VALIDATION OF A MACHINE LEARNING MODEL IN A PROSPECTIVE COHORT: THE RISQ STUDY

Introduction Previously our group had identified 20 features which were associated with the development of upper gastrointestinal (UGI) cancers using a machine learning approach.[1] We sought to refine this model and to validate this in an independent dataset to assess its generalisability in an interim analysis. Methods We selected patients who were recruited for the multicentre Saliva to Predict rIsk of disease using Transcriptomics and epigenetics (SPIT) study to develop our model. Patients were recruited from 2-week wait suspected UGI pathways and additionally enriched with patients with confirmed oesophageal adenocarcinoma admitted as inpatients. We used regularised logistic regression (glmnet) from the caret package in R software to create the model. 60% of the data with 10-fold cross validation was used for training, with the remaining 40% for testing. For validation, we used data from the predicting RIsk of disease uSing detailed Questionnaires (RISQ) study, an ongoing prospective multicentre study using the questionnaire based on the our previous work.1 We evaluated the model using area under the receiver operating characteristic curve (AUC). Results We included 93 cancer and 715 non-cancer patients for training and testing and 21 cancer and 203 non-cancer patients for validation. We further reduced the model to 18 features without significant detriment to model performance. In the training and testing data AUC was 0.86 (95%CI: 0.81- 0.91) and 0.75 (95%CI: 0.67-0.83) respectively. We set a threshold of 0.03 as a cut off based on a cost function where false negatives had a 50-time greater impact than false positive cases (figure 1). For the validation cohort we achieved an AUC of 0.95 (95%CI: 0.90-1.00). This equated to a sensitivity 0.952 and a specificity of 0.897 for detecting cancer. Conclusions Initial results from our model compare favourably with the Edinburgh Dysphagia Scale, which has a sensitivity and specificity of 0.984 and 0.093 respectively.2 It also appears to have a high specificity, potentially helping to reduce unnecessary endoscopies. We aim to further increase the size of the validation cohort to ensure its robustness and generalisability. Our model could be applied to triaging and prioritising endoscopic referral backlogs as a result of COVID- 19.3.

USING MACHINE LEARNING TO DEVELOP MODELS FOR THE PREDICTION OF UPPER GASTROINTESTINAL CANCERS

Introduction Machine learning methods have been used to develop predictive models in gastroenterology.1 Previously we identified features including age, history of psychological disorders and severity of dysphagia symptoms which were correlated with upper gastrointestinal (UGI) cancers.2 We sought to create a machine learning based model which could be used to predict the presence of UGI in patients referred for endoscopy. Methods Patients were recruited as part of the Saliva to Predict rIsk of disease using Transcriptomics and epigenetics (SPIT) study. Patients were recruited from 2-week wait suspected UGI pathway referrals at 20 hospitals in the United Kingdom. We enriched the cohort with additional patients admitted with confirmed oesophageal adenocarcinoma. 60% of the data was used for model generation with 10-fold cross validation, while the models were tested on the remaining 40% of the data. We used seven methods to generate our models: Linear Discriminant Analysis (lda), Classification and Regression Tree (cart), k-Nearest Neighbour (knn), Support Vector Machines (svm), Random Forest (rf), Logistic Regression (glm) and Regularised Logistic Regression (glmnet). Model performance was assessed using area under the receiver operating characteristic curve (AUC) and DeLong test was used for model comparison. Results 93 cancer and 715 non-cancer patients were included. The best three models with 18 features were glmnet, lda and glm which all achieved an AUC of greater than 0.80 (figure 1). For the testing dataset, AUC was 0.75 (95%CI: 0.67- 0.83), 0.74 (95%CI: 0.66-0.82) and 0.75 (95%CI: 0.68-0.83) (p=ns for all 3 pairwise comparisons) respectively. When applying a cost function, the three models all achieved a sensitivity of 0.973 and a specificity of 0.234 to 0.388 for the testing dataset. Conclusions Our models compare favourably with the Edinburgh Dysphagia Scale, which has a sensitivity and specificity of 0.984 and 0.093 respectively.3 Our models have the advantage of an improved specificity, which could equate to fewer endoscopies being performed for low risk patients. Given rising waiting lists as a direct result of COVID-19, our tool could be used to prioritise patients who should be investigated sooner.4 We plan next to validate our models on a validation cohort to assess its generalisability.

Multiparametric CMR Findings in Mrna COVID-19 Vaccine-induced Myocardit is: A Case Series

Introduction: Within weeks of the pediatric coronavirus disease 2019 (COVID-19) vaccination campaign beginning, reports of acute myocarditis after adolescents' second vaccination began. The present research describes the clinical and cardiovascular magnetic resonance (CMR) imaging characteristics of three adolescents recently vaccinated with a mRNA vaccine and admitted for myopericarditis treatment. Case Description: This retrospective case-series investigated adolescents admitted within a week of their second mRNA COVID-19 vaccination. The electronic medical record was queried for all patients ≥12 years old, admitted for acute myocarditis or pericarditis (International Classification Diseases-Version 10;I30.xx, I40.xx respectively) since April 1, 2021. Patients were included if they had a documented mRNA vaccination in the prior seven days. Three patients met inclusion criteria. All three had acute onset chest pain within 48 hours of receiving their second mRNA vaccine. All had elevated troponins, all were eventually admitted and had mild clinical courses. All met Lake Louise criteria for acute myocarditis despite only one patient having mild depression of cardiac function on echocardiography. All patients were negative for COVID-19 and none had a clinical history or immunologic evidence of prior COVID-19. The patient with the most diffuse pattern of late gadolinium enhancement on CMR (Figure 1) developed ventricular tachycardia three weeks after discharge. Discussion: Vaccine induced myopericarditis is rare in inactivated vaccines, but is a known entity with live vaccines, especially the smallpox vaccine. Since the 1950's, cases of myocarditis and pericarditis have been reported in association with vaccination. Research using VAERS has previously found that from 1990- 2018, 0.1% of reports were for myopericarditis associated with vaccination. The rates of mRNA vaccine-induced myocarditis are currently unknown, but our clinical findings are similar to other recently published case series of pediatric mRNA associated myopericarditis. We have observed differences in CMR patterns between our patients from this series and previous reports of patients with cardiac involvement from COVID-19 (Table 1). We remain uncertain regarding the precise pathophysiology in these patients with myocardial inflammation following mRNA vaccine administration. However, the relatively focal pattern of involvement, and the relative preservation of global function, suggest a milder involvement of the myocardium-in most of these patients-than has previously been observed in classic viral and COVID-19 myocarditis. Conclusion: Three adolescent males developed acute myocarditis within days of their second mRNA COVID-19 vaccination. CMR in combination with serum troponin measurements was critical for diagnosis, and arrythmia monitoring was critical in their follow up. Repeat CMR studies over the six months following diagnosis will be important to rule out development of post-inflammatory fibrosis and long-term arrhythmias. Legend: A and B: LGE (Magnitude IR) and PSIR (Phase sensitive IR), respectively, showing patchy epicardial enhancement at the basal inferolateral and inferior segments;C: Abnormal ECV at basal anterolateral, inferolateral and inferior segments;D and E: ECV and T1 bullseye maps with abnormal values;F and G: Patchy visible edema at basal inferolateral, anterolateral and inferior segments on the T2 and T2 color map;H: Bullseye map showing T2 values;I: Asymmetric Right axillary lymphadenopathy secondary to vaccination in the right arm.

Optimizing social and economic activity while containing SARS-CoV-2 transmission using DAEDALUS

To study the trade-off between economic, social and health outcomes in the management of a pandemic, DAEDALUS integrates a dynamic epidemiological model of SARS-CoV-2 transmission with a multi-sector economic model, reflecting sectoral heterogeneity in transmission and complex supply chains. The model identifies mitigation strategies that optimize economic production while constraining infections so that hospital capacity is not exceeded but allowing essential services, including much of the education sector, to remain active. The model differentiates closures by economic sector, keeping those sectors open that contribute little to transmission but much to economic output and those that produce essential services as intermediate or final consumption products. In an illustrative application to 63 sectors in the United Kingdom, the model achieves an economic gain of between £161 billion (24%) and £193 billion (29%) compared to a blanket lockdown of non-essential activities over six months. Although it has been designed for SARS-CoV-2, DAEDALUS is sufficiently flexible to be applicable to pandemics with different epidemiological characteristics. © 2022, The Author(s), under exclusive licence to Springer Nature America, Inc.

Power on: The rapid transition of a large interdisciplinary behavioral health department to telemental health during the COVID-19 pandemic.

BACKGROUND: The COVID-19 pandemic necessitated a rapid transition to telemental health (TMH) for behavioral health services in the behavioral health department of a large integrated primary care organization. Although the COVID-19 pandemic was the initial trigger for rapid organizational change, systems were developed with a focus on longer term scalability and sustainability. METHODS: This paper discusses the process of organizational change within our healthcare delivery system using the Strengths, Opportunities, Aspirations, and Results (SOAR) framework. Within this framework a structured mixed methods survey of 38 clinicians representing 5 different disciplines was conducted. Internal and survey data were analyzed to evaluate and guide the iterative change process. RESULTS: The majority of BH clinicians reported that they were as or more effective with TMH. The transition to TMH in our organization resulted in increased access to care, with a 10.3% increase in BH visit completions. The transition to TMH may benefit clinician work-life balance, but requires resources to support clinical, technological, and communication/teamwork changes. IMPLICATIONS/CONCLUSIONS: TMH is a feasible treatment modality for integrated care settings. It is cost-effective and well-accepted by clinicians. The SOAR framework can be used to guide rapid organizational change and ongoing QI processes.

A symptom and risk factor questionnaire accurately predicts upper gastrointestinal cancer

Introduction Waiting times for endoscopy are rising rapidly following the COVID-19 pandemic. In addition, cancers may be missed as patients are placed on routine waiting lists but not monitored. Some hospitals use the Edinburgh Dysphagia Score to assess and prioritise patients for investigation. This offers a sensitivity of 98.4% and specificity of 9.3% to detect malignancy in patients presenting with dysphagia.4 However, it is not designed for detecting gastric cancer. We aimed to create a more accurate screening questionnaire as an aid to triaging referrals. Methods Patients were recruited as part of the Saliva to Predict rIsk of disease using Transcriptomics and epigenetics (SPIT) study. Patients were recruited from 2 week-wait suspected upper gastrointestinal cancer pathway referrals at 20 hospitals in the United Kingdom. The cohort was further enriched with patients found to have oesophageal adenocarcinoma on emergency hospital admission. They completed over 200 questions about a wide variety of symptoms and risk factors. After data cleaning, 800 patients were available for evaluation. Of these, 80 had upper GI cancer. A machine learning model was developed to identify those at highest risk of having upper GI cancer using a 'cost-based' approach which maximises the chance of detecting cancer. Information gain was followed by correlated feature selection and a multivariable logistic regression curve was created with scores from 0 (cancer very unlikely) to 100 (cancer very likely). The training dataset used 80% of the data and the model was tested with the other 20%. Results 20 features were found to be important and reproducible. They included age, sex, dysphagia, odynophagia, early satiety, weight loss, duration of chest pain and regurgitation, frequency of acid taste in the mouth, a previous history of smoking, cancer or psychological disorders, current anxiety level and frequency of vegetable intake. The area under the receiver operator curve to detect cancer was 0.83. 50% of cancers scored greater than 85 whereas 50% of normals scored less than 25. At a cut-off score of 10, sensitivity was 98.7% with specificity 26.8% to detect cancer (figure). Conclusions We have created a simple, reproducible risk score to identify patients at high and low risk of upper GI cancer. It performs better than previous scores but now needs testing in the real world. It might be usable to both upgrade routine patients to urgent endoscopy and remove patients at very low risk from waiting lists, thereby helping to prioritise patients with a greater clinical need and reducing the endoscopic backlog.

A symptom and risk factor questionnaire accurately predicts upper gastrointestinal cancer

Introduction: Waiting times for endoscopy are rising rapidly following the COVID-19 pandemic, leading to significant backlogs.1 Modelling has demonstrated that delays in presentation to health services and delays in completing diagnostic procedures will lead to excess mortality.2 In addition, many cancers are likely to be missed as patients are placed on routine waiting lists but are not regularly monitored. Some hospitals use the Edinburgh Dysphagia Score to risk assess and prioritise patients for investigation.3 This offers a sensitivity of 98.4% and specificity of 9.3% to detect malignancy in patients presenting with dysphagia.4 However, it is primarily not designed for detecting gastric cancer. We aimed to create a more accurate screening questionnaire to risk assess patients and prioritise those who need early endoscopy. Aims & Methods: Patients were recruited as part of the Saliva to Predict rIsk of disease using Transcriptomics and epigenetics (SPIT) study. Ethical approval was gained from the Coventry and Warwickshire Regional Ethics Committee (17/WM/0079). Patients were recruited from 2 week-wait pathway referrals at 20 hospitals in the United Kingdom, which is used by physicians to refer patients who have may suspected cancer for further investigation The cohort was further enriched with patients found to have oesophageal adenocarcinoma on emergency hospital admission. They completed over 200 questions about a wide variety of symptoms and risk factors. After data cleaning, 800 patients were available for evaluation. Of these, 80 had upper GI cancer. A machine learning model was developed to identify those at highest risk of having upper GI cancer using a 'cost-based' approach which maximises the chance of detecting cancer. Information gain was followed by correlated feature selection and a multivariable logistic regression curve was created with scores from 0 (cancer very unlikely) to 100 (cancer very likely). The training dataset used 80% of the data and the model was tested with the other 20%. Results: 20 features were found to be important and reproducible. They included age, sex, dysphagia, odynophagia, early satiety, weight loss, duration of chest pain and regurgitation, frequency of acid taste in the mouth, a previous history of smoking, cancer or psychological disorders, current anxiety level and frequency of vegetable intake. The area under the receiver operator curve to detect cancer was 0.83. 50% of cancers scored greater than 85 whereas 50% of normals scored less than 25. At a cut-off score of 10, sensitivity was 98.7% with specificity 26.8% to detect cancer. Conclusion: We have created a simple, reproducible risk score to identify patients at high and low risk of upper GI cancer. It performs better than previous scores but now needs testing in the real world. It might be usable to both upgrade routine patients to urgent endoscopy and remove patients at very low risk from waiting lists, thereby helping to prioritise patients with a greater clinical need and reducing the endoscopic backlog.

COVIDCube: An RDF Data Cube for Exploring Among-Country COVID-19 Correlations

We present an RDF Data Cube { integrated from numerous sources on the Web { that describes countries in terms of general vari-Ables (e.g., GDP, population density) and COVID-19 variables. On top of this data cube, we develop a system that computes and visualises cor-relations between these variables, providing insights into the factors that correlate with COVID-19 cases, deaths, etc., on an international level. © 2021 CEUR-WS. All rights reserved.

Behaviour and heart rate response in children and adolescents undergoing anesthesia with considerably altered (Covid-19-Related) safety precautions

Introduction The COVID-19 pandemic imposed on us the requirement for fast and significant change to our anesthetic practice. Some anesthetic safety precautions (e.g. full personal protective equipment and restricting parent-attendance at anesthesia induction) had the potential to alter behavioral and physiological (anxiety-related) responses in children and adolescents undergoing surgery during this time, despite increased use of oral pre-sedation. We explored our unique dataset in order to provide preliminary and opportunistic data addressing this issue. Methods 93 children and adolescents (1-16 years) underwent anesthesia for surgery at our large tertiary hospital during the first U.K. peak of the COVID-19 pandemic (approximately April 2020). All anesthetics were performed with COVID-19 safety precautions. A control group consisted of 91 children and adolescents undergoing anesthesia for surgery immediately before the peak-pandemic. Behavior was assessed by anesthetist-rating of cooperation and calmness of children and adolescents at anesthesia induction, and subsequently by recovery nurse subjective rating. Multiple heart rate values obtained for each child / adolescent pre- and post-procedure were age-normalised and explored for incidence of tachycardia, which can relate to anxiety. Results Accounting for age group, behavior and heart rate values were comparable across groups. This was despite significant changes to anesthetic practice including reduced use of inhalational induction and therefore increased awake intravenous cannulation. Conclusion Behavior and heart rate data indicate some stability in children's and adolescent's acute response to anesthesia in the context of otherwise significant change to our practice. However, efforts should still be directed at assessing possible late-emerging behavioral and emotional responses to this altered-anesthesia experience, as this may yet influence how this cohort of children and adolescents engages with any future healthcare procedures.