Finn Catling

Imperial College London | Imperial · Section of Anaesthetics, Pain Medicine, and Intensive Care (APMIC)

Background: Patients’ encounters with healthcare services must undergo clinical coding. These codes are typically derived from free-text notes. Manual clinical coding is expensive, time-consuming and prone to error. Automated clinical coding systems have great potential to save resources, and realtime availability of codes would improve oversight o...

Background Ventilator-associated pneumonia (VAP) often presents ambiguously and is treated empirically. Definitions of VAP vary widely across the literature, reflecting this ambiguity. Clinical coding of VAP is consequently unreliable, and discouraged in institutions using VAP incidence as a quality indicator.[1,2] An alternative, more sensitive de...

Objective: Clinical interventions and death in the intensive care unit (ICU) depend on complex patterns in patients' longitudinal data. We aim to anticipate these events earlier and more consistently so that staff can consider preemptive action. Materials and methods: We use a temporal convolutional network to encode longitudinal data and a feed...

Predictive models can estimate risk of death in emergency laparotomy, but currently only make point predictions of mortality risk. This masks uncertainty in the predictions which can lead to over-confident decision making. In this study, we develop and validate an alternative model to address this, using data from 127148 UK patients. This novel app...

Clinical prediction models typically make point estimates of risk. However, values of key variables are often missing during model development or at prediction time, meaning that the point estimates mask significant uncertainty and can lead to over-confident decision making. We present a model of mortality risk in emergency laparotomy which instead...

Background: We conducted a scoping review of machine learning systems that inform individualised cardiovascular resuscitation of adults in hospital with sepsis. Our study reviews the resuscitation tasks that the systems aim to assist with, system robustness and potential to improve patient care, and progress towards deployment in clinical practice....

Clinical decisions in the ICU depend on the patient's background, and on complex short- and long-term patterns in their longitudinal data. This study uses neural networks to derive a rich mathematical representation of patients' status, and uses this to better predict clinical interventions and death. We aim to anticipate these events earlier and m...

Aim: The GMC have mandated Local Education and Training Boards and medical schools to collect evidence on how trainers meet specified “trainer criteria.” Trainees are accustomed to collating such evidence using online logbooks. The aim was to compare consultant and trainee uptake of an online logbook to provide this GMC evidence. Method: Anonymous...

Objectives (1) Review current validated scar assessment tools and (2) describe the impact of scar cosmesis perception on body image and quality of life. Methods Three independent reviewers performed comprehensive searches and identified 680 English language studies published between 1950 and 2014 (data sources: Medline, EMBASE, Cochrane Library, a...

UK medical students' confidence in their prescribing skills is low, and a significant proportion of prescriptions written by foundation year 1 (FY1) doctors contain errors. The Prescribing Safety Assessment (PSA) is a new national examination aimed at ensuring prescribing competence in undergraduates, but few PSA-specific preparatory resources are...