An emphasis on overly broad notions of generalisability as it pertains to applications of machine learning in health care can overlook situations in which machine learning might provide clinical utility. We believe that this narrow focus on generalisability should be replaced with wider considerations for the ultimate goal of building machine learning systems that are useful at the bedside.

The myth of generalisability in clinical research and machine learning in health care.

We developed an explainable artificial intelligence (AI) early warning score (xAI-EWS) system for early detection of acute critical illness. While maintaining a high predictive performance, our system explains to the clinician on which relevant electronic health records (EHRs) data the prediction is grounded. Acute critical illness is often preceded by deterioration of routinely measured clinical parameters, e.g., blood pressure and heart rate. Early clinical prediction is typically based on manually calculated screening metrics that simply weigh these parameters, such as Early Warning Scores (EWS). The predictive performance of EWSs yields a tradeoff between sensitivity and specificity that can lead to negative outcomes for the patient. Previous work on EHR-trained AI systems offers promising results with high levels of predictive performance in relation to the early, real-time prediction of acute critical illness. However, without insight into the complex decisions by such system, clinical translation is hindered. In this letter, we present our xAI-EWS system, which potentiates clinical translation by accompanying a prediction with information on the EHR data explaining it.

Explainable artificial intelligence model to predict acute critical illness from electronic health records

Background: Continuous monitoring of vital signs by using wearable wireless devices may allow for timely detection of clinical deterioration in patients in general wards in comparison to detection by standard intermittent vital signs measurements. A large number of studies on many different wearable devices have been reported in recent years, but a systematic review is not yet available to date.
Objective: The aim of this study was to provide a systematic review for health care professionals regarding the current evidence about the validation, feasibility, clinical outcomes, and costs of wearable wireless devices for continuous monitoring of vital signs.
Methods: A systematic and comprehensive search was performed using PubMed/MEDLINE, EMBASE, and Cochrane Central Register of Controlled Trials from January 2009 to September 2019 for studies that evaluated wearable wireless devices for continuous monitoring of vital signs in adults. Outcomes were structured by validation, feasibility, clinical outcomes, and costs. Risk of bias was determined by using the Mixed Methods Appraisal Tool, quality assessment of diagnostic accuracy studies 2nd edition, or quality of health economic studies tool.
Results: In this review, 27 studies evaluating 13 different wearable wireless devices were included. These studies predominantly evaluated the validation or the feasibility outcomes of these devices. Only a few studies reported the clinical outcomes with these devices and they did not report a significantly better clinical outcome than the standard tools used for measuring vital signs. Cost outcomes were not reported in any study. The quality of the included studies was predominantly rated as low or moderate.
Conclusions: Wearable wireless continuous monitoring devices are mostly still in the clinical validation and feasibility testing phases. To date, there are no high quality large well-controlled studies of wearable wireless devices available that show a significant clinical benefit or cost-effectiveness. Such studies are needed to help health care professionals and administrators in their decision making regarding implementation of these devices on a large scale in clinical practice or in-home monitoring.

/pdf/current-evidence-for-continuous-vital-signs-monitoring-by-5juyaseli3.pdf

Current Evidence for Continuous Vital Signs Monitoring by Wearable Wireless Devices in Hospitalized Adults: Systematic Review

Background: Patient-reported outcome measures (PROMs) are important in clinical practice and research. The growth of electronic health technologies provides unprecedented opportunities to systematically collect information via PROMs.
Objective: The aim of this study was to provide an objective and comprehensive overview of the benefits, barriers, and disadvantages of the digital collection of qualitative electronic patient-reported outcome measures (ePROMs).
Methods: We performed a systematic review of articles retrieved from PubMED and Web of Science. The Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines were followed during all stages. The search strategy yielded a total of 2333 records, from which 32 met the predefined inclusion and exclusion criteria. The relevant ePROM-related information was extracted from each study.
Results: Results were clustered as benefits and disadvantages. Reported benefits of ePROMs were greater patient preference and acceptability, lower costs, similar or faster completion time, higher data quality and response rates, and facilitated symptom management and patient-clinician communication. Tablets were the most used ePROM modality (14/32, 44%), and, as a platform, Web-based systems were used the most (26/32, 81%). Potential disadvantages of ePROMs include privacy protection, a possible large initial financial investment, and exclusion of certain populations or the “digital divide.”
Conclusions: In conclusion, ePROMs offer many advantages over paper-based collection of patient-reported outcomes. Overall, ePROMs are preferred over paper-based methods, improve data quality, result in similar or faster completion time, decrease costs, and facilitate clinical decision making and symptom management. Disadvantages regarding ePROMs have been outlined, and suggestions are provided to overcome the barriers. We provide a path forward for researchers and clinicians interested in implementing ePROMs.
Trial Registration: PROSPERO CRD42018094795; https://www.crd.york.ac.uk/prospero/display_record.php?RecordID=94795

Benefits and Disadvantages of Electronic Patient-reported Outcome Measures: Systematic Review.

The impact of continuous versus intermittent vital signs monitoring in hospitals: A systematic review and narrative synthesis.

Strengths and limitations of early warning scores: A systematic review and narrative synthesis


 
 
 Small bowel obstruction (SBO) is associated with significant morbidity and mortality (90 day mortality: 13%). Common causes include: adhesions, hernias or tumours. Conservative management includes nil per oral, intravenous hydration and naso-gastric decompression. If conservative management fails (>72 hours) patients usually require surgery.
 Recent national and international guidelines (ASGBI, Bologna, etc.) however advocate early use of gastrografin for the management of SBO.
 The aim of the audit is therefore to review current HDFT practice of gastrografin use for the management of small bowel obstruction.
 
 
 
 Retrospective data collection between April 2020- January 2022. All patients who underwent surgery for SBO were included. Exclusion: ischaemic or strangulated bowel, large bowel obstruction.
 
 
 
 A total of 50 patients were included in the analysis (25 male: 25 female) with a mean age of 73 years. 46% of the patients had undergone previous abdominal surgery. Cardiovascular conditions were the commonest pre-operative co-morbidity (32% had cardiac co-morbidities on presentation).
 Adhesions were the commonest cause of SBO (64%). Gastrografin was only administered in 12% of patients prior to surgery. Majority of the patients underwent laparotomy (92%) with 58% requiring bowel resection during surgery (27% of these required a stoma).
 52% of the patients undergoing surgery were admitted to ICU post-operatively with a mean stay of 4 days. The mean length of hospital stay was 19 days. The 30 and 90 day post-operative complication rate was 52% and 6%, respectively. Commonest post-operative complication was related to surgery (44% either developed wound infection, ileus, collections, leak, obstruction, etc.).
 
 
 
 Emergency laparotomy for SBO is associated with significant morbidity and hospital stay. Use of Gastrografin for diagnostic/ therapeutic effect is suboptimal in HDFT and requires further work.


/pdf/egs-p27-evaluating-the-use-of-gastrografin-for-the-28sar6bi.pdf

W. Tahir

Papers

Strengths and limitations of early warning scores: A systematic review and narrative synthesis

EGS P27 Evaluating the use of gastrografin for the management of small bowel obstruction at Harrogate District Foundation Trust