Jessica Caterson

Jessica Caterson

Auto-generated profile, based on publicly available information

Jessica Caterson

Hamlyn Centre, Department of Surgery and Cancer, Imperial College London, London, United Kingdom

Research Focus & Metrics

Academic Career Length

7Years (2020-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
1 January 2024 in Digital Health
Verified
Journal Article
Open access
JCJessica Caterson
Jessica Caterson
ALAlexandra Lewin
Alexandra Lewin
EWElizabeth Williamson
Elizabeth Williamson
Machine Learning (ML) and Deep Learning (DL) models show potential in surpassing traditional methods including generalised linear models for healthcare predictions, particularly with large, complex datasets. However, low interpretability hinders practical implementation. To address this, Explainable Artificial Intelligence (XAI) methods are proposed, but a comprehensive evaluation of their effectiveness is currently limited. The aim of this scoping review is to critically appraise the application of XAI methods in ML/DL models using Electronic Health Record (EHR) data. In accordance with PRISMA scoping review guidelines, the study searched PUBMED and OVID/MEDLINE (including EMBASE) for publications related to tabular EHR data that employed ML/DL models with XAI. Out of 3220 identified publications, 76 were included. The selected publications published between February 2017 and June 2023, demonstrated an exponential increase over time. Extreme Gradient Boosting and Random Forest models were the most frequently used ML/DL methods, with 51 and 50 publications, respectively. Among XAI methods, Shapley Additive Explanations (SHAP) was predominant in 63 out of 76 publications, followed by partial dependence plots (PDPs) in 11 publications, and Locally Interpretable Model-Agnostic Explanations (LIME) in 8 publications. Despite the growing adoption of XAI methods, their applications varied widely and lacked critical evaluation. This review identifies the increasing use of XAI in tabular EHR research and highlights a deficiency in the reporting of methods and a lack of critical appraisal of validity and robustness. The study emphasises the need for further evaluation of XAI methods and underscores the importance of cautious implementation and interpretation in healthcare settings.

3 April 2023 in Bone & Joint Open
Verified
Journal Article
Open access
PDF file
APArwel T. Poacher
Arwel T. Poacher
JFJoseph L. J. Froud
Joseph L. J. Froud
JCJessica Caterson
Jessica Caterson
DCDaniel L. Crook
Daniel L. Crook
GRGregor Ramage
Gregor Ramage
ECEleanor C. Carpenter
Eleanor C. Carpenter
The cost effectiveness of potential risk factors for developmental dysplasia of the hip within a national screening programme

1 September 2020 in Bone & Joint Open
Verified
Journal Article
Open access
JCJessica Caterson
Jessica Caterson
MWMatthew A Williams
Matthew A Williams
CMCatherine McCarthy
Catherine McCarthy
NANicholas Athanasou
Nicholas Athanasou
HTH Thomas Temple
H Thomas Temple
CGChristopher L.Max H. Gibbons
Christopher L.Max H. Gibbons
Aims The aticularis genu (AG) is the least substantial and deepest muscle of the anterior compartment of the thigh and of uncertain significance. The aim of the study was to describe the anatomy of AG in cadaveric specimens, to characterize the relevance of AG in pathological distal femur specimens, and to correlate the anatomy and pathology with preoperative magnetic resonance imaging (MRI) of AG. Methods In 24 cadaveric specimens, AG was identified, photographed, measured, and dissected including neurovascular supply. In all, 35 resected distal femur specimens were examined. AG was photographed and measured and its utility as a surgical margin examined. Preoperative MRIs of these cases were retrospectively analyzed and assessed and its utility assessed as an anterior soft tissue margin in surgery. In all cadaveric specimens, AG was identified as a substantial structure, deep and separate to vastus itermedius (VI) and separated by a clear fascia! plane with a discrete neurovascular supply. Mean length of AG was 16.1 cm ( +/- 1.6 cm) origin anterior aspect distal third femur and insertion into suprapatellar bursa. In 32 of 35 pathological specimens, AG was identified (mean length 12.8 cm ( +/- 0.6 cm)). Where AG was used as anterior cover in pathological specimens all surgical margins were clear of disease. Of these cases, preoperative MRI identified AG in 34 of 35 cases (mean length 8.8 cm ( +/- 0.4 cm)). Results AG was best visualized with T1-weighted axial images providing sufficient cover in 25 cases confirmed by pathological findings.These results demonstrate AG as a discrete and substantial muscle of the anterior compartment of the thigh, deep to VI and useful in providing anterior soft tissue margin in distal femoral resection in bone tumours. Conclusion Preoperative assessment of cover by AG may be useful in predicting cases where AG can be dissected, sparing the remaining quadriceps muscle, and therefore function.

6 August 2026 in Annals of Surgery
Verified
Journal Article
Open access
JCJessica Caterson
Jessica Caterson
MGMary S.L. Goble
Mary S.L. Goble
VCVirginia Caddick
Virginia Caddick
ZLZhenchen Lin
Zhenchen Lin
ADAra Darzi
Ara Darzi
Daniel LeffDaniel Leff
Daniel Leff
Objective: This study aimed to characterize neural activation in novice surgical residents with increasing cognitive load and its association with performance. Background: Evaluating the impact of cognitive load on surgical performance is crucial for training and patient safety. Excessive workload can lead to “cognitive overload,” impairing performance. Prior work links overload in senior residents to deactivation in the prefrontal cortex, associated with attention and concentration, but neurocognitive responses in novice surgical residents remain unclear. Methods: Twenty novice surgeons performed a suturing task under 3 conditions: self-paced (SP), time-pressured (TP), and time and cognitive stress (CS). Functional near-infrared spectroscopy (fNIRS) recorded activation across the prefrontal and premotor cortices. Suturing performance was assessed with a validated score, subjective cognitive load with the SURG-TLX questionnaire, and physiological stress with heart rate variability (HRV). Results: Subjective cognitive load increased significantly with task complexity ( P <0.00001), while suturing performance declined ( P <0.00001). HRV did not differ significantly between tasks ( P =0.5905). Neural activation increased in the dorsolateral prefrontal cortex (DLPFC), premotor cortex (PMC), and supplementary motor area (SMA) with escalating load. Left DLFPC activation increased in “maintainers” who preserved performance (HbO 2.Ch20 = 10.8, P =0.005) but attenuated in “decliners” with performance deterioration (HHb Ch21 =6.4, P =0.041). Frontal PFC deactivation was greater in decliners (SP: ∆HHb Ch57 P =0.048; ∆HHb Ch57 P =0.011). Conclusion: The DLPFC supports planning and attention, the frontal PFC multitasking, and PMC/SMA motor preparation. These findings contribute to our understanding of cognitive overload resilience in early surgical training and provide a foundation for informing safer surgical training.

6 October 2026 in SSRN Electronic Journal
Verified
Preprint
Open access
JCJessica Caterson
Jessica Caterson
EFEvelynne Fulda
Evelynne Fulda
JRJamie Robinson
Jamie Robinson
ATAnya Topiwala
Anya Topiwala
JDJoshua C. Denny
Joshua C. Denny
SDStefan van Duijvenboden
Stefan van Duijvenboden
Background: Pre-operative cardiorespiratory fitness assessment informs surgical risk stratification, but objective in-hospital assessment is challenging and limited to high-risk cohorts. Wearable devices, already used by millions worldwide, may provide an accessible objective measure for broader surgical populations.Methods: This retrospective cohort study evaluated wearable-derived daily step count as a scalable pre-operative risk marker in 9,187 participants undergoing surgery across UK Biobank (n=5,034) and All of Us (n=4,153). Step counts within 1 year before surgery, stratified by established step count groupings. Outcomes of interest, post-operative complications and hospital length of stay, and covariates were ascertained from linked electronic health record data.Findings: After adjustment for demographic, comorbidity and surgical factors, participants taking <2,500 versus ≥10,000 steps daily had higher odds of 30-day post-operative complications (adjusted odds ratio (aOR), 2.24; 95% CI, 1.47–3.41) and longer hospital stay (adjusted incidence rate ratio (aIRR), 1.74; 95% CI, 1.42–2.14). Those taking 2,500–4,999 steps also had higher complication risk (aOR, 1.46; 95% CI, 1.08–1.97) and longer stay (aIRR, 1.22; 95% CI, 1.09–1.36).Interepretation: These findings support wearable-derived pre-operative step count as an independent, externally validated marker of peri-operative risk. By identifying patients at higher risk of complications and prolonged hospital stay, step count can complement existing pre-operative risk assessment and help target optimisation strategies before surgery.FundingCancer Research UK, National Insititute of Health Researc Oxford BRC, Wellcome Trust [223100/Z/21/Z, 227093/Z/23/Z, 306069/Z/23/Z], GSK, Boehringer Ingelheim, Google, National Institutes of Health’s Oxford Cambridge Scholars Program, EPSRC Centre for Doctoral Training in Health Data Science (EP/S02428X/1), British Heart Foundation Centre of Research Excellence (grant number RE/18/3/34214), Danish National Research Foundation, National Human Genome Research Institute intramural research program (ZIA HG200417).