The prediction of hospital length of stay using unstructured data

Objective This study aimed to assess the performance improvement for machine learning-based hospital length of stay (LOS) predictions when clinical signs written in text are accounted for and compared to the traditional approach of solely considering structured information such as age, gender and major ICD diagnosis. Methods This study was an observational retrospective cohort study and analyzed patient stays admitted between 1 January to 24 September 2019. For each stay, a patient was admitted through the Emergency Department (ED) and stayed for more than two days in the subsequent service. LOS was predicted using two random forest models. The first included unstructured text extracted from electronic health records (EHRs). A word-embedding algorithm based on UMLS terminology with exact matching restricted to patient-centric affirmation sentences was used to assess the EHR data. The second model was primarily based on structured data in the form of diagnoses coded from the International Classification of Disease 10th Edition (ICD-10) and triage codes (CCMU/GEMSA classifications). Variables common to both models were: age, gender, zip/postal code, LOS in the ED, recent visit flag, assigned patient ward after the ED stay and short-term ED activity. Models were trained on 80% of data and performance was evaluated by accuracy on the remaining 20% test data. Results The model using unstructured data had a 75.0% accuracy compared to 74.1% for the model containing structured data. The two models produced a similar prediction in 86.6% of cases. In a secondary analysis restricted to intensive care patients, the accuracy of both models was also similar (76.3% vs 75.0%). Conclusions LOS prediction using unstructured data had similar accuracy to using structured data and can be considered of use to accurately model LOS.

Optimization of Hill Cipher Method for Encryption and Decryption of Prescription Drugs at Puskesmas Twano Jayapura City

A drug prescription is a written request from a doctor to a pharmacist that must be kept secret because it contains certain doses of drugs and types of drugs that cannot be known by just anyone, especially those who are not interested. From time to time technological advances have a rapid impact on all sectors, both private and government agencies, including the health sector. One form of service in the health sector that can utilize information technology is the manufacture of electronic drug prescriptions that can be sent via an application from a doctor to a pharmacist. The frequent misuse of prescription drugs by unauthorized persons, as well as errors by officers at the pharmacy in reading prescriptions can be fatal for the community, so a solution is needed to overcome this problem. This application is designed using the Hill Cipher Algorithm which is one of the classic types of algorithms in the field of cryptography, but to get the maximum level of security, the algorithm key will be modified using a postal code pattern as a matrix key. Broadly speaking, the Encryption Stage is the first starting from the plaintext which is the type of drug and drug dose, the second key matrix using a POS code pattern, the three plaintexts are converted into blocks, the fourth is arranged into a 2x2 matrix, the fifth is multiplied between the key and the sixth plaintext is multiplied into mod 26 to generate an encrypted ciphertext or recipe. The success rate of system functionality testing using the blackbox method is 100%

Identification of Tobacco-Related Cancer Diagnoses among Individuals with Psychiatric Disorders: A Population-Based Matched Cohort Study Using a Competing Risks Approach from British Columbia

Background: Individuals with psychiatric disorders (PD) have a high prevalence of tobacco use. Therefore, we assessed the hazard of receiving a tobacco-related (TR) cancer diagnosis among individuals with PD. Methods: Several population-based provincial databases were used to identify individuals in BC diagnosed with depression, schizophrenia, bipolar disorder, anxiety disorders, or multiple PD between 1990 and 2013. A primary population proxy comparison group (appendicitis) was also identified and matched to the psychiatric cohort based on age at cohort entry, gender, year of cohort entry, and postal code. We linked individuals in the cohort and comparison groups with the BC Cancer Registry. Using a competing risks approach, we estimated the effect of having a PD on the risk of receiving a TR cancer diagnosis, in light of the competing risk of mortality. Results: In total, 165,289 patients were included. Individuals with depression (HR = 0.81; p < 0.01; 95% CI: 0.73–0.91), anxiety disorders (HR = 0.84; p = 0.02; 95% CI: 0.73–0.97), or multiple PD (HR = 0.74; p < 0.01; 95% CI: 0.66–0.83) had a statistically significant lower risk of a TR cancer diagnosis compared to the comparison group. Individuals with schizophrenia (HR = 0.86; p = 0.40; 95% CI: 0.62–1.21) or bipolar disorder (HR = 0.58; p = 0.12; 95% CI: 0.29–1.14), however, showed no evidence of a statistically significant difference from the comparison group. Interpretation: We found that individuals with depression, anxiety disorders, or multiple PD diagnoses had a significantly reduced risk of receiving a tobacco-related cancer diagnosis. These results were unexpected and could be explained by individuals with a PD having barriers to a cancer diagnosis rather than a true decreased incidence.

Longitudinal SARS-CoV-2 RNA Wastewater Monitoring Across a Range of Scales Correlates with Total and Regional COVID-19 Burden in a Well-Defined Urban Population

Wastewater-based epidemiology (WBE) is an emerging surveillance tool that has been used to monitor the ongoing COVID-19 pandemic by tracking SARS-CoV-2 RNA shed into wastewater. WBE was performed to monitor the occurrence and spread of SARS-CoV-2 from three wastewater treatment plants (WWTP) and six neighborhoods in the city of Calgary, Canada (population 1.3 million). A total of 222 WWTP and 192 neighborhood samples were collected from June 2020 to May 2021, encompassing the end of the first-wave (June 2020), the second-wave (November end to December, 2020) and the third-wave of the COVID-19 pandemic (mid-April to May, 2021). Flow-weighted 24-hour composite samples were processed to extract RNA that was then analyzed for two SARS-CoV-2-specific regions of the nucleocapsid gene, N1 and N2, using reverse transcription-quantitative polymerase chain reaction (RT-qPCR). Using this approach SARS-CoV-2 RNA was detected in 98.06 percent (406/414) of wastewater samples. SARS-CoV-2 RNA abundance was compared to clinically diagnosed COVID-19 cases organized by the three-digit postal code of affected individuals primary residences, enabling correlation analysis at neighborhood, WWTP and city-wide scales. Strong correlations were observed between N1 and N2 gene signals in wastewater and new daily cases for WWTPs and neighborhoods. Similarly, when flow rates at Calgarys three WWTPs were used to normalize observed concentrations of SARS-CoV-2 RNA and combine them into a city-wide signal, this was strongly correlated with regionally diagnosed COVID-19 cases and clinical test percent positivity rate. Linked census data demonstrated disproportionate SARS-CoV-2 in wastewater from areas of the city with lower socioeconomic status and more racialized communities. WBE across a range of urban scales was demonstrated to be an effective mechanism of COVID-19 surveillance.

Epigenetic And Transcriptomic Alterations in Offspring Born To Women With Type 1 Diabetes (The EPICOM Study)

Background: Offspring born to women with pregestational type 1 diabetes (T1DM) are exposed to an intrauterine hyperglycemic milieu and has an increased risk of metabolic disease in later in life. In this present study we hypothesize that in utero exposure to T1DM alters offspring DNA methylation and gene expression, thereby altering their risk of future disease. Design: Follow-up study using data from the Epigenetic, Genetic and Environmental Effects on Growth, Metabolism and Cognitive Functions in Offspring of Women with Type 1 Diabetes (EPICOM) collected between 2012-2013.Setting: Exploratory sub study using data from the nationwide EPICOM study.Participants: Adolescent offspring born to women with T1DM (n=20) and controls (n=20) matched on age, sex and postal code. Main outcome measures: This study investigates DNA methylation using the 450K-Illumina Infinium assay® and RNA expression (RNA sequencing) of leucocytes from peripheral blood samples. Results: We identified 9 hypermethylated and 5 hypomethylated positions (p < 0.005, |DM-value| > 1). RNA expression profiling identified 38 up- and 1 down-regulated genes (p < 0.005, log2FC ≥ 0.3.). Functional enrichment analysis revealed enrichment in ontologies related to diabetes, carbohydrate and glucose metabolism, pathways including MAPK1/MAPK3 and MAPK family signaling and genes related to T1DM, obesity and atherosclerosis. Lastly, by integrating the DNA methylation and RNA expression data, we identified six genes where relevant methylation changes corresponded with RNA expression (CIITA, TPM1, PXN, ST8SIA1, LIPA, DAXX). Conclusions: Findings suggest the possibility for intrauterine hyperglycemia to impact later life methylation and gene expression, a profile that may be linked to the increased risk of metabolic disease.

Use of Antibiotics among Residents Living Close to Poultry or Goat Farms: A Nationwide Analysis in The Netherlands

Prior regional studies found a high risk of pneumonia for people living close to poultry and goat farms. This epidemiological study in the Netherlands used nationwide antibiotic prescription data as a proxy for pneumonia incidence to investigate whether residents of areas with poultry and goat farms use relatively more antibiotics compared to areas without such farms. We used prescription data on antibiotics most commonly prescribed to treat pneumonia in adults and livestock farming data, both with nationwide coverage. Antibiotic use was expressed as defined daily doses per (4-digit Postal Code (PC4) area)-(age group)-(gender)-(month) combination for the year 2015. We assessed the associations between antibiotic use and farm exposure using negative binomial regression. The amoxicillin, doxycycline, and co-amoxiclav use was significantly higher (5–10% difference in use) in PC4 areas with poultry farms present compared to areas without, even after adjusting for age, gender, smoking, socio-economic status, and goat farm presence. The adjusted models showed no associations between antibiotic use and goat farm presence. The variables included in this study could only partly explain the observed regional differences in antibiotic use. This was an ecological study that precludes inference about causal relations. Further research using individual-level data is recommended.

623. Exploring ‘Slicer Dicer’, an Extraction Tool in EPIC, for Clinical and Epidemiological Analysis

Background Electronic Health Record (EHR) implementation has created an unprecedented library of patient data. Data extraction tools provide an opportunity to retrieve clinico-epidemiological information on a wide scale. Slicer Dicer is a data exploration tool in the EPIC EHR that allows one to customize searches on large patient populations. This software contains a variety of models that present de-identified information from EPIC’s Caboodle database. We explored the applicability and potential utility of this tool utilizing the diagnosis of Lyme disease as an example. Methods The following steps outline an overview of data extraction utilizing ICD-10 codes around Lyme disease at our health system. Step 1-3: Denominator chosen as ‘All Patients’ over a 3-year period, ‘Slicing’ of the data by ‘Lyme disease, unspecified’ was applied to these results, and the ‘sliced’ data was categorized by year of diagnosis (Slide 1). Step 4: This data was further arranged by month of diagnosis for trend analysis (Slide 2). Step 5: Sub-diagnosis was applied for Lyme arthritis (Slide 3). Step 6: Further ‘slicing’ was/can be done by other variables, such as ‘Hospitalization,’ ‘Encounter Diagnosis,’ and ‘ED Diagnosis’ (Slide 4). Step 7-8: Output was ‘sliced’ by ‘Age’ (Slide 5) and ‘Postal Code’ (Slide 6). Slide 1. EPIC EHR screen capture showing 3-year period data Data shown here represents 'All patients' chosen as the denominator further sliced by 'Lyme disease, unspecified' and categorized by the year of diagnosis. Slide 2. EPIC EHR screen capture showing data further arranged by month of diagnosis Results Macro-level data of period prevalence on Lyme disease over 3 years (Slide 1), seasonal trends (Slide 2), specific sub-diagnosis (Slide 3), output by setting of diagnosis (Slide 4), and demographic information of our patient population (Slides 5, 6) was revealed by application of these parameters. Slide 3. EPIC EHR screen capture showing application of sub-diagnosis for Lyme arthritis Slide 4. EPIC EHR screen capture showing further slicing by multiple variables like hospitalization and diagnosis Slide 5. EPIC EHR screen capture showing slicing of data by demographic information (Age) Conclusion Slicer Dicer can provide a snapshot for preliminary data analysis prior to investing time and commitment to a project. The appeal of this tool is that it mines de-identified data and thus does not require initial IRB approval. This opens an avenue for potential full research projects based on the results obtained and helps generate preliminary hypotheses through analysis of healthcare. Slide 6. EPIC EHR screen capture showing slicing of data by demographic information (Postal Code) Disclosures All Authors: No reported disclosures

Peer review declaration

All papers published in this volume of Journal of Physics: Conference Series have been peer reviewed through processes administered by the Editors. Reviews were conducted by expert referees to the professional and scientific standards expected of a proceedings journal published by IOP Publishing. • Type of peer review: Double-blind • Conference submission management system: Konfrenzi • Number of submissions received: 37 • Number of submissions sent for review: 23 • Number of submissions accepted: 14 • Acceptance Rate (Number of Submissions Accepted / Number of Submissions Received X 100): 37.8% • Average number of reviews per paper: 2 • Total number of reviewers involved: 7 • Any additional info on review process: • Contact person for queries: Prof. Zaki Su’ud Email: [email protected] Dept. of Physics, Faculty Mathematics and Natural Sciences Institut Teknologi Bandung, Indonesia Address: Jl. Ganesa No 10 Bandung, West Java, Indonesia Postal Code: 40132 Tel : +62-22-2500834 Fax: +62-22-2506452

Leveraging open data to reconstruct the Singapore Housing Index and other building-level markers of socioeconomic status for health services research

Background Socioeconomic status (SES) is an important determinant of health, and SES data is an important confounder to control for in epidemiology and health services research. Individual level SES measures are cumbersome to collect and susceptible to biases, while area level SES measures may have insufficient granularity. The ‘Singapore Housing Index’ (SHI) is a validated, building level SES measure that bridges individual and area level measures. However, determination of the SHI has previously required periodic data purchase and manual parsing. In this study, we describe a means of SHI determination for public housing buildings with open government data, and validate this against the previous SHI determination method. Methods Government open data sources (e.g. data.gov.sg, Singapore Land Authority OneMAP API, Urban Redevelopment Authority API) were queried using custom Python scripts. Data on residential public housing block address and composition from the HDB Property Information dataset (data.gov.sg) was matched to postal code and geographical coordinates via OneMAP API calls. The SHI was calculated from open data, and compared to the original SHI dataset that was curated from non-open data sources in 2018. Results Ten thousand seventy-seven unique residential buildings were identified from open data. OneMAP API calls generated valid geographical coordinates for all (100%) buildings, and valid postal code for 10,012 (99.36%) buildings. There was an overlap of 10,011 buildings between the open dataset and the original SHI dataset. Intraclass correlation coefficient was 0.999 for the two sources of SHI, indicating almost perfect agreement. A Bland-Altman plot analysis identified a small number of outliers, and this revealed 5 properties that had an incorrect SHI assigned by the original dataset. Information on recently transacted property prices was also obtained for 8599 (85.3%) of buildings. Conclusion SHI, a useful tool for health services research, can be accurately reconstructed using open datasets at no cost. This method is a convenient means for future researchers to obtain updated building-level markers of socioeconomic status for policy and research.

Distance as explanatory factor for sexual health centre utilization: an urban population-based study in the Netherlands

Background The central sexual health centre (SHC) in the greater Rotterdam area in the Netherlands helps finding people unaware of their STI/HIV status. We aimed to determine a possible association between SHC utilization and travel distance in this urban and infrastructure-rich area. Insight in area-specific utilization helps adjust outreach policies to enhance STI testing. Methods The study population consists of all residents aged 15–45 years in the greater Rotterdam area (2015–17). We linked SHC consultation data from STI tested heterosexual clients to the population registry. The association between SHC utilization and distance was investigated by multilevel modelling, adjusting for sociodemographic and area-specific determinants. The data were also stratified by age (aged &lt; 25 years) and migratory background (non-Western), since SHC triage may affect their utilization. We used straight-line distance between postal code area centroid and SHC address as a proxy for travel distance. Results We found large area variation in SHC utilization (range: 1.13–48.76 per 1000 residents). Both individual- and area-level determinants determine utilization. Travel distance explained most area variation and was inversely associated with SHC utilization when adjusted for other sociodemographic and area-specific determinants [odds ratio (OR) per kilometre: 0.95; 95% confidence interval (CI): 0.93–0.96]. Similar results were obtained for residents &lt;25 years (OR: 0.95; 95% CI: 0.94–0.96), but not for non-Western residents (OR: 0.99; 95% CI: 0.99–1.00). Conclusions Living further away from a central SHC shows a distance decline effect in utilization. We recommend to enhance STI testing by offering STI testing services closer to the population.