Denna studie undersöker användarupplevelsen av utbildning i journalsystemet Cosmic i Region Jönköpings län. Utbildning är av största vikt för personalens välmående samt för optimal användning av systemet. Syftet med undersökningen är att utforma en enkät för kontinuerligt förbättringsarbete inom området. Enkäten undersöker både kvantitativa och kvalitativa element hos ett urval som stratifierats utifrån användarnas yrkesroller. Enkäten skickades till deltagarnas respektive arbetsmejl, och svaren samlades in och bearbetades med hjälp av enkätprogrammet EsMaker. Ordinalskalor användes som mätverktyg i många av enkätens kvantitativa frågor, medan de kvalitativa frågorna analyserades med hjälp av The constant comparative method.  Studiens resultat visar en godtycklighet gentemot det material som finns samt med kollegor som instruktörer. Dock framkommer starka önskemål om organiserade utbildningar. Ett tydligt mönster är att användarna föredrar utbildningsmetoder som bygger på synkron kommunikation, samt att metoder som bygger på demonstration av programvaran är mer uppskattade än andra. Resultaten visar dock att dessa metoder bör kombineras med övningar för bästa effekt.  Slutsatser som undersökningen genererat är att kommande utbildningsinsatser bör innebära organiserade utbildningar på arbetsplatsen. Vidare forskning kopplat till Ställföreträdande lärande och Aktivitetsbaserat lärande skulle kunna användas för att optimera utbildningens resultat samt användarnas nöjdhet. En djupare analys av enkätresultatet med avseende på yrkesrollernas respektive behov skulle ytterligare kunna höja kvalitén och effektivisera utbildningarna. Studiens absolut viktigaste fynd är vikten av att chefer avsätter tid för sina medarbetare att ta del av de utbildningsmöjligheter som finns. Detta är kärnan i allt, för utan tid till utbildning spelar utbildningsmaterialets kvalitet ingen som helst roll. / This study investigates the end-user experience of education in Cosmic, a system for electronic health records, in Region Jönköping County. Training is of paramount importance for the well-being of the staff and for optimal use of the system. The purpose of the survey is to design a questionnaire that can be used for continuous improvement of the end-user training in the county.  The survey examines both quantitative and qualitative elements of a sample that is stratified based on the end‑users' professions. The survey was sent to the participants' work emails, and the responses were collected and processed using the EsMaker survey program. Ordinal scales were used as a measurement tool in many of the survey's quantitative questions, while the qualitative questions were analyzed using The constant comparative method.  The results of the study show an arbitrary attitude towards the available training material as well as towards colleagues as instructors. However, there are strong desires for organized training. A clear pattern is that users prefer training methods based on synchronous communication, as well as methods based on demonstration of the software. However, the results show that these methods should be combined with individual tasks for the best effect.  Conclusions generated by the survey are that future training efforts should involve organized training at the workplace. Further research linked to vicarious modeling and enactive learning could be used to optimize the results of the education as well as end-user satisfaction. A deeper analysis of the survey results regarding the respective needs of the professional roles could further increase the quality and streamline the education. The study's most important finding is the importance of managers to dedicate time for their employees to use the training opportunities available. This is the essence of everything, because without time for training, the quality of the educational material does not matter whatsoever. / <p>Examensarbete i vårdadministration, YH-utbildning: 20 Yh-poäng.</p>

Disease subtyping has been a critical aim of precision and personalized medicine. With the potential to improve patient outcomes, unsupervised and semi-supervised methods for determining phenotypes of subtypes have emerged with a recent focus on matrix and tensor factorization. However, interpretability of proposed models is debatable. Principal component analysis (PCA), a traditional method of dimensionality reduction, does not impose non-negativity constraints. Thus coefficients of the principal components are, in cases, difficult to translate to real physical units. Non-negative matrix factorization (NMF) constrains the factorization to positive numbers such that representative types resulting from the factorization are additive. Archetypal analysis (AA) extends this idea and seeks to identify pure types, archetypes, at the extremes of the data from which all other data can be expressed as a convex combination, or by proportion, of the archetypes. Using AA, this study sought to evaluate the sufficiency of AKI staging criteria through unsupervised subtyping. Archetype analysis failed to find a direct 1:1 mapping of archetypes to physician staging and also did not provide additional insight into patient outcomes. Several factors of the analysis such as quality of the data source and the difficulty in selecting features contributed to the outcome. Additionally, after performing feature selection with lasso across data subsets, it was determined that current staging criteria is sufficient to determine patient phenotype with serum creatinine at time of diagnosis to be a necessary factor.

In recent years the number of electronic healthcare records (EHRs)has increased rapidly. EHR represents a systematized collection of patient health information in a digital format. EHR systems maintain diagnoses, medications, procedures, and lab tests associated with the patients at each time they visit the hospital or care center. Since the information is available into multiple visits to hospitals or care centers, the EHR can be used to increasing quality care. This is especially useful when working with chronic diseases because they tend to evolve. There have been many deep learning methods that make use of these EHRs to solve different prediction tasks. Transformers have shown impressive results in many sequence-to-sequence tasks within natural language processing. This paper will mainly focus on using transformers, explicitly using a sequence of visits to do prediction tasks. The model presented in this paper is called CondBEHRT. Compared to previous state-of-art models, CondBEHRT will focus on using as much available data as possible to understand the patient’s trajectory. Based on all patients, the model will learn the medical ontology between diagnoses, medications, and procedures. The results show that the inferred medical ontology that has been learned can simulate reality quite well. Having the medical ontology also gives insights about the explainability of model decisions. We also compare the proposed model with the state-of-the-art methods using two different use cases; predicting the given codes in the next visit and predicting if the patient will be readmitted within 30 days.

Electronic health records (EHRs) are used in Sweden's healthcare systems to store patients' medical information. Patients in Sweden have the right to access and read their health records. Unfortunately, the language used in EHRs is very complex and presents a challenge for readers who lack medical knowledge. Simplifying the language used in EHRs could facilitate the transfer of information between medical staff and patients. This project investigates the possibility of generating Swedish medical synonyms automatically. These synonyms are intended to be used in future systems for lexical simplification that can enhance the readability of Swedish EHRs and simplify medical terminology. Current publicly available Swedish corpora that provide synonyms for medical terminology are insufficient in size to be utilized in a system for lexical simplification. To overcome the obstacle of insufficient corpora, machine learning models are trained to generate synonyms and terms that convey medical concepts in a more understandable way. With the purpose of establishing a foundation for analyzing complex medical terms, a simple mechanism for Complex Word Identification (CWI) is implemented. The mechanism relies on matching strings and substrings from a pre-existing corpus containing hand-curated medical terms in Swedish. To find a suitable strategy for generating medical synonyms automatically, seven different machine learning models are queried for synonym suggestions for 50 complex sample terms. To explore the effect of different input data, we trained our models on different datasets with varying sizes. Three of the seven models are based on BERT and four of them are based on Word2Vec. For each model, results for the 50 complex sample terms are generated and raters with medical knowledge are asked to assess whether the automatically generated suggestions could be considered synonyms. The results vary between the different models and seem to be connected to the amount and quality of the data they have been trained on. Furthermore, the raters involved in judging the synonyms exhibit great disagreement, revealing the complexity and subjectivity of the task to find suitable and widely accepted medical synonyms. The method and models applied in this project do not succeed in creating a stable source of suitable synonyms. The chosen BERT approach based on Masked Language Modelling cannot reliably generate suitable synonyms due to the limitation of generating one term per synonym suggestion only. The Word2Vec models demonstrate some weaknesses due to the lack of context consideration. Despite the fact that the current performance of our models in generating automatic synonym suggestions is not entirely satisfactory, we have observed a promising number of accurate suggestions. This gives us reason to believe that with enhanced training and a larger amount of input data consisting of Swedish medical text, the models could be improved and eventually effectively applied.

Recent advancements in machine learning and the widespread adoption of electronic healthrecords have enabled breakthroughs for several predictive modelling tasks in health care. One such task that has seen considerable improvements brought by deep neural networks is length of stay (LOS) prediction, in which research has mainly focused on adult patients in the intensive care unit. This thesis uses multivariate time series extracted from the publicly available Medical Information Mart for Intensive Care III database to explore the potential of deep learning for classifying the remaining LOS of newborns in the neonatal intensive care unit (NICU) at each hour of the stay. To investigate this, this thesis describes experiments conducted with various deep learning models, including long short-term memory cells, gated recurrentunits, fully-convolutional networks and several composite networks. This work demonstrates that modelling the remaining LOS of newborns in the NICU as a multivariate time series classification problem naturally facilitates repeated predictions over time as the stay progresses and enables advanced deep learning models to outperform a multinomial logistic regression baseline trained on hand-crafted features. Moreover, it shows the importance of the newborn’s gestational age and binary masks indicating missing values as variables for predicting the remaining LOS. / Framstegen inom maskininlärning och det utbredda införandet av elektroniska hälsoregister har möjliggjort genombrott för flera prediktiva modelleringsuppgifter inom sjukvården. En sådan uppgift som har sett betydande förbättringar förknippade med djupa neurala nätverk är förutsägelsens av vistelsetid på sjukhus, men forskningen har främst inriktats på vuxna patienter i intensivvården. Den här avhandlingen använder multivariata tidsserier extraherade från den offentligt tillgängliga databasen Medical Information Mart for Intensive Care III för att undersöka potentialen för djup inlärning att klassificera återstående vistelsetid för nyfödda i den neonatala intensivvårdsavdelningen (neonatal-IVA) vid varje timme av vistelsen. Denna avhandling beskriver experiment genomförda med olika djupinlärningsmodeller, inklusive longshort-term memory, gated recurrent units, fully-convolutional networks och flera sammansatta nätverk. Detta arbete visar att modellering av återstående vistelsetid för nyfödda i neonatal-IVA som ett multivariat tidsserieklassificeringsproblem på ett naturligt sätt underlättar upprepade förutsägelser över tid och gör det möjligt för avancerade djupa inlärningsmodeller att överträffaen multinomial logistisk regressionsbaslinje tränad på handgjorda funktioner. Dessutom visar det vikten av den nyfödda graviditetsåldern och binära masker som indikerar saknade värden som variabler för att förutsäga den återstående vistelsetiden.

No description available.

This research explores the role of blockchain technology in advanced information sharing systems with the applications of energy systems and healthcare. Essay 1 proposes a blockchain application to improve resilience in smart grids by addressing cyber security and peer-to-peer trading potentials. The results show that blockchain-based smart contracts are positively related to smart grid resilience. The findings also show that blockchain-based smart contracts significantly contribute to zero trust cybersecurity, which results in a better capability to mitigate cyber-attacks. Essay 2 proposes a blockchain application to improve electronic health record (EHR) systems by increasing patient's empowerment. Confirmatory factor analysis is used for the validity and reliability tests of the model. The results show that blockchain-based information systems can empower patients by providing the perception of control over their health records. The usage of blockchain technology motivates patients to share information with healthcare provider systems and has the advantage of reducing healthcare costs and improving diagnosis management. Essay 3 contributes to the existing literature by using a multimethod approach to propose three new mediators for blockchain-based healthcare information systems: digital health care, healthcare improvement, and peer-to-peer trade capability. Based on the findings from the text analysis, we propose a research model drawing upon stimulus-organism-response theory. Through three experimental studies, we test the research model to explain the patient's willingness to share their health records with others, including researchers. A post hoc analysis is conducted to segment patients and predict their behavior using four machine learning algorithms. The finding was that merely having peer-to-peer trade capability by ignoring healthcare improvement does not necessarily incentivize patients to share their medical reports.

Sepsis is the leading cause of death among critically ill patients and the 10th most common cause of death overall in the United States. The mortality rates increase with severity of the disease, ranging from 15% for sepsis to 60% for septic shock. Patient with sepsis can present varied clinical symptoms depending on the personal predisposition, causal microorganism, organ system involved, and disease severity. To facilitate sepsis diagnosis, the first sepsis consensus definitions was published in 1991 and then updated in 2001. Early recognition of a sepsis patient followed with timely and appropriate treatment and management strategies have been shown to significantly reduce sepsis-related mortality, and allows care to be provided at lower costs. Despite the rapid progress in the knowledge of pathophysiological mechanisms of sepsis and its treatment in the last two decades, identifying patient with sepsis and therapeutic approaches to sepsis and its complications remains challenging to critical care clinicians. Hence, the objectives of this thesis were to 1) evaluate the test characteristics of the two sepsis consensus definitions and delineate the differences in patient profile among patients meeting or not meeting sepsis definitions; 2) determine the relationship between the changes in several physiological parameters before sepsis onset and sepsis, and to determine whether these parameters could be used to identify sepsis in critically ill adults; 3) evaluate the effect of corticosteroids therapy on patient mortality. Data used in this thesis were prospectively collected from an electronic medical record system for all the adult patients admitted into the seven critical care units (ICUs) in a tertiary medical center. Besides analyzing data at the ICU stay level, we investigated patient information in various time frames, including 24-hour, 12-hour, and 6-hour time windows. In the first study of this thesis, the 1991 sepsis definition was found to have a high sensitivity of 94.6%, but a low specificity of 61.0%. The 2001 sepsis definition had a slightly increased sensitivity but a decreased specificity, which was 96.9% and 58.3%, respectively. The areas under the ROC curve for the two consensus definitions were similar, but less than optimal. The sensitivity and area under the ROC curve of both definitions were lower at the 24-hour time window level than those of the unit stay level, though the specificity increased slightly. At the time window level, the 1991 definitions performed slightly better than the 2001 definition. In the second study, minimum systolic blood pressure performed the best, followed by maximum respiratory rate in discriminating sepsis patients from SIRS patients. Maximum heart rate and maximum respiratory rate can differentiate sepsis patients from non-SIRS patients fairly well. The area under ROC of the combination of five physiological parameters was 0.74 and 0.90 for comparing sepsis to non-infectious SIRS patients and comparing sepsis to non-SIRS patients, respectively. Parameters typically performed better in 24-hour windows compared to 6-hour or 12-hour windows. In the third study, significantly increased hospital mortality and ICU mortality were observed in the group treated with low-dose corticosteroids than the control group based on the propensity score matched comparisons, and multivariate logistic regression analyses after adjustment for propensity score alone, covariates, or propensity score (in deciles) and covariates. This thesis advances the existing knowledge by systemically evaluating the test characteristics for the 1991 and 2001 sepsis consensus definitions, delineating physiological signs and symptoms of deterioration in the preceding 24 hours prior to sepsis onset, assessing the prediction performances of single or combined physiological parameters, and examining the use of corticosteroids treatment and survival among septic shock patients. In addition, this thesis sets an innovative example on how to use data from electronic medical records as these surveillance systems are becoming increasingly popular. The results of these studies suggest that a more parsimonious set of definitional criteria for sepsis diagnosis are needed to improve sepsis case identification. In addition, continuously monitored physiological parameters could help to identify patients who show signs of deterioration prior to developing sepsis. Last but not least, caution should be used when considering a recommendation on the use of low dose corticosteroids in clinical practice guidelines for the management of sepsis.

Healthcare is a complex system under great pressure for meeting the patients’ needs. Implementing technology at inpatient wards might possibly support healthcare professionals and improve quality of care. However, these technologies might come with issues and the system might not be used as intended. This master thesis project investigates how healthcare professionals communicate at an inpatient ward and how this might be affected by implementing a Mobile Healthcare Solution (MHS). Further, it sought to question why healthcare professions might, or might not, use the MHS as a support of their daily work and what some reasons for this might be. Research methods were of qualitative approach. Field studies were performed at an inpatient ward and further, two healthcare professionals were interviewed. Grounded Theory (GT) was chosen as a method to process the data and obtain understanding for communication at the inpatient ward. The results showed that healthcare professionals communicate verbally, written and by reading, using different tools. The most prominent ways of communication were verbally, where it was common to report or discuss about a patient. The means for communication did not get drastically affected by implementing the MHS and reasons for this were of social, technical and organizational types. Some reasons for not using the MHS were habits and due to healthcare professionals perceiving the MHS as more time consuming than manual handling. However, a specific investigation of whether this might affect the usage of the MHS is yet needed

Colorectal cancer (CRC) is a health challenge worldwide and early detection of the disease is crucial to improve patient prognosis. It is common for the first contact with care to occur in primary care centers where general practitioners often face the challenge of distinguishing CRC from other diseases with similar symptoms. In this master thesis, patient records from primary care were used to create, optimize, and evaluate a machine learning model that classifies patients with CRC for early detection of the disease. The data used in the project included parts of electronic health records (EHRs) from both public (SLSO) and privately run (Capio and Praktikertjänst) primary care centers in the Stockholm region. The available dataset was cleaned and pre- processed, and then tested on four separate models. After selecting and optimizing the most promising model, LightGBM, a detailed evaluation of the model was performed. To simulate realistic clinical conditions, data from the three months prior to diagnosis were excluded from two of the datasets. The results were then compared with a baseline machine learning model that utilized ICD codes extracted from EHRs in primary care for early detection of CRC.The results showed that the final developed model had a generally good performance with an AUROC score of a maximum of 85.8%, which indicates very good ability to distinguish between the classes. The performance dropped when using the datasets with 3 months of data removed, but the ROC curves still showed a better ability than random classification to distinguish between the classes with a AUROC score of maximum 60,8%. The results also showed that the model developed in this master thesis outperforms the baseline model, which was based on ICD codes, from a performance perspective. For future development and before a possible clinical implementation, a larger data set should be used for training and testing. / Tjock- och ändtarmscancer, kolorektal cancer (KRC) är en hälsoutmaning över hela världen och tidig upptäckt av sjukdomen är avgörande för att förbättra patientens prognos. Det är vanligt att den första kontakten med vården inträffar på vårdcentraler där allmänläkare ofta står inför utmaningen att skilja KRC från andra sjukdomar med liknande symtom. I denna masteruppsats kommer patientjournaler från primärvården att användas för att skapa, optimera och utvärdera en maskininlärningsmodell som klassificerar patienter med KRC för tidig upptäckt av sjukdomen.De data som använts i projektet omfattade delar av elektroniska patientjournaler (EHR) från både offentliga (SLSO) och privatägda (Capio och Praktikertjänst) primärvårdscentraler i Stockholmsregionen. Den tillgängliga datamängden städades och förbehandlades, och testades sedan på fyra separata modeller. Efter att ha valt ut och optimerat den mest lovande modellen, LightGBM, utfördes en detaljerad utvärdering av modellen. För att simulera realistiska kliniska tillstånd utvärderades modellen på två datamängder där data från tre månader före diagnos uteslöts. Resultaten jämfördes sedan med en baslinjemodell som använde ICD-koder som hämtats från journalsystem inom primärvården för tidig upptäckt av CRC.Resultaten visade att den slutliga utvecklade modellen hade en generellt bra prestanda med en AUROC-poäng på högst 85,8%, vilket indikerar mycket god förmåga att skilja mellan klasserna. Prestandan sjönk vid användning av datasatserna med 3 månaders data borttagen, men ROC-kurvorna visade fortfarande en bättre förmåga än slumpmässig klassificering att skilja mellan klasserna med en AUROC-poäng på högst 60,8%. Resultaten visade också att den modell som utvecklats i denna masteruppsats överträffar baslinjemodellen, som baserades på ICD-koder, ur ett prestationsperspektiv. För framtida utveckling och före en eventuell klinisk implementation bör en större datamängd användas för träning och testning av modellen.