Human-artificial intelligence interaction in gastrointestinal endoscopy.

The number and variety of applications of artificial intelligence (AI) in gastrointestinal (GI) endoscopy is growing rapidly. New technologies based on machine learning (ML) and convolutional neural networks (CNNs) are at various stages of development and deployment to assist patients and endoscopists in preparing for endoscopic procedures, in detection, diagnosis and classification of pathology during endoscopy and in confirmation of key performance indicators. Platforms based on ML and CNNs require regulatory approval as medical devices. Interactions between humans and the technologies we use are complex and are influenced by design, behavioural and psychological elements. Due to the substantial differences between AI and prior technologies, important differences may be expected in how we interact with advice from AI technologies. Human-AI interaction (HAII) may be optimised by developing AI algorithms to minimise false positives and designing platform interfaces to maximise usability. Human factors influencing HAII may include automation bias, alarm fatigue, algorithm aversion, learning effect and deskilling. Each of these areas merits further study in the specific setting of AI applications in GI endoscopy and professional societies should engage to ensure that sufficient emphasis is placed on human-centred design in development of new AI technologies.

A corporeal conundrum: Challenges posed by remote consultation for postgraduate medical education.

BACKGROUND: The COVID-19 pandemic accelerated the use of remote consultation in hospital outpatient clinics. Remote consultation alters the clinical environment and the learning environment in ways that are incompletely understood. This research sought to explore how trainees negotiate training and learning in such an environment when it is novel to them. METHODS: Purposive sampling was used to recruit eight doctors from the gastroenterology department of an academic teaching hospital. Four consultants and four trainees participated in individual, semi-structured interviews. Interpretative phenomenological analysis of interview transcripts was employed and themes developed from the analysis, to characterise the experience of learning and teaching in remote consultation clinics, as described by participants. RESULTS: Participants described how they try to create mental representations of each patient they review by remote consultation. Whilst consultants found this task relatively easy, trainee physicians found remote consultation more challenging and highlighted the importance of the physical presence of the patient to help them form a holistic sense of the patient's condition. Doctors in training also struggled to develop a workable mental model of the patient's condition when physical examination was precluded by remote consultation. CONCLUSIONS: This study highlights the place of the patient's physical presence as an essential educational stimulus to facilitate teaching and learning. Further research is needed to characterise the processes clinicians use to formulate mental models of patients who are physically absent from the consultation room.

Metabolic dysfunction and diabetes mellitus during long-term follow-up of severe acute pancreatitis: A case-matched study.

BACKGROUND/OBJECTIVES: Endocrine insufficiency following severe acute pancreatitis (SAP) leads to diabetes of the exocrine pancreas, (type 3c diabetes mellitus), however it is not known how this metabolic phenotype differs from that of type 2 diabetes, or how the two subtypes can be differentiated. We sought to determine the prevalence of diabetes following SAP, and to analyse the behaviour of glucose and pancreatic hormones across a 2-h oral glucose tolerance test (OGTT). METHODS: Twenty-six patients following SAP (mean (range) duration of first SAP episode to study time of 119.3 (14.8-208.9) months) along with 26 matched controls underwent an OGTT with measurement of glucose, insulin, c-peptide, glucagon and pancreatic polypeptide (PP) at fasting/15/90/120min. Beta-cell area was estimated using the 15min c-peptide/glucose ratio, and insulin resistance (IR) using homeostasis model assessment (HOMA) and oral glucose insulin sensitivity (OGIS) models. RESULTS: The prevalence of diabetes/prediabetes was 54% following SAP (38.5% newly-diagnosed compared to 19.2% newly-diagnosed controls). Estimated beta-cell area and IR did not differ between groups. AUC c-peptide was lower in SAP versus controls. AUC insulin and AUC c-peptide were lower in SAP patients with diabetes versus controls with diabetes; between-group differences were observed at the 90 and 120 min time-points only. Half of new diabetes cases in SAP patients were only identified at the 120min timepoint. CONCLUSIONS: Diabetes and pre-diabetes occur frequently following SAP and are difficult to distinguish from type 2 diabetes in controls but are characterised by reduced insulin and c-peptide at later stages of an OGTT. Consistent with this observation, most new post SAP diabetes cases were diagnosed by 2-h glucose levels only.