Electronic Patient Record Data Research Articles

Method to apply temporal graph analysis on electronic patient record data to explore healthcare professional–patient interaction intensity: a cohort study

AimInteractions between patients and healthcare professionals (HCP) during hospital admissions are complex and difficult to interrogate using traditional analysis of electronic patient record (EPR) data. The objective of this study...

The use of locum doctors in the NHS: understanding and improving the quality and safety of care.

The use of locum doctors in the National Health Service is widely believed to have increased, and there have been widespread and sustained concerns among policy-makers, healthcare providers, professional associations and professional regulators about the quality/safety, cost and effective use of locum doctors. To provide evidence on the extent, quality and safety of medical locum practice and the implications of medical locum working for health service organisation and delivery in primary and secondary care in the English National Health Service, to support policy and practice. Four interlinked work packages involving surveys of National Health Service trusts and of general practices in England; semistructured interviews and focus groups across 11 healthcare organisations in England; analysis of existing routine data sets on the medical workforce in primary care and in National Health Service trusts in England from National Health Service Digital and National Health Service Improvement; and analysis of data from the Clinical Practice Research Datalink in primary care and of electronic patient record data from two National Health Service hospitals in secondary care. In primary care, about 6% of general practice medical consultations were undertaken by locums in 2010 and this had risen slightly to about 7.1% in 2021. In National Health Service trusts (mostly secondary care and mental health), about 4.4% of medical staff full-time equivalent was provided by locum doctors. But those overall national rates of locum use hide a great deal of variation. In primary care, we found the National Health Service Digital workforce returns showed the rate of locum use by Clinical Commissioning Group varied from 1% to almost 31%. Among National Health Service trusts, the reported rate of locum use varied from < 1% to almost 16%. We found that there was poor awareness of and adherence to national guidance on locum working arrangements produced by National Health Service England. Our research showed that locum working can have adverse consequences for the quality and safety of care, but that such consequences were probably more likely to result from the organisational setting and the working arrangements than they were from the locum doctors themselves and their competence, clinical practice or behaviours. Our research was hampered in some respects by the COVID pandemic which both resulted in some delays and other challenges. Our efforts to use electronic patient record data in secondary care to explore locum doctor working were stymied by the problems of data access and quality. Locum doctors are a key component of the medical workforce in the National Health Service, and provide necessary flexibility and additional capacity for healthcare organisations and services. We found that the extent of reliance on locum doctors varied considerably, but that an over-reliance on locums for service provision was undesirable. Some differences in practice and performance between locum and permanent doctors were found, but these seemed often to arise from organisational characteristics. We found that patients were more concerned with the clinical expertise and skills of the doctor they saw than whether they were a locum or not. Organisational arrangements for locum working could be improved in many respects. This award was funded by the National Institute for Health and Care Research (NIHR) Health and Social Care Delivery Research programme (NIHR award ref: NIHR128349) and is published in full in Health and Social Care Delivery Research; Vol. 12, No. 37. See the NIHR Funding and Awards website for further award information.

Digital Exclusion, Social Deprivation, and Clinical Outcomes of Patients Undergoing Hyperacuity Home Monitoring.

Digital exclusion is a growing challenge when deploying digital patient care pathways and a potential barrier to widespread implementation, especially in the field of smartphone-based self-monitoring of vision. This retrospective case series seeks to examine the characteristics of individuals who adhere to a smartphone homemonitoring programme using the Alleye app for retinal disease, with a focus on digital exclusion, social deprivation and clinical outcomes. We conducted a retrospective analysis of 89 patients with retinal pathologies including diabetic retinopathy and retinal vein occlusions at Moorfields Eye Hospital participating in an Alleye homemonitoring programme between April 2020 and November 2022. Postcodes were used to determine the Digital Exclusion Risk Index (DERI) and the Index of Multiple Deprivation (IMD) rebased for London. Clinical information from the electronic patient record and Alleye app usage data were extracted for each patient. Associations between the DERI/IMD, clinical parameters and app use were examined using multivariable regression models. Mean DERI was 2.56 (standard deviation [SD] = 0.36), IMD was 6.25 (SD = 2.79), visual acuity (VA) in the better eye at study entry was 83.28 Early Treatment Diabetic Retinopathy Study (ETDRS) letters (SD = 7.92), and mean follow-up was 344.46days (SD = 260.13). During the observation period, 36% received an intravitreal injection (IVI) and VA fell by at least ten letters in approximately one in four patients. In 87.5% of patients requiring IVI, the use of the app increased. We found no association between clinical parameters and programme adherence for DERI or IMD. We found no association between high digital exclusion risk and high social deprivation with monitoring adherence to smartphone-based self-monitoring of vision, contrary to the currently available evidence. This suggests that smartphone-based self-monitoring of vision is accessible to population groups of varying digital exclusion and social deprivation risk, and can be safely employed to monitor clinical progression.

Complications and health-related quality of life following lumbosacral hemivertebra resection and short segmental fusion: a retrospective cohort study

Previous literature has provided descriptions of postoperative complications. However, these studies have included heterogeneous cohorts of hemivertebra patients undergoing different surgical methods and addressing deformities at various locations. To assess independent risk factors for complications following lumbosacral hemivertebra (LSHV) resection and to further explored the potential association between complications and health-related quality of life (HRQOL). Retrospective cohort study. We retrospectively analyzed consecutive patients who underwent LSHV resection and short segment fusion at our institution between 2010 and 2022. In the current study, the outcome measure were complications and HRQOL. Between 2010 and 2022, a total of 64 consecutive patients with LSHV underwent hemivertebra resection. The diagnostic criteria for LSHV were delineated that the hemivertebra resides caudal to L5 and beyond. Electronic patient record and radiographic data were reviewed retrospectively in our electronic database. According to complications, the patients were divided into with and without complications, and further stratified as neurological, mechanical, and infectious complications. HRQOL outcomes were collected preoperatively, 3 months postoperatively, and at last follow-up using the Scoliosis Research Society-22 (SRS-22) questionnaire. Multivariate logistic regression models were used to determine independent risk factors affecting complications. A total of 64 patients were enrolled in this study for at least 2-year follow-up. The mean follow-up was 7.3±4.2 years. In the coronal plane, the mean main curve was corrected from 27.3° preoperatively to 8.0° postoperatively, and 9.3° at last follow-up. Global spine balance was significantly improved after surgery, and kept stable during the follow-up. In all, 18 complications were recorded in 15 (23.4%) patients. In the multivariate logistic regression model, 3 independent risk factors for complication were screened out, including age (OR=1.16, 95% CI 1.01-1.34, p=.042), anterior reconstruction (OR=3.40, 95% CI 1.03-11.2, p=.045), and preoperative CB (OR=1.06, 95% CI 1.00-1.12, p=.036). The mean SRS-22 total and domain scores for the entire cohort showed improvement at last follow-up. However, in the noncomplication group, the improvements in SRS-22 scores were more substantial. Postoperative neurological and mechanical complications are common medical events following LSHV resection. The age, anterior column reconstruction, and preoperative coronal balance were the independent risk factors. Compared to the noncomplication group, the improvements in HRQOL are blunted in the complication group.

Dose-response relationship between perioperative allogeneic blood transfusion and surgical site infections following spinal surgery

BACKGROUND CONTEXTPerioperative allogeneic transfusion was generally considered to be safe. However, there had been some literatures reporting a potential association between surgical site infections (SSI) and blood transfusion. PURPOSETo determine whether perioperative blood transfusion increased the risk of SSI and to further explored whether there was a dose-response relationship. DESIGNRetrospective nested case-control study. PATIENT SAMPLEWe retrospectively analyzed consecutive patients who underwent spinal fusion surgery at our institution between July 2011 and July 2021. OUTCOME MEASUREIn the current study, the primary outcome measure was SSI. METHODSAll consecutive patients who underwent elective spinal surgery at our institution between July 2011 and July 2021 formed the retrospective cohort. Electronic patient record and radiographic data were reviewed retrospectively in our electronic database. To examine the effects of mismatched variables, we further adjusted for possible confounding factors using conditional logistic regression models. Then, we explored the nonlinear relationship between perioperative blood transfusion and SSI by a smoothed curve, with the adjustments for potential confounders. If a nonlinear relationship was observed, a two-piecewise regression model would be performed to calculate the threshold effect. RESULTSThe average time from surgery to diagnosis of SSI was 20.5 days. We matched 248 controls to 124 SSI cases. Of the 124 patients who developed SSI, 84 patients (67.7%) had deep SSI, 40 patients (32.3%) had superficial SSI. In the fully adjusted model, the risk of SSI increased by 27% for each additional unit of blood transfusion. It can be seen from the curve fitting plot that the risk of SSI has a greater increase after blood transfusion >3U. Subsequent piecewise regression identified an inflection point of 3U. CONCLUSIONWe determined that 3U was a threshold volume of allogeneic blood transfusion that shifted the risk of SSI following spinal surgery, and there was a dose-response effect.

Encryption-then-embedding-based hybrid data hiding scheme for medical images

The smart healthcare system plays a vital role in modern healthcare, facilitating the exchange of Electronic Patient Records (EPR), and improving medical care. Nevertheless, safeguarding the inherent security and confidentiality of EPR data persists as a formidable challenge. This issue demanding rigorous attention and innovative solutions. Digital watermarking safeguards the genuineness and integrity of digital images, and is widely employed. In medical imaging, it is vital to guarantee the confidentiality and security of patient data. In this paper, we propose an Encryption-Then-Embedding-Based data hiding scheme for medical images that combines cryptography and watermarking techniques. The proposed technique first encrypts the patient information using Advanced Encryption Standard-Galois/Counter Mode (AES-GCM) before embedding. The intensity-based image segmentation method is then used to select the Region of Non-interest (RONI) for embedding the encrypted patient information and the watermark using a Fused Transform-Based Method (FTBM). The experiments conducted in this study utilize a large dataset of medical data, The Cancer Genome Atlas Lung Adenocarcinoma Collection (TCGA-LUAD). The findings of this study demonstrate the efficacy of the proposed technique in safeguarding patient data while simultaneously preserving the quality of medical images. In comparison to existing techniques, the proposed approach demonstrates superior performance in terms of security, authenticity, and integrity.

Pathways to interoperable electronic patient records for health and social care: Part 1: for those involved directly in care

Here, we illustrate the clinical and technical implementation of interoperable health data for direct care from the viewpoint of practicing clinicians using examples from primary care. Interoperability allows people involved in the provision and receipt of care to seamlessly exchange and use the coded, free text and documentary data they need to inform care decisions. The pathway toward NHS interoperability to support direct care has been long, but substantial progress has been made. GP computer systems and data-recording standards, national infrastructure and hospital trusts have come a long way on the journey to fully interoperable records. GPs can now receive and utilise laboratory data; via GP2GP, they can transfer full electronic patient records when patients move practice; share health data with other health organisations through GP Connect to support patient care; and provide patients with online access to their full GP record, a cornerstone of person-centred care. Here, we describe the effective technical assets and standards that have been developed to enable electronic patient record data to be shared reliably and securely.

Peripheral Blood Cell Profile and Monocyte–High-density Lipoprotein Ratio in Alzheimer’s Disease: A Hospital-based Case-control Study

IntroductionAlzheimer’s disease (AD) is a neurodegenerative disorder characterized by a progressive decline in memory and cognitive abilities. The monocyte/HDL-C ratio (MHR) has emerged as a new marker of inflammation in recent years. The purpose of this research was to examine MHR alterations in AD and to assess its feasibility as a straightforward and easily computable biomarker for evaluating the severity of the disease.Material and methodsA retrospective case-control study was conducted with 101 patients with AD and 81 age and sex-matched controls from the hospital records. AD was diagnosed according to the National Institute of Neurological and Communicative Disorders and Stroke/Alzheimer’s Disease and Related Disorders Association (NINCDS-ADRDA) criteria. The hospital’s electronic patient record data between September 2021 and April 2022 were screened.ResultsIndividuals diagnosed as having AD were classified into three subgroups based on the progression of the disease, mild, moderate, and severe. MHR was higher in the moderate and severe dementia subgroups compared with the controls according to subgroup analysis (P=0.013). An increase in MHR was found in patients with AD. The multivariate logistic regression analysis model revealed that a one-unit increase in MHR resulted in a 1.081 times increase in the risk of AD (OR: 1.081, 95% CI: [1.005-1.162]; P= 0.035).ConclusionsHigh MHR values could not be used as a diagnostic test for AD. Instead, because it negatively correlates with MMSE, it could be a good index reflecting the increased AD risk and disease severity.

Suboptimal outcomes and treatment burden of anti-vascular endothelial growth factor treatment for diabetic macular oedema in phakic patients

ObjectivesIn England and Wales, treatment options were limited for patients with diabetic macular oedema (DMO) with phakic eyes that failed anti-vascular endothelial growth factor (anti-VEGF) treatment pre-2022. This study aimed to quantify the response to, and treatment burden of, anti-VEGF treatment in phakic eyes.MethodsRetrospective, cohort study using electronic patient record data from two UK centres between 2015 and 2020. Primary objective was proportion of phakic eyes with a suboptimal response after initial 6 months of anti-VEGF treatment. Data were available for 500 eyes from 399 patients.ResultsAt 6 months significantly more eyes had a suboptimal response to anti-VEGF treatment: 65.8% (95% CI 61.5–70.0%) vs 34.2% (95% CI 30.0–38.5%), p < 0.0001. Baseline visual acuity (VA) predicted VA outcome, however, despite greater gains in eyes with poorer VA, such eyes did not achieve the same VA levels as those who started treatment with better VA. Only 53.6% of eyes had more than three injections in the first 6 months indicating difficulties in delivering high volume/high frequency treatment. Treatment and review burden were similar over the following years regardless of response to anti-VEGF treatment.ConclusionsData confirm previous real world evidence around response to anti-VEGF treatment, importance of baseline VA and frequency of injections in predicting outcomes in a UK setting. Continuing treatment beyond 6 months in suboptimal responders imposes unnecessary treatment burden without significant change in VA. In suboptimal responders, consideration of early switch to longer acting steroid treatments may help to reduce treatment burden, whilst maintaining or improving vision.

Increase in hypoglycaemia and hyperglycaemia in people with diabetes admitted to hospital during COVID-19 pandemic

BackgroundWe used detailed information on patients with diabetes admitted to hospital to determine differences in clinical outcomes before and during the COVID-19 pandemic in the UK. MethodsThe study used electronic patient record data from Imperial College Healthcare NHS Trust. Hospital admission data for patients coded for diabetes was analysed over three time periods: pre-pandemic (31st January 2019–31st January 2020), Wave 1 (1st February 2020–30th June 2020), and Wave 2 (1st September 2020–30th April 2021). We compared clinical outcomes including glycaemia and length of stay. ResultsWe analysed data obtained from 12,878, 4008 and 7189 hospital admissions during the three pre-specified time periods. The incidence of Level 1 and Level 2 hypoglycaemia was significantly higher during Waves 1 and 2 compared to the pre-pandemic period (25 % and 25.1 % vs. 22.9 % for Level 1 and 11.7 % and 11.5 % vs. 10.3 % for Level 2). The incidence of hyperglycaemia was also significantly higher during the two waves. The median hospital length of stay increased significantly (4.1[1.6, 9.8] and 4.0[1.4, 9.4] vs. 3.5[1.2, 9.2] days). ConclusionsDuring the COVID-19 pandemic in the UK, hospital in-patients with diabetes had a greater number of hypoglycaemic/hyperglycaemic episodes and an increased length of stay when compared to the pre-pandemic period. This highlights the necessity for a focus on improved diabetes care during further significant disruptions to healthcare systems and ensuring minimisation of the impact on in-patient diabetes services. SummaryDiabetes is associated with poorer outcomes from COVID-19. However the glycaemic control of inpatients before and during the COVID-19 pandemic is unknown. We found the incidence of hypoglycaemia and hyperglycaemia was significantly higher during the pandemic highlighting the necessity for a focus on improved diabetes care during further pandemics.

A comparative study of chronic spontaneous urticaria and chronic mast cell mediated angioedema

Background: Mast cell-mediated angioedema (MC-AE) is considered a form of chronic spontaneous urticaria (CSU). Objective: To investigate the clinical and laboratory features that distinguish MC-AE from antihistamine-responsive CSU (CSU), and antihistamine-resistant CSU (R-CSU) with and without concomitant AE. Methods: A retrospective observational study using the electronic patient record data base of patients with MC-AE, CSU, R-CSU, and sex- and age-matched control group (control), with a case-control ratio of 1:2. Results: A total of 986 subjects in the CSU group, 148 in the R-CSU group, 64 in the MC-AE group, and 1198 in the control group were compared. The R-CSU group without AE was characterized by lower total IgE levels (118.5 ± 84.7 IU/mL) and higher High sensitivity-C reactive protein (hs-CRP) levels (138.9 ± 94.2 IU/mL, p = 0.027; and 7.4 ± 6.9 mg/L versus 5.1 ± 6.8 mg/L, p = 0.001) than the CSU without AE group. The R-CSU group with AE was characterized by lower total IgE levels (112.1 ± 81.3 IU/mL) than the CSU group with AE (141.7 ± 89.5 IU/mL; p < 0.001), higher hs-CRP levels (7.1 ± 6.1 mg/L versus 4.7 ± 5.9 mg/L; p < 0.001). There were fewer female subjects in the MC-AE group (31 [48.4%]) than in the CSU with AE and in the R-CSU with AE 223 (67.8%) and 18 (66.7%), respectively; p = 0.012). MC-AE group was characterized by less eyelid/perioral/facial involvement and more limb involvement than in the CSU with AE and R-CSU with AE groups (p < 0.001). Conclusion: Low IgE in MC-AE and higher IgE in CSU may signify two distinct types of immune dysregulation. Due to clinical and laboratory differences between MC-AE and CSU, we suggest questioning the assumption that MC-AE is a form of CSU.

1334 THE OXFORD COGNITIVE COMORBIDITY AND AGEING RESEARCH DATABASE (ORCHARD): DESCRIPTION OF A LARGE ACUTE CARE RESEARCH DATABASE

Abstract Background Guidelines recommend that all older hospital patients are screened for cognitive comorbidity (i.e. dementia, delirium) and frailty to inform care and target multidisciplinary team resources based mainly on evidence from studies in outpatient or specialty-specific settings. Unselected hospital-wide data are needed to inform guidance and service design and delivery, so we set up the Oxford Cognitive Comorbidity and Ageing Research Database (ORCHARD) using routinely-acquired electronic patient record (EPR) data. Methods ORCHARD includes pseudonymised EPR data on all patients &amp;gt;65 years with unplanned admission to one of four general hospitals in Oxfordshire, serving a population of 660,000. Data collected include cognitive screening (mandatory for &amp;gt;70 years) comprising dementia history, delirium diagnosis (Confusion Assessment Method—CAM), and 10-point Abbreviated Mental Test; together with nursing risk assessments, frailty, diagnoses, comorbidities (Charlson index), observations, illness acuity, laboratory tests and brain imaging. Outcomes include length of stay, delayed transfers of care, discharge destination, readmissions, death and dementia through linkage to electronic mental health records. Results ORCHARD (2017-2019) includes data from 90,303 consecutive, unselected hospital admissions across all specialties (n=60,535 [67%] inpatient versus n=29,768 [33%] day case; n=73,385 [81%] medical versus n=16,918 [19%] surgical or other). Admissions data were linked to 46,533 unique individuals (n=23,724 [52%] female) with a mean/SD age of 78.3/9.6, Index of Multiple Deprivation Decile of 7.5/2.3 and Charlson Comorbidity of 8.1/9.4 at first admission. Frailty was prevalent, with 15,423 (33%) scoring moderate and 2,471 (5%) high on the Hospital Frailty Risk Score. Complete cognitive screening data are available for 34,416 (67%) unique individuals ≥70 years with inpatient admission. Conclusion ORCHARD is a large and rich data resource that will enable studies on the burden and impact of cognitive and physical frailty in-hospital, with relevance to the design and delivery of clinical services, and understanding of healthcare resource use hospital-wide and by specialty.

Rendezvous between ambulances and prehospital physicians in the Capital Region of Denmark: a descriptive study

BackgroundIn a two-tier Emergency Medical Services response system with ambulances and physician-staffed rapid response vehicles, both units are ideally dispatched simultaneously when a physician is needed. However, when advanced resources are dispatched secondarily, a meeting point (rendezvous) is established to reduce time to advanced care. This study aims to assess the extent of rendezvous tasks, patient groups involved and physician contribution when rendezvous is activated between the primary ambulances and rapid response vehicles in the Capital Region of Denmark.MethodsWe analysed prehospital electronic patient record data from all rendezvous cases in the Capital Region of Denmark in 2018. Variables included the number of times rendezvous was activated, patient demographics, dispatch criteria, on-scene diagnosis, and prehospital treatment.ResultAmbulances requested rendezvous 2340 times, corresponding to 1.3% of all ambulance tasks and 10.7% of all rapid response vehicle dispatches. The most frequently used dispatch criterion was unclear problem n = 561 (28.8%), followed by cardiovascular n = 439 (22.5%) and neurological n = 392 (20.1%). The physician contributed with technical skills like medication n = 760 (39.0%) and advanced airway management n = 161 (8.3%), as well as non-technical skills like team leading during advanced life support n = 152 (7.8%) and decision to end futile treatment and death certificate issuance n = 73 (3.7%).ConclusionRendezvous between ambulances and physician-staffed rapid response vehicles was activated in 1.3% of all ambulance cases corresponding to 10.7% of all RRV dispatches in 2018. The three largest patient groups in rendezvous presented cardiovascular, neurological, and respiratory problems. The prehospital physician contributed with technical skills like medication and advanced airway management as well as non-technical skills like team leading during advanced life support and ending futile treatment. The high percentage of dispatch criterion unclear problem illustrates the challenge of precise dispatch and optimal use of prehospital resources. Therefore, it seems necessary to have a safe and rapid rendezvous procedure to cope with this uncertainty.

A 10-Year Review of Sunburn Injuries Presenting to the Manchester Adult and Paediatric Specialist Burn Services

Background: The incidence of sunburn injuries continues to rise despite increased awareness of the risks of sun exposure and availability of sun protection. Whilst not a significant burden on burns care services, patients remain exposed to associated risks for future development of skin malignancies. The aims of this study were to determine the burden and severity of sunburn injury presentations to the Manchester adult and children’s burns services. Methods: A 10-year retrospective review was performed of patients with sunburn injuries, presenting to the Manchester burn services between 2010 and 2019 (inclusive). Data were collected from the International Burn Injury Database (iBID), electronic patient record (EPR) and local data collection systems. The data extracted included patient demographics, sunburn characteristics and management of the burn injury including need for admission and any documented surgical interventions. Temporal correlation was determined by linear regression analysis. Results: In total, 131 paediatric and 228 adult patients with sunburn injuries were managed by the Manchester burns services over the 10-year period. Mean % total body surface area burned was low (2.00% and 2.12% in adult and paediatric patients, respectively), with the majority of injuries either superficial or superficial partial thickness. Thirty percent (30.2%) of adult and 40.5% of paediatric patients were admitted with a mean length of stay of 3.51 and 1.11 days, respectively. The presentation of sunburn injuries progressively increased over the study period with a peak in 2017 (n = 58). Similar trends in patient demographics, burn size and depth and temporal trends were observed in national data from the same period for both adult and paediatric patients. Conclusion: This 10-year retrospective cohort study demonstrates an increasing trend of sunburn injury presentations to the Manchester specialist burns services; a pattern replicated in national data from England and Wales. The majority of sunburn injuries do not present to specialist burn services; therefore, these reported injuries reflect only a fraction of the true burden of sunburn nationwide. Despite increased awareness, an obvious need for enhanced public awareness campaigns regarding sun protection is therefore needed to address this trend. The educational and preventative role of burns care services is a key component in tackling both consequences of burn injuries themselves and associated risks such as future skin cancer development.

Discrete-time survival analysis in the critically ill: a deep learning approach using heterogeneous data

Prediction of survival for patients in intensive care units (ICUs) has been subject to intense research. However, no models exist that embrace the multiverse of data in ICUs. It is an open question whether deep learning methods using automated data integration with minimal pre-processing of mixed data domains such as free text, medical history and high-frequency data can provide discrete-time survival estimates for individual ICU patients. We trained a deep learning model on data from patients admitted to ten ICUs in the Capital Region of Denmark and the Region of Southern Denmark between 2011 and 2018. Inspired by natural language processing we mapped the electronic patient record data to an embedded representation and fed the data to a recurrent neural network with a multi-label output layer representing the chance of survival at different follow-up times. We evaluated the performance using the time-dependent concordance index. In addition, we quantified and visualized the drivers of survival predictions using the SHAP methodology. We included 37,355 admissions of 29,417 patients in our study. Our deep learning models outperformed traditional Cox proportional-hazard models with concordance index in the ranges 0.72–0.73, 0.71–0.72, 0.71, and 0.69–0.70, for models applied at baseline 0, 24, 48, and 72 h, respectively. Deep learning models based on a combination of entity embeddings and survival modelling is a feasible approach to obtain individualized survival estimates in data-rich settings such as the ICU. The interpretable nature of the models enables us to understand the impact of the different data domains.

Machine learning forecasting for COVID-19 pandemic-associated effects on paediatric respiratory infections

ObjectiveThe COVID-19 pandemic and subsequent government restrictions have had a major impact on healthcare services and disease transmission, particularly those associated with acute respiratory infection. This study examined non-identifiable routine...

Artificial intelligence and clinical deterioration.

To provide an overview of the systems being used to identify and predict clinical deterioration in hospitalised patients, with focus on the current and future role of artificial intelligence (AI). There are five leading AI driven systems in this field: the Advanced Alert Monitor (AAM), the electronic Cardiac Arrest Risk Triage (eCART) score, Hospital wide Alert Via Electronic Noticeboard, the Mayo Clinic Early Warning Score, and the Rothman Index (RI). Each uses Electronic Patient Record (EPR) data and machine learning to predict adverse events. Less mature but relevant evolutions are occurring in the fields of Natural Language Processing, Time and Motion Studies, AI Sepsis and COVID-19 algorithms. Research-based AI-driven systems to predict clinical deterioration are increasingly being developed, but few are being implemented into clinical workflows. Escobar et al. (AAM) provide the current gold standard for robust model development and implementation methodology. Multiple technologies show promise, however, the pathway to meaningfully affect patient outcomes remains challenging.

Personas for Better Targeted eHealth Technologies: User-Centered Design Approach.

BackgroundThe full potential of eHealth technologies to support self-management and disease management for patients with chronic diseases is not being reached. A possible explanation for these lacking results is that during the development process, insufficient attention is paid to the needs, wishes, and context of the prospective end users. To overcome such issues, the user-centered design practice of creating personas is widely accepted to ensure the fit between a technology and the target group or end users throughout all phases of development.ObjectiveIn this study, we integrate several approaches to persona development into the Persona Approach Twente to attain a more holistic and structured approach that aligns with the iterative process of eHealth development.MethodsIn 3 steps, a secondary analysis was carried out on different parts of the data set using the Partitioning Around Medoids clustering method. First, we used health-related electronic patient record data only. Second, we added person-related data that were gathered through interviews and questionnaires. Third, we added log data.ResultsIn the first step, 2 clusters were found, with average silhouette widths of 0.12 and 0.27. In the second step, again 2 clusters were found, with average silhouette widths of 0.08 and 0.12. In the third step, 3 clusters were identified, with average silhouette widths of 0.09, 0.12, and 0.04.ConclusionsThe Persona Approach Twente is applicable for mixed types of data and allows alignment of this user-centered design method to the iterative approach of eHealth development. A variety of characteristics can be used that stretches beyond (standardized) medical and demographic measurements. Challenges lie in data quality and fitness for (quantitative) clustering.

Keratoconjunctivitis sicca in dogs under primary veterinary care in the UK: an epidemiological study.

To estimate the frequency and breed-related risk factors for keratoconjunctivitis sicca (KCS) in dogs under UK primary veterinary care. Analysis of cohort electronic patient record data through the VetCompass Programme. Risk factor analysis used multivariable logistic regression. There were 1456 KCS cases overall from 363,898 dogs [prevalence 0.40%, 95% confidence interval (CI) 0.38-0.42] and 430 incident cases during 2013 (1-year incidence risk 0.12%, 95% CI 0.11-0.13). Compared with crossbreds, breeds with the highest odds ratio (aOR) for KCS included American cocker spaniel (aOR 52.33: 95% CI 30.65-89.37), English bulldog (aOR 37.95: 95% CI 26.54-54.28), pug (aOR 22.09: 95% CI 15.15-32.2) and Lhasa apso (aOR 21.58: 95% CI 16.29-28.57). Conversely, Labrador retrievers (aOR 0.23: 95% CI 0.1-0.52) and border collie (aOR 0.30: 95% CI 0.11-0.82) had reduced odds. Brachycephalic dogs had 3.63 (95% CI 3.24-4.07) times odds compared to mesocephalics. Spaniels had 3.03 (95% CI 2.69-3.40) times odds compared to non-spaniels. Dogs weighing at or above the mean bodyweight for breed/sex had 1.25 (95% CI 1.12-1.39) times odds compared to body weights below. Advancing age was strongly associated with increased odds. Quantitative tear tests are recommended within yearly health examinations for breeds with evidence of predisposition to KCS and might also be considered in the future within eye testing for breeding in predisposed breeds. Breed predisposition to KCS suggests that breeding strategies could aim to reduce extremes of facial conformation.

MO1045DEVELOPMENT OF A HOME CARE SUPPORT SYSTEM FOR PERITONEAL DIALYSIS PATIENTS I - USING DIGITAL WATERMARKING AND STEGANOGRAPHY*

Background and AimsAs it is necessary to refrain from going out due to the COVID-19 pandemic, a system that allows dialysis patients to be treated at a remote location or at home, i.e., a home care support system, is required. Information and communications technology (ICT) used for these purposes is widely applied in various medical fields. Using ICT has the advantage of allowing the sharing of patients’ electronic patient records (EPR) among medical staff, but increases the risk of copyright infringement and privacy leaks during archiving and transmission.We have developed a home care support system for peritoneal dialysis patients using information hiding technology consisting of both digital watermarking technology for copyright protection and steganography technology for communication security when treating patients at home using ICT. In addition, we evaluated the developed system.MethodThe system for sharing medical information was developed in the PHP programming language on a personal computer system using Microsoft Azure cloud services. Figure 1 shows an explanation of the digital watermarking technology and steganography technology used in the developed system.1. Digital watermarking technologyThe patient’s data, such as EPR data, facility name, etc., were hidden in the region of non-interest (RONI) of the patient’s chest CT image series and stored in a database.2. Steganography technologyWe call scene photos “cover pictures.” Medical information (CT images, etc.) was hidden in the cover picture. In this study, the cover picture containing the medical information was designated as a Stego image. A body CT image series (16-bit, 512 × 512, 100 slices) was used to verify the steganography technique. These CT images were compressed using 7-Zip and then saved in a folder, which was then embedded in the cover photo. The Stego image was then sent from the patient’s home to the medical institution via the home care support system.ResultsWe investigated the hash value, peak signal-to-noise ratio (PSNR), and structural similarity (SSIM) of the image series (Fig. 2). If the structure of the image or photograph was exactly the same, the SSIM shows a value of 1. If the PSNR is ≥ 40 dB, the image quality can be maintained without affecting diagnosis. If part of the ROI is changed during transmission, the hash value decoded from the received Stego image will be different from that before transmission. For Stego images containing watermarked or hidden CT images with 4000 words embedded, SSIM and PSNR were ≥ 0.99 dB and 65.3 dB, respectively.If the medical information was embedded in a low bit plane, such as a 1-bit or 2-bit plane, the radiologist could not identify the embedded information.When our technology was applied, there were no changes in the capacity of CT images or Stego images before and after embedding. Therefore, it was not possible to tell that medical information was embedded due to changes in capacity.ConclusionUsing ICT, we have built a home care support system that can conceal medical information by combining digital watermarking technology and steganography technology to ensure the copyright of images and to ensure privacy and secure transmission of EPR and CT images. Using the developed system, daily medical information of dialysis patients could be transmitted safely to the institute, and the medical staff could share the information safely. Both techniques can be applied to all digital image information, and is not just limited to CT images.