QTNet: Predicting Drug-Induced QT Prolongation With Artificial Intelligence-Enabled Electrocardiograms.

BACKGROUND: Prediction of drug-induced long QT syndrome (diLQTS) is of critical importance given its association with torsades de pointes. There is no reliable method for the outpatient prediction of diLQTS. OBJECTIVES: This study sought to evaluate the use of a convolutional neural network (CNN) applied to electrocardiograms (ECGs) to predict diLQTS in an outpatient population. METHODS: We identified all adult outpatients newly prescribed a QT-prolonging medication between January 1, 2003, and March 31, 2022, who had a 12-lead sinus ECG in the preceding 6 months. Using risk factor data and the ECG signal as inputs, the CNN QTNet was implemented in TensorFlow to predict diLQTS. RESULTS: Models were evaluated in a held-out test dataset of 44,386 patients (57% female) with a median age of 62 years. Compared with 3 other models relying on risk factors or ECG signal or baseline QTc alone, QTNet achieved the best (P < 0.001) performance with a mean area under the curve of 0.802 (95% CI: 0.786-0.818). In a survival analysis, QTNet also had the highest inverse probability of censorship-weighted area under the receiver-operating characteristic curve at day 2 (0.875; 95% CI: 0.848-0.904) and up to 6 months. In a subgroup analysis, QTNet performed best among males and patients ≤50 years or with baseline QTc <450 ms. In an external validation cohort of solely suburban outpatient practices, QTNet similarly maintained the highest predictive performance. CONCLUSIONS: An ECG-based CNN can accurately predict diLQTS in the outpatient setting while maintaining its predictive performance over time. In the outpatient setting, our model could identify higher-risk individuals who would benefit from closer monitoring.

Development and external validation of a dynamic risk score for early prediction of cardiogenic shock in cardiac intensive care units using machine learning.

AIMS: Myocardial infarction and heart failure are major cardiovascular diseases that affect millions of people in the USA with morbidity and mortality being highest among patients who develop cardiogenic shock. Early recognition of cardiogenic shock allows prompt implementation of treatment measures. Our objective is to develop a new dynamic risk score, called CShock, to improve early detection of cardiogenic shock in the cardiac intensive care unit (ICU). METHODS AND RESULTS: We developed and externally validated a deep learning-based risk stratification tool, called CShock, for patients admitted into the cardiac ICU with acute decompensated heart failure and/or myocardial infarction to predict the onset of cardiogenic shock. We prepared a cardiac ICU dataset using the Medical Information Mart for Intensive Care-III database by annotating with physician-adjudicated outcomes. This dataset which consisted of 1500 patients with 204 having cardiogenic/mixed shock was then used to train CShock. The features used to train the model for CShock included patient demographics, cardiac ICU admission diagnoses, routinely measured laboratory values and vital signs, and relevant features manually extracted from echocardiogram and left heart catheterization reports. We externally validated the risk model on the New York University (NYU) Langone Health cardiac ICU database which was also annotated with physician-adjudicated outcomes. The external validation cohort consisted of 131 patients with 25 patients experiencing cardiogenic/mixed shock. CShock achieved an area under the receiver operator characteristic curve (AUROC) of 0.821 (95% CI 0.792-0.850). CShock was externally validated in the more contemporary NYU cohort and achieved an AUROC of 0.800 (95% CI 0.717-0.884), demonstrating its generalizability in other cardiac ICUs. Having an elevated heart rate is most predictive of cardiogenic shock development based on Shapley values. The other top 10 predictors are having an admission diagnosis of myocardial infarction with ST-segment elevation, having an admission diagnosis of acute decompensated heart failure, Braden Scale, Glasgow Coma Scale, blood urea nitrogen, systolic blood pressure, serum chloride, serum sodium, and arterial blood pH. CONCLUSION: The novel CShock score has the potential to provide automated detection and early warning for cardiogenic shock and improve the outcomes for millions of patients who suffer from myocardial infarction and heart failure.

Ulotaront, a Trace Amine-Associated Receptor 1/Serotonin 5-HT1A Agonist, in Patients With Parkinson Disease Psychosis: A Pilot Study.

Background and Objectives: Ulotaront (SEP-363856) is a trace amine-associated receptor 1 agonist with 5-HT1A receptor agonist activity currently in phase 3 clinical development for the treatment of schizophrenia. In this exploratory, flexibly dosed study, ulotaront was evaluated for the treatment of Parkinson disease psychosis (PDP). Methods: Patients with PDP requiring antipsychotic therapy were randomized, double-blind to ulotaront (25, 50, or 75 mg/d) or placebo. Mixed Model for Repeated Measures was used to assess change from baseline in the Scale for the Assessment of Positive Symptoms for Parkinson Disease (SAPS-PD) at 6 weeks (primary end point). Results: The efficacy analysis sample comprised 38 patients (ulotaront, n = 24; placebo, n = 14). SAPS-PD total scores were numerically reduced in ulotaront-treated vs placebo-treated patients from week 1 to week 6: Least squares mean (95% confidence interval) difference in change from baseline at week 6 was -1.1 (-6.5, 4.3, p = 0.681). PDP symptom complete remission (≥100% improvement [reduction] from baseline in SAPS-PD total score) was observed in 25% of ulotaront-treated vs 0% of placebo-treated patients. SAPS-PD and Neuropsychiatric Inventory hallucinations subscales were numerically reduced vs placebo, and SAPS-PD total scores were reduced in patients with greater cognitive impairment (baseline Mini-Mental State Examination [MMSE] scores ≤24). Ulotaront improved Scales for Outcomes in Parkinson Disease Sleep Scale - Daytime Sleepiness scores (p = 0.022). There was no worsening of Unified Parkinson Disease Rating Scale Part III motor score, MMSE, or vital signs. Adverse events (≥10%) with ulotaront vs placebo included hallucinations (24% vs 14%), confusional state (20% vs 14%), dizziness (16% vs 7%), nausea (12% vs 7%), and falls (12% vs 21%). Discussion: In this exploratory pilot study, ulotaront may decrease PDP symptoms without worsening motor function, particularly in patients with cognitive impairment. Trial Registration Information: ClinicalTrials.gov identifier: NCT02969369; submitted: November 17, 2016; study start date: December 31, 2016. Classification of Evidence: This Class II study was an exploratory pilot study that was underpowered to detect a statistically significant difference between ulotaront and placebo in the treatment of patients with Parkinson disease psychosis without worsening motor function.

Oral ENT-01 Targets Enteric Neurons to Treat Constipation in Parkinson Disease : A Randomized Controlled Trial.

BACKGROUND: Parkinson disease (PD) is associated with α-synuclein (αS) aggregation within enteric neurons. ENT-01 inhibits the formation of αS aggregates and improved constipation in an open-label study in patients with PD. OBJECTIVE: To evaluate the safety and efficacy of oral ENT-01 for constipation and neurologic symptoms in patients with PD and constipation. DESIGN: Randomized, placebo-controlled phase 2b study. (ClinicalTrials.gov: NCT03781791). SETTING: Outpatient. PATIENTS: 150 patients with PD and constipation. INTERVENTION: ENT-01 or placebo daily for up to 25 days. After baseline assessment of constipation severity, daily dosing was escalated to the prokinetic dose, the maximum dose (250 mg), or the tolerability limit, followed by a washout period. MEASUREMENTS: The primary efficacy end point was the number of complete spontaneous bowel movements (CSBMs) per week. Neurologic end points included dementia (assessed using the Mini-Mental State Examination [MMSE]) and psychosis (assessed using the Scale for the Assessment of Positive Symptoms adapted for PD [SAPS-PD]). RESULTS: The weekly CSBM rate increased from 0.7 to 3.2 in the ENT-01 group versus 0.7 to 1.2 in the placebo group (P < 0.001). Improvement in secondary end points included SBMs (P = 0.002), stool consistency (P < 0.001), ease of passage (P = 0.006), and laxative use (P = 0.041). In patients with dementia, MMSE scores improved by 3.4 points 6 weeks after treatment in the ENT-01 group (n = 14) versus 2.0 points in the placebo group (n = 14). Among patients with psychosis, SAPS-PD scores improved from 6.5 to 1.7 six weeks after treatment in the ENT-01 group (n = 5) and from 6.3 to 4.4 in the placebo group (n = 6). ENT-01 was well tolerated, with no deaths or drug-related serious adverse events. Adverse events were predominantly gastrointestinal, including nausea (34.4% [ENT-01] vs. 5.3% [placebo]; P < 0.001) and diarrhea (19.4% [ENT-01] vs. 5.3% [placebo]; P = 0.016). LIMITATION: Longer treatment periods need to be investigated in future studies. CONCLUSION: ENT-01 was safe and significantly improved constipation. PRIMARY FUNDING SOURCE: Enterin, Inc.

Learning Invariant Representations with Missing Data.

Spurious correlations allow flexible models to predict well during training but poorly on related test populations. Recent work has shown that models that satisfy particular independencies involving correlation-inducing nuisance variables have guarantees on their test performance. Enforcing such independencies requires nuisances to be observed during training. However, nuisances, such as demographics or image background labels, are often missing. Enforcing independence on just the observed data does not imply independence on the entire population. Here we derive MMD estimators used for invariance objectives under missing nuisances. On simulations and clinical data, optimizing through these estimates achieves test performance similar to using estimators that make use of the full data.

Survival Mixture Density Networks.

Survival analysis, the art of time-to-event modeling, plays an important role in clinical treatment decisions. Recently, continuous time models built from neural ODEs have been proposed for survival analysis. However, the training of neural ODEs is slow due to the high computational complexity of neural ODE solvers. Here, we propose an efficient alternative for flexible continuous time models, called Survival Mixture Density Networks (Survival MDNs). Survival MDN applies an invertible positive function to the output of Mixture Density Networks (MDNs). While MDNs produce flexible real-valued distributions, the invertible positive function maps the model into the time-domain while preserving a tractable density. Using four datasets, we show that Survival MDN performs better than, or similarly to continuous and discrete time baselines on concordance, integrated Brier score and integrated binomial log-likelihood. Meanwhile, Survival MDNs are also faster than ODE-based models and circumvent binning issues in discrete models.

Care of patients with mental health issues in the emergency department: a quality improvement project.

Mental health issues have increased significantly over the past decade in the US and emergency departments (EDs) often serve as a safety net to protect and support patients with mental health issues. However, many EDs in the US are not equipped to provide the level of advanced care that patients experiencing a mental health crisis require. This article describes a quality improvement project conducted in an ED in Colorado, US, to identify and address deficiencies in the care provided to patients with mental health issues. The care of patients with mental health issues in the ED can be enhanced through measures such as implementing a validated suicide risk assessment tool, reducing ligature risks, improving the physical environment and training staff in de-escalation techniques.

Clinical, biochemical, and hematological characteristics of community-dwelling adolescent and young adult males with anorexia nervosa.

OBJECTIVE: Data regarding clinical characteristics in males with AN are limited. We aimed to delineate clinical, biochemical, and hematological features in community-dwelling adolescent and young adult males with AN. METHOD: A retrospective chart review of electronic medical records from 2000 to 2016 was conducted for 53 males aged 10-23 years old; AN (n = 36) and healthy controls (n = 17) were similar for Tanner stage. RESULTS: Adolescent and young adult males with AN were diagnosed at a mean age of 15.9 ± 3.0 years. The most prevalent strategy for weight loss (following calorie restriction) was over-exercising. Labs demonstrated polycythemia, leukopenia, and thrombocytopenia, but no electrolyte abnormalities. Compared with healthy controls of similar Tanner stage, males with AN had lower total testosterone levels. A significant proportion of males with AN had traumatic bone fractures. DISCUSSION: Over-exercising is a common secondary weight loss strategy in males with AN. Testosterone levels are lower than in controls, but electrolyte abnormalities are rare. With enhanced provider awareness, diagnostic delays may be prevented.

On the prevention and treatment of Alzheimer's disease: Control the promoters and look beyond the brain.

Alzheimer's disease (AD) is a progressive incurable neurodegenerative disease of the brain afflicting a third of the population aged 85 and older. Pathologic hallmarks include extracellular plaques of amyloid-beta (Aß), intraneuronal neurofibrillary tangles of hyperphosphorylated tau protein, synaptic destruction, neuronal death, and brain atrophy. Neuroinflammation, mediated by microglia, is a central component of the disease, and is intricately connected with peripheral inflammation. The clinical manifestations include progressive memory loss and eventual death. The present treatment of AD is largely ineffective. Nearly all AD is late-onset and presents age 65 or older, and the most common genetic risk factor is carriage of an apolipoprotein (APO) E4 allele, seen in about 25% of the general population. Individuals carrying an APOE4 allele produce more Aß and clear it less efficiently from the brain throughout life. There has been accumulating pathologic and clinical evidence that microbes, particularly the herpes simplex virus (HSV), is a causative factor for AD, most notable in carriers of the APOE4 allele. Eighty percent of the adult population harbors HSV and it resides in the trigeminal ganglion in latent state throughout life, but periodically reactivates, traveling antegrade resulting in herpes labialis and traveling retrograde into the brain leading to neuroinflammation. Functioning as an antimicrobial peptide, Aß inactivates HSV and the recurring process culminates in a buildup of Aß plaque and other hallmarks of AD over time. Periodontal disease exists in 20-50% of the adult population and is also a causative factor for AD. Accordingly, bacteria causing periodontal disease and their byproducts can enter the brain directly via the trigeminal nerve or indirectly through the bloodstream, resulting in AD pathology over time. There are many other promoters of AD, particularly inflammatory conditions outside of the brain, that can be mitigated. Small trials are finally in progress testing antimicrobial drugs for the prevention and treatment of AD. In the meantime, a more proactive approach to the prevention and treatment of AD is posited, with an emphasis on prevention, since the pathologic underpinnings of the disease start decades before the clinical manifestations. Individuals can be stratified in risk categories using family history, periodontal disease presence, APOE4 carriage, and HSV IgG positivity. Moderate- and high-risk individuals can be treated safely with various preventive measures and appropriate antimicrobial agents as discussed. Importantly, the proposed treatments are concordant with the accepted practice of medicine, and if utilized, could significantly decrease AD prevalence.

Barriers and facilitators of implementing an antimicrobial stewardship intervention for urinary tract infection in a long-term care facility.

BACKGROUND: Fifty percent of antibiotic courses in long-term care facilities (LTCFs) are unnecessary, leading to increased risk of harm. Most studies to improve antibiotic prescribing in LTCFs showed modest and unsustained results. We aimed to identify facilitators, barriers and strategies in implementing a urinary tract infection (UTI)-focused antimicrobial stewardship (AS) intervention at a LTCF, with the secondary objective of exploring the pharmacist's potential roles. METHODS: The study used a qualitative descriptive design. Participants attended either a focus group or one-on-one interview. Data were analyzed inductively using a codebook modified in an iterative analytic process. Barrier and facilitator themes were mapped using the capability, opportunity, motivation and behaviour (COM-B) model. Similarly, themes were identified from the transcripts regarding the pharmacist's roles. RESULTS: Sixteen participants were interviewed. Most barriers and facilitators mapped to the opportunities domain of the COM-B model. The main barrier themes were lack of access, lack of knowledge, ineffective communication, lack of resources and external factors, while the main facilitator themes were education, effective collaboration, good communication, sufficient resources and access. For the pharmacist's role, the barrier themes were ineffective collaboration and communication. CONCLUSION: This study supports the importance of tailoring interventions to target factors underlying barriers to behaviour change. At this LTCF, an effective antimicrobial stewardship intervention should incorporate strategies to improve access, knowledge, communication and collaboration in its design, having sufficient resources and addressing external factors to optimize its success and long-term sustainability. Can Pharm J (Ott) 2021;154:xx-xx.

Effect of Transdermal Estradiol and Insulin-like Growth Factor-1 on Bone Endpoints of Young Women With Anorexia Nervosa.

CONTEXT: Anorexia nervosa (AN) is prevalent in adolescent girls and is associated with bone impairment driven by hormonal alterations in nutritional deficiency. OBJECTIVE: To assess the impact of estrogen replacement with and without recombinant human insulin-like growth factor-1 (rhIGF-1) administration on bone outcomes. DESIGN: Double-blind, randomized, placebo-controlled 12-month longitudinal study. PARTICIPANTS: Seventy-five adolescent and young adult women with AN age 14 to 22 years. Thirty-three participants completed the study. INTERVENTION: Transdermal 17-beta estradiol 0.1 mg/day with (i) 30 mcg/kg/dose of rhIGF-1 administered subcutaneously twice daily (AN-IGF-1+) or (ii) placebo (AN-IGF-1-). The dose of rhIGF-1 was adjusted to maintain levels in the upper half of the normal pubertal range. MAIN OUTCOME MEASURES: Bone turnover markers and bone density, geometry, microarchitecture, and strength estimates. RESULTS: Over 12 months, lumbar areal bone mineral density increased in AN-IGF-1- compared to AN-IGF-1+ (P = 0.004). AN-IGF-1+ demonstrated no improvement in areal BMD in the setting of variable compliance to estrogen treatment. Groups did not differ for 12-month changes in bone geometry, microarchitecture, volumetric bone mineral density (vBMD), or strength (and results did not change after controlling for weight changes over 12 months). Both groups had increases in radial cortical area and vBMD, and tibia cortical vBMD over 12 months. Levels of a bone resorption marker decreased in AN-IGF-1- (P = 0.042), while parathyroid hormone increased in AN-IGF-1+ (P = 0.019). AN-IGF-1- experienced irregular menses more frequently than did AN-IGF-1+, but incidence of all other adverse events did not differ between groups. CONCLUSIONS: We found no additive benefit of rhIGF-1 administration for 12 months over transdermal estrogen replacement alone in this cohort of young women with AN.

Inverse-Weighted Survival Games.

Deep models trained through maximum likelihood have achieved state-of-the-art results for survival analysis. Despite this training scheme, practitioners evaluate models under other criteria, such as binary classification losses at a chosen set of time horizons, e.g. Brier score (BS) and Bernoulli log likelihood (BLL). Models trained with maximum likelihood may have poor BS or BLL since maximum likelihood does not directly optimize these criteria. Directly optimizing criteria like BS requires inverse-weighting by the censoring distribution. However, estimating the censoring model under these metrics requires inverse-weighting by the failure distribution. The objective for each model requires the other, but neither are known. To resolve this dilemma, we introduce Inverse-Weighted Survival Games. In these games, objectives for each model are built from re-weighted estimates featuring the other model, where the latter is held fixed during training. When the loss is proper, we show that the games always have the true failure and censoring distributions as a stationary point. This means models in the game do not leave the correct distributions once reached. We construct one case where this stationary point is unique. We show that these games optimize BS on simulations and then apply these principles on real world cancer and critically-ill patient data.

Understanding Failures in Out-of-Distribution Detection with Deep Generative Models.

Deep generative models (dgms) seem a natural fit for detecting out-of-distribution (ood) inputs, but such models have been shown to assign higher probabilities or densities to ood images than images from the training distribution. In this work, we explain why this behavior should be attributed to model misestimation. We first prove that no method can guarantee performance beyond random chance without assumptions on which out-distributions are relevant. We then interrogate the typical set hypothesis, the claim that relevant out-distributions can lie in high likelihood regions of the data distribution, and that ood detection should be defined based on the data distribution's typical set. We highlight the consequences implied by assuming support overlap between in- and out-distributions, as well as the arbitrariness of the typical set for ood detection. Our results suggest that estimation error is a more plausible explanation than the misalignment between likelihood-based ood detection and out-distributions of interest, and we illustrate how even minimal estimation error can lead to ood detection failures, yielding implications for future work in deep generative modeling and ood detection.