711-$9,575). Thus, the hospital experienced a median loss in net hospital revenue of $2,076 per IAD patient compared with a patient for which antibiotic duration was appropriate. CONCLUSION: The majority of patients with influenza RTI received antibiotics on admission, and 34.5% were inappropriately continued on antibiotics without evidence of bacterial infection, which led to increased LOS, loss of net revenue, and no improvement in outcome. Thus, stewardship initiatives aimed at this population are warranted.

Respiratory problems and management in people with spinal cord injury.

Spinal cord injury (SCI) is characterised by profound respiratory compromise secondary to the level of loss of motor, sensory and autonomic control associated with the injury. This review aims to detail these anatomical and physiological changes after SCI, and outline their impact on respiratory function. Injury-related impairments in strength substantially alter pulmonary mechanics, which in turn affect respiratory management and care. Options for treatments must therefore be considered in light of these limitations. KEY POINTS: Respiratory impairment following spinal cord injury (SCI) is more severe in high cervical injuries, and is characterised by low lung volumes and a weak cough secondary to respiratory muscle weakness.Autonomic dysfunction and early-onset sleep disordered breathing compound this respiratory compromise.The mainstays of management following acute high cervical SCI are tracheostomy and ventilation, with noninvasive ventilation and assisted coughing techniques being important in lower cervical and thoracic level injuries.Prompt investigation to ascertain the extent of the SCI and associated injuries, and appropriate subsequent management are important to improve outcomes. EDUCATIONAL AIMS: To describe the anatomical and physiological changes after SCI and their impact on respiratory function.To describe the changes in respiratory mechanics seen in cervical SCI and how these changes affect treatments.To discuss the relationship between injury level and respiratory compromise following SCI, and describe those at increased risk of respiratory complications.To present the current treatment options available and their supporting evidence.

Carbapenem-Resistant Enterobacteriaceae: A Case Series of Infections at Hartford Hospital.

Infections due to carbapenem-resistant Enterobacteriaceae (CRE) represent a growing problem nationally, are difficult to treat, and are associated with high mortality rates. Features of CRE infection within Connecticut, a nonendemic area, have not been described in the medical literature. We report a case series of five patients with CRE isolates who were treated at Hartford Hospital in Hartford, Connecticut from 2011-2013. CRE isolates included three Klebsiella pneumoniae, one Klebsiella oxytoca, and two E. coli. Three isolates were obtained from urine cultures, one from respiratory culture, and one from blood and wound cultures. Four patients survived the hospital course, including three patients who had a hospital stay of > 100 days. In this nonendemic hospital, CRE infections were identified mainly in patients with multiple comorbidities who underwent catheter placement and surgical procedures, had contact with the health care system in the preceding 90 days, and tended to have prolonged, complicated hospital courses.

Risk of self-harm and suicide in people with specific psychiatric and physical disorders: comparisons between disorders using English national record linkage.

Background Psychiatric illnesses are known risk factors for self-harm but associations between self-harm and physical illnesses are less well established. We aimed to stratify selected chronic physical and psychiatric illnesses according to their relative risk of self-harm. Design Retrospective cohort studies using a linked dataset of Hospital Episode Statistics (HES) for 1999-2011. Participants Individuals with selected psychiatric or physical conditions were compared with a reference cohort constructed from patients admitted for a variety of other conditions and procedures. Setting All admissions and day cases in National Health Service (NHS) hospitals in England. Main outcome measures Hospital episodes of self-harm. Rate ratios (RRs) were derived by comparing admission for self-harm between cohorts. Results The psychiatric illnesses studied (depression, bipolar disorder, alcohol abuse, anxiety disorders, eating disorders, schizophrenia and substance abuse) all had very high RRs (> 5) for self-harm. Of the physical illnesses studied, an increased risk of self-harm was associated with epilepsy (RR = 2.9, 95% confidence interval [CI] 2.8-2.9), asthma (1.8, 1.8-1.9), migraine (1.8, 1.7-1.8), psoriasis (1.6, 1.5-1.7), diabetes mellitus (1.6, 1.5-1.6), eczema (1.4, 1.3-1.5) and inflammatory polyarthropathies (1.4, 1.3-1.4). RRs were significantly low for cancers (0.95, 0.93-0.97), congenital heart disease (0.9, 0.8-0.9), ulcerative colitis (0.8, 0.7-0.8), sickle cell anaemia (0.7, 0.6-0.8) and Down's syndrome (0.1, 0.1-0.2). Conclusions Psychiatric illnesses carry a greatly increased risk of self-harm as well as of suicide. Many chronic physical illnesses are also associated with an increased risk of both self-harm and suicide. Identifying those at risk will allow provision of appropriate monitoring and support.

Auto-titrating continuous positive airway pressure treatment for obstructive sleep apnoea after acute quadriplegia (COSAQ): study protocol for a randomized controlled trial.

BACKGROUND: Quadriplegia is a severe, catastrophic injury that predominantly affects people early in life, resulting in lifelong physical disability. Obstructive sleep apnoea is a direct consequence of quadriplegia and is associated with neurocognitive deficits, sleepiness and reduced quality of life. The usual treatment for sleep apnoea is nasal continuous positive airway pressure (CPAP); however, this is poorly tolerated in quadriplegia. To encourage patients to use this therapy, we have to demonstrate that the benefits outweigh the inconvenience. We therefore propose a prospective, multinational randomized controlled trial of three months of CPAP for obstructive sleep apnoea after acute quadriplegia. METHODS/DESIGN: Specialist spinal cord injury centres across Australia, New Zealand, the UK and Canada will recruit medically stable individuals who have sustained a (new) traumatic quadriplegia (complete or incomplete second cervical to first thoracic level lesions). Participants will be screened for obstructive sleep apnoea using full, portable sleep studies. Those with an apnoea hypopnoea index greater than 10 per hour will proceed to an initial three-night trial of CPAP. Those who can tolerate CPAP for at least 4 hours on at least one night of the initial trial will be randomized to either usual care or a 3-month period of auto-titrating CPAP. The primary hypothesis is that nocturnal CPAP will improve neuropsychological functioning more than usual care alone. The secondary hypothesis is that the magnitude of improvement of neuropsychological function will be predicted by the severity of baseline sleepiness measures, sleep fragmentation and sleep apnoea. Neuropsychological tests and full polysomnography will be performed at baseline and 3 months with interim measures of sleepiness and symptoms of autonomic dysfunction measured weekly. Spirometry will be performed monthly. Neuropsychological tests will be administered by blinded assessors. Recruitment commenced in July 2009. DISCUSSION: The results of this trial will demonstrate the effect of nocturnal CPAP treatment of obstructive sleep apnoea in acute quadriplegia. If CPAP can improve neurocognitive function after injury, it is likely that rehabilitation and subsequent community participation will be substantially improved for this group of predominantly young and severely physically disabled people. TRIAL REGISTRATION: Australian New Zealand Clinical Trial Registry ACTRN12605000799651.

Cardiac tamponade as a result of Mycobacterium avium-complex pleural effusion.

Patients with advanced AIDS are prone to opportunistic infections, including disseminated disease from Mycobacterium avium-complex (MAC). We describe the first reported case of MAC related pleural effusion causing cardiac tamponade.

Patterns of adherence to antiretroviral medications: the value of electronic monitoring.

OBJECTIVE: To investigate the patterns of intra-subject (between medication) adherence to antiretroviral therapy. DESIGN: A prospective, observational, 3-month study of adherence to antiretroviral therapy at an inner-city clinic in 40 HIV-infected subjects. METHODS: Adherence was monitored monthly by the use of medication event monitoring system (Aprex) caps placed on each antiretroviral drug in a subject's regimen. Agreement between different drug classes and dosing schedules, for each subject, was quantified by estimating the mean difference in adherence, with 95% limits of agreement. An analysis of variance model was used to estimate the variance of the differences. Individual dosing calendars were examined for each subject. RESULTS: The dosing schedule was a strong predictor of intra-subject adherence. Regardless of the subject's overall adherence rate, high or low, when subjects missed a dose of one medication, they missed a dose of both medications taken at that dosing time. Conversely, when medications were scheduled to be taken together, regardless of the drug class, the medications were taken at the same times. The majority of the subjects took medications at obviously incorrect times. Problematical adherence was related to thrice-daily dosing and food restrictions. CONCLUSION: This is the first report objectively to quantify intra-subject adherence to antiretroviral therapy and report the findings in detail. We observed clear patterns of drug-taking behavior among the subjects in our study. To the extent that medication scheduling is a controllable factor, our report provides an insight into specific patterns of behavior that may be targets for adherence counseling.

Bioterrorism preparedness.

The Connecticut Department of Public Health (DPH) entered into a cooperative agreement with the Centers for Disease Control and Prevention (CDC) to establish public health preparedness and a response plan for bioterrorism. With funds from the CDC and an additional grant from the Health Resources and Services Administration (HRSA), the DPH designated Hartford Hospital as one of two Centers of Excellence that will coordinate and manage a statewide system for bioterrorism preparedness. This paper reviews the progress that Hartford Hospital has made in meeting this challenge. Highlighted are the development of a Web application to use for statewide preparedness and response, and the preparation for a smallpox vaccination program at Hartford Hospital.