A Deep Learning Framework for Predicting Patient Decannulation on Extracorporeal Membrane Oxygenation Devices: Development and Model Analysis Study.

BACKGROUND: Venovenous extracorporeal membrane oxygenation (VV-ECMO) is a therapy for patients with refractory respiratory failure. The decision to decannulate someone from extracorporeal membrane oxygenation (ECMO) often involves weaning trials and clinical intuition. To date, there are limited prognostication metrics to guide clinical decision-making to determine which patients will be successfully weaned and decannulated. OBJECTIVE: This study aims to assist clinicians with the decision to decannulate a patient from ECMO, using Continuous Evaluation of VV-ECMO Outcomes (CEVVO), a deep learning-based model for predicting success of decannulation in patients supported on VV-ECMO. The running metric may be applied daily to categorize patients into high-risk and low-risk groups. Using these data, providers may consider initiating a weaning trial based on their expertise and CEVVO. METHODS: Data were collected from 118 patients supported with VV-ECMO at the Columbia University Irving Medical Center. Using a long short-term memory-based network, CEVVO is the first model capable of integrating discrete clinical information with continuous data collected from an ECMO device. A total of 12 sets of 5-fold cross validations were conducted to assess the performance, which was measured using the area under the receiver operating characteristic curve (AUROC) and average precision (AP). To translate the predicted values into a clinically useful metric, the model results were calibrated and stratified into risk groups, ranging from 0 (high risk) to 3 (low risk). To further investigate the performance edge of CEVVO, 2 synthetic data sets were generated using Gaussian process regression. The first data set preserved the long-term dependency of the patient data set, whereas the second did not. RESULTS: CEVVO demonstrated consistently superior classification performance compared with contemporary models (P<.001 and P=.04 compared with the next highest AUROC and AP). Although the model's patient-by-patient predictive power may be too low to be integrated into a clinical setting (AUROC 95% CI 0.6822-0.7055; AP 95% CI 0.8515-0.8682), the patient risk classification system displayed greater potential. When measured at 72 hours, the high-risk group had a successful decannulation rate of 58% (7/12), whereas the low-risk group had a successful decannulation rate of 92% (11/12; P=.04). When measured at 96 hours, the high- and low-risk groups had a successful decannulation rate of 54% (6/11) and 100% (9/9), respectively (P=.01). We hypothesized that the improved performance of CEVVO was owing to its ability to efficiently capture transient temporal patterns. Indeed, CEVVO exhibited improved performance on synthetic data with inherent temporal dependencies (P<.001) compared with logistic regression and a dense neural network. CONCLUSIONS: The ability to interpret and integrate large data sets is paramount for creating accurate models capable of assisting clinicians in risk stratifying patients supported on VV-ECMO. Our framework may guide future incorporation of CEVVO into more comprehensive intensive care monitoring systems.

Use of an Opt-Out vs Opt-In Strategy Increases Use of Residency Mental Health Services.

Background Residents report high levels of distress but low utilization of mental health services. Prior research has shown several barriers that prevent residents from opting into available mental health services. Objective To determine the impact of a mental health initiative centered around an opt-out versus an opt-in approach to help-seeking, on the use of psychotherapy. Methods Resident use of psychotherapy was compared between 2 time frames. During the first time frame (July 1, 2020 to January 31, 2021), residents were offered access to therapy that they could self-initiate by calling to schedule an appointment (opt-in). The second time frame (February 1, 2021 to April 30, 2021) involved the switch to an opt-out structure, during which the same residents were scheduled for a session but could choose to cancel. Additional changes were implemented to reduce stigma and minimize barriers. The outcome was psychotherapy use by residents. Results Of the 114 residents, 7 (6%) self-initiated therapy during the opt-in period. When these same residents were placed in an opt-out context, 59 of the remaining 107 residents (55%) kept their initial appointment, and 23 (39%) self-initiated additional sessions. Altogether, across both phases, a total of 30 of the 114 residents initiated therapy (ie, 7 during the opt-in and 23 during the opt-out). The differences in therapy use between the 2 phases are statistically significant (P<.001 by McNemar's test). Conclusions There was a substantial increase in residents' use of psychotherapy after the opt-out initiative that included efforts to reduce stigma and encourage mental health services.

AAV-based delivery of RNAi targeting Ataxin-2 improves survival, strength, and pathology in mouse models of rapidly and slowly progressive sporadic ALS.

Amyotrophic lateral sclerosis (ALS) is characterized by motor neuron death due to nuclear loss and cytoplasmic aggregation of the splice factor TDP-43. Pathologic TDP-43 associates with stress granules (SGs) and downregulating the SG-associated protein Ataxin-2 (Atxn2) using antisense oligonucleotides (ASO) prolongs survival in the TAR4/4 sporadic ALS mouse model, a strategy now in clinical trials. Here, we used AAV-mediated RNAi delivery to achieve lasting and targeted Atxn2 knockdown after a single injection. To achieve this, a novel AAV with improved transduction potency of our target cells was used to deliver Atxn2 -targeting miRNAs. Mouse dosing studies demonstrated 55% Atxn2 knockdown in frontal cortex and 25% knockdown throughout brainstem and spinal cord after intracerebroventricular injection at a dose 40x lower than used in other recent studies. In TAR4/4 mice, miAtxn2 treatment increased mean and median survival by 54% and 45% respectively (p<0.0003). Mice showed robust improvement across strength-related measures ranging from 24-75%. Interestingly, treated mice showed increased vertical activity above wildtype, suggesting unmasking of an FTD phenotype with improved strength. Histologically, lower motor neuron survival improved with a concomitant reduction in CNS inflammatory markers. Additionally, phosphorylated TDP-43 was reduced to wildtype levels. Bulk RNA sequencing revealed correction of 153 genes in the markedly dysregulated transcriptome of mutant mice, several of which are described in the human ALS literature. In slow progressing hemizygous mice, treatment rescued weight loss and improved gait at late time points. Cumulatively the data support the utility of AAV-mediated RNAi against Atxn2 as a robust and translatable treatment strategy for sporadic ALS.