A5057037167- Machine Learning in Healthcare
- Topic Modeling
- Biomedical Text Mining and Ontologies
- Sepsis Diagnosis and Treatment
- Data-Driven Disease Surveillance
- Research Data Management Practices
- Scientific Computing and Data Management
- Natural Language Processing Techniques
- Nosocomial Infections in ICU
- Text and Document Classification Technologies
- Distributed and Parallel Computing Systems
- Advanced Clustering Algorithms Research
- Chronic Disease Management Strategies
- Data Quality and Management
- Advanced Text Analysis Techniques
- Clostridium difficile and Clostridium perfringens research
- COVID-19 epidemiological studies
- Emergency and Acute Care Studies
- Microscopic Colitis
- Anomaly Detection Techniques and Applications
- COVID-19 Impact on Reproduction
- Clinical Reasoning and Diagnostic Skills
- SARS-CoV-2 and COVID-19 Research
- Algorithms and Data Compression
- COVID-19 and healthcare impacts
Karolinska Institutet
2021-2024
Osaka University
2023-2024
Karolinska University Hospital
2014-2022
Stockholm University
2012-2016
Kista Photonics Research Center
2013
Hospital-acquired infections pose a significant risk to patient health, while their surveillance is an additional workload for hospital staff. Our overall aim build system that reliably detects all records potentially include hospital-acquired infections. This reduce the burden of having staff manually check records. study focuses on application text classification using support vector machines and gradient tree boosting problem. Support have never been applied problem detecting in Swedish...
An understanding of differences in clinical phenotypes and outcomes COVID-19 compared with other respiratory viral infections is important to optimise the management patients plan healthcare. Herein we sought investigate such positive for SARS-CoV-2 influenza, syncytial virus (RSV) viruses. We performed a retrospective cohort study hospitalised adults children (≤15 years) who tested SARS-CoV-2, influenza A/B, RSV, rhinovirus, enterovirus, parainfluenza viruses, metapneumovirus, seasonal...
Abstract Sepsis is a leading cause of mortality and early identification improves survival. With increasing digitalization health care data automated sepsis prediction models hold promise to aid in prompt recognition. Most previous studies have focused on the intensive unit (ICU) setting. Yet only small proportion develops ICU there an apparent clinical benefit identify patients earlier disease trajectory. In this cohort 82,852 hospital admissions 8038 episodes classified according Sepsis-3...
The prevalence of healthcare-associated infections (HAI) stresses the need for automatic surveillance in order to follow effect preventive measures. A number detection systems have been set up several languages, but none is known Swedish hospitals. We plan a series infection type specific programs HAI electronic health records at university hospital. Also, we aim detecting patients entering hospital with from previous care, task that not often addressed. This first study aims urinary tract...
BackgroundSurveillance for healthcare-associated infections such as urinary tract (HA-UTI) is important directing resources and evaluating interventions. However, traditional surveillance methods are resource-intensive subject to bias.AimTo develop validate a fully automated algorithm HA-UTI using electronic health record (EHR) data.MethodsFive algorithms were developed EHR data from 2979 admissions at Karolinska University Hospital 2010 2011: (1) positive urine culture (UCx); (2) UCx + UTI...
Background Universal SARS-CoV-2 testing at hospital admission has been proposed to prevent nosocomial transmission. Aim To investigate positivity in patients tested with low clinical COVID-19 suspicion admission. Methods We characterised a retrospective cohort of admitted Karolinska University Hospital for by PCR from March September 2020, supplemented an in-depth chart review (16 March–12 April). compared rates and without Spearman’s rank correlation coefficient. used multivariable logistic...
We developed and validated a set of fully automated surveillance algorithms for healthcare-onset CDI using electronic health records. In validation data 750 manually annotated admissions, the algorithm based on International Classification Disease, Tenth Revision (ICD-10) code A04.7 had insufficient sensitivity. Algorithms microbiological test results with or without addition symptoms performed well.
Research must be reproducible to verifiable. Provenance, which describes how data was produced, is one of the metadata that can improve reproducibility. In this paper, we propose a method construct provenance research produced in high-performance computing (HPC) systems. Our high-level and user-perspective by integrating information available HPC systems, such as workload manager, with low-level about running programs' behavior captured an operating system kernel. The enables users systems...
Continuous surveillance for healthcare-associated infections such as central venous catheter-related bloodstream (CVC-BSI) is crucial prevention. However, traditional methods are resource-intensive and prone to bias. This study aimed develop validate fully-automated algorithms CVC-BSI. Two were developed using electronic health record data from 1000 admissions with a positive blood culture (BCx) at Karolinska University Hospital 2017: (1) Combining microbiological findings in BCx CVC...
Many natural language processing applications rely on the availability of domain-specific terminologies containing synonyms.To that end, semi-automatic methods for extracting additional synonyms a given concept from corpora are useful, especially in low-resource domains and noisy genres such as clinical text, where nonstandard use misspellings prevalent.In this study, prototype embeddings based seed words were used to create representations (i) specific urinary tract infection (UTI) symptoms...
Background: Healthcare-associated infection (HAI) surveillance is essential for most prevention programs and continuous epidemiological data can be used to inform healthcare personal, allocate resources, evaluate interventions prevent HAIs. Many HAI systems today are based on time-consuming resource-intensive manual reviews of patient records. The objective HAI-proactive, a Swedish triple-helix innovation project, develop implement fully automated system electronic health record data....