Pharamacovigilance relates to activities involving drug safety monitoring in the post-marketing phase of the drug development life-cycle. Despite rigorous trials and experiments that drugs undergo before they are available in the market, they can still cause previously unobserved side-effects (also known as adverse events) due to drug–drug interaction, genetic, physiological or demographic reasons. The Uppsala Monitoring Centre (UMC) is the custodian of the global reporting system, VigiBase, for adverse drug reactions in collaboration with the World Health Organization (WHO). VigiBase houses over 20 million case reports of suspected adverse drug reactions from all around the world. However, not all case reports that the UMC receives pertains to adverse reactions that are novel in the safety profile of the drugs. In fact, many of the reported reactions found in the database are known adverse events for the reported drugs. With more than 3 million potential associations between all possible drugs and all possible adverse events present in the database, identifying associations that are likely to represent previously unknown safety concerns requires powerful statistical methods and knowledge of the known safety profiles of the drugs. Therefore, there is a need for a knowledge base with mappings of drugs to their known adverse reactions. To-date, such a knowledge base does not exist. The purpose of this thesis is to develop a deep-learning model that learns to extract adverse reactions from product labels — regulatory documents providing the current state of knowledge of the safety profile of a given product — and map them to a standardized terminology with high precision. To achieve this, I propose a two-phase algorithm, with a first scanning phase aimed at finding regions of the text representing adverse reactions, and a second mapping phase aiming at normalizing the detected text fragments into Medical Dictionary for Regulatory Activities (MedDRA) terms, the terminology used at the UMC to represent adverse reactions. A previous dictionary-based algorithm developed at the UMC achieved a scanning F1 of 0.42 (0.31 precision, 0.66 recall) and mapping macro-averaged F1 of 0.43 (0.39 macro-averaged precision, 0.64 macro-averaged recall). State-of-the-art methods achieve F1 above 0.8 and above 0.7 for the scanning and mapping problems respectively. To develop algorithms for adverse reaction extraction, I use the 2019 ADE Evaluation Challenge data, a dataset made by the FDA with 100 product labels annotated for adverse events and their mappings to MedDRA. This thesis explores three architectures for the scanning problem: 1) a Bidirectional Long Short-Term Memory (BiLSTM) encoder followed by a softmax classifier, 2) a BiLSTM encoder with Conditional Random Field (CRF) classifier and finally, 3) a BiLSTM encoder with CRF classifier with Embeddings from Language Model (ELMo) embeddings. For the mapping problem, I explore Information Retrieval techniques using the search engines whoosh and Solr, as well as a Learning to Rank algorithm. The BiLSTM encoder with CRF gave the highest performance on finding the adverse events in the texts, with an F1 of 0.67 (0.75 precision, 0.61 recall), representing a 0.06 absolute increase in F1 over the simpler BiLSTM encoder with softmax. Using the ELMo embeddings was proven detrimental and lowered the F1 to 0.62. Error analysis revealed the adopted Inside, Beginning, Outside (IOB2) labelling scheme to be poorly adapted for denoting discontinuous and compound spans while introducing ambiguity in the training data. Based on the gold standard annotated mappings, I also evaluated the whoosh and Solr search engines, with and without Learning to Rank. The best performing search engine on this data was Solr, with a macro-averaged F1 of 0.49 compared to the macro-averaged F1 of 0.47 for the whoosh search engine. Adding a Learning to Rank algorithm on top of each engine did not improve mapping performance, as both macro-averaged F1 dropped by over 0.1 when using the re-ranking approach. Finally, the best performing scanning and mapping algorithms beat the aforementioned dictionary-based baseline F1 by 0.25 in the scanning phase and 0.06 in the mapping phase. A large source of error for the Solr search engine came from tokenisation issues, which had a detrimental impact on the performance of the entire pipeline. In conclusion, modern Natural Language Processing (NLP) techniques can significantly improve the performance of adverse event detection from free-formtext compared to dictionary-based approaches, especially in cases where context is important. / Farmakovigilans berör de aktiviteter som förbättrar förståelsen av biverkningar av läkemedel. Trots de stränga prövningar som behövs för läkemedelsutvecklingen finns ändå en del biverkningar som är okända p.g.a. genetik, fysiologiska eller demografiska faktorer. Uppsala Monitoring Centre (UMC), i samarbete med World Health Organization (WHO) är vårdnadshavare till den globala databasen av rapporter på medicinska biverkningar, VigiBase. VigiBase innehåller över 20 miljoner misstänkta rapporter från hela världen. Dock, en andel av dessa rapporter beskriver biverkningar som är redan kända. Egentligen finns det över 3 miljoner potentiella samband mellan alla läkemedel och biverkningar i databasen. Att hitta den riktiga och okända biverkningar behövs kraftfulla statistiska metoder samt kunskap om det kända säkerhetsprofil av läkemedlet. Det finns ett behöv för ett databas som kartlägger läkemedel med alla kända biverkningar men, inget sådant databas finns idag. Syftet med detta examensarbete är att utveckla en djup-lärandemodell som kan läsa av texter på läkemedels etiketter — tillsynsdokument som beskriver säkerhetsprofil av läkemedel — och kartlägga dem till ett standardiserat terminologi med hög precision. Problemet kan brytas in i två fas, den första scanning och den andra mapping. Scanning handlar om att kartlägga position av text-fragmentet i etiketter. Mapping handlar om att kartlägga de detekterade text-fragmentet till Medical Dictionary for Regulatory Activities (MedDRA), den terminologi som används i UMC för biverkningar. Tidigare försök, s.k. dictionary-based approach på UMC uppnådde scanning F1 i 0,42 (0,31 precision; 0,64 recall) och mapping macro-averaged F1 i 0,43 (0,39 macro-averaged precision; 0,64 macro-averaged recall). De bästa systemen (s.k. state-of-the-art) uppnådde scanning F1 över 0,8 och 0,7 för den scanning respektive mapping problemet. Jag använder den 2019 ADE Evaluation Challenge dataset att utveckla algoritmerna i projektet. Detta dataset innehåller 100 läkemedels etiketter annoterad med biverkningar och deras kartläggning i MedDRA. Denna avhandling utforskar tre arkitekturer till scanning problemet: 1) Bidirectional Long Short-Term Memory (BiLSTM) och softmax för klassificering, 2) BiLSTM med Conditional Random Field (CRF) klassificering och, till sist, 3) BiLSTM med CRF klassificering och Embeddings from Language Model (ELMo) embeddings. Med avseende till mapping problematiken utforskar jag metoder inom Information Retrieval genom användning av sökmotorerna whoosh och Solr. För att förbättra prestandan i mapping utforskar jag Learning to Rank metoder. BiLSTM med CRF presterade bäst inom scanning problematiken med F1 i 0,67 (0,75 precision; 0,61 recall) som är ett 0,06 absolut ökning över den BiLSTM encoder med softmax klassificering. Med ELMo försämrade F1 till 0,62. Analys av felet visade att Inside, Beginning, Outside (IOB2) märkning som jag har valt att använda passar inte till att beteckna diskontinuerliga och sammansatta spans, och tillför betydande osäkerhet i träningsdata. Med avseende till mapping problematiken har jag kollat på sökmotorn Solr och whoosh, med, och utan Learning to Rank. Solr visade sig som den bäst presterande sökmotorn med macro-averaged F1 i 0,49 jämfört med whoosh som visade macro-averaged F1 i 0,47. Learning to Rank algoritmerna försämrade F1 med över 0,1 för båda sökmotorer. Den bäst presterande scanning och mapping algoritmer slog den baseline systemets F1 med 0,25 i scanning faset, och 0,06 i mapping fasen. Ett stor källa av fel för den Solr sökmotorn har kommit från tokeniserings-fel, som hade en försämringseffekt i prestanda genom hela pipelinen. I slutsats, moderna Natural Language Processing (NLP) tekniker kan kraftigt öka prestanda inom detektering av biverkningar från etiketter och texter, jämfört med gamla dictionary metoder, särskilt när kontexten är viktigt.

In active sensing, sensory organs are actively controlled by the motor system to optimize stimuli acquisition. Coupling behaviour of animal models with electrophysiological signals could give us important insights in the workings of this active sensing in health and disease. In this work, a contribution is made towards this aim. A modern machine learning method, called DeepLabCut, is evaluated and optimized for the use of tracking whiskers from videos of head-fixed, free whisking mice from two groups. Novel to the approach is the longer observation of the mice, without markers or trimming. This tracking is then used to investigate the natural statistics of whisking in healthy and Parkinsons Disease (PD) modeled animals. Two groups of mice (6 healthy versus 4 PD) were filmed for 5 minutes on 1-3 days per animal during head-fixed free air whisking, resulting in 21 filmed days in total. 6 Whiskers were tracked with 4 points per whisker using DeepLabCut and as few as 105 manually labeled images.A tracking accuracy of <3 pixels training error and <15 pixels test error was achieved. Qualitatively, tracking errors occurred both along and perpendicular to the whisker, but the perpendicular error reduced after 30 Hz low-pass filtering. The observed whisking was non-stationary and predominantly slow (<10 Hz). Whisking was also synchronized in time, as the cross-spectra were significantly higher than the random shuffled cross-values (pKS<0.05), but this synchronicity was lost on the right side in lesion animals with whisking frequencies between 15 and 25 Hz. In this PD-modeled group, a difference between left and right side co-synchronization was found (JSD=0.067, pKS<0.05) for higher frequencies (15-25 Hz). In total, control animals where whisking 34.1% of the time versus 35.6% for the lesion animals. Upon examining these active bouts, it was found that lesion animals have relatively more long bouts of co-activation than healthy controls across all subgroups (pKS<0.001). Furthermore, in lesion animals the activity distributions are laterally different (LL vs RR, JSD=0.178, pKS<0.001), but this cannot be concluded for healthy controls. Thus, it can be concluded that DeepLabCut is a promising method to speed up whisker tracking. Future improvements could be made in sharpness and contrast of the tips of the whiskers; incorporating information on the relation of whisker position in time; extending the tracking to 3D; and evaluating tracking by means of biomechanical models. These improvements will enable observation in a more realistic environment, and for a longer time. Finally, coupling of whisker position data to electrophysiological recordings will hopefully result in new insights in the relation between sensorimotor behaviour and global brain dynamics. / Vid 'active sensing' styrs sinnesorgan aktivt av motorsystemet för att optimera insamling av stimuli. Koppling av djurmodeller med elektrofysiologiska signaler skulle kunna ge oss viktig insikt i funktionen av denna 'active sensing' i hälsa och sjukdom. I detta arbete görs ett bidrag till detta mål. En modern maskininlärningsmetod, kallad DeepLabCut, utvärderades och optimerades för användning av spårning av morrhår från videor där möss i två grupper hade huvudet fixerat men morrhåren kunde rör sig fritt. Nytt för denna metod är den längre observationen av mössen utan märkning eller hårklippning. Denna spårning användes sedan för att undersöka den naturliga statistiken över morrhårens rörelse i kontrollgruppen och gruppen med Parkinsons sjukdom (PD). Två grupper av möss (6 kontroll mot 4 PD) filmades i 5 minuter på 1-3 dagar per djur där huvudet var fixerat men morrhåren kunde röra sig fritt, vilket resulterade i totalt 21 filmade dagar. Sex morrhår spårades med 4 punkter per hår med hjälp av DeepLabCut med så få som 105 manuellt märkta bilder. En spårningsnoggrannhet med träningsfel på <3 pixlar och testfel på <15 pixlar uppnåddes. Kvalitativt inträffade spårningsfel både längs med och vinkelrätt mot morrhåret, men det vinkelräta felet minskade efter 30 Hz lågpassfiltrering. Den observerade rörelsen av morrhåret var föränderlig och övervägande långsam (<10 Hz). Rörelsen av morrhåren synkroniserades också i tid, eftersom korsspektra var signifikant högre än de blandade korsvärdena (pKS<0.05), men denna synkronitet förlorades på höger sida i PD djur med frekvenser mellan 15 och 25 Hz. I denna PD grupp hittades en skillnad mellan vänster och höger sida i samsynkronisering (JSD=0.067, pKS<0.05) för högre frekvenser (15-25 Hz). Kontrollgruppen rörde på morrhåren i totalt 34,1% av tiden mot 35,6% för PD gruppen. Inom denna aktiva period visade det sig att PD djur hade relativt mer långa perioder av samaktivering än kontrolldjur i alla delgrupper (pKS<0.001). Dessutom var aktivitetsfördelning laterallt olika i PD gruppen (LL vs RR, JSD=0,178,pKS<0.001), men detta kunde inte påvisas i kontrollgruppen. Således kan man dra slutsatsen att DeepLabCut är en lovande metod för att påskynda spårning av rörelse av morrhår. Framtida förbättringar skulle kunna göras med förbättrad skärpa och kontrast av morrhårens ändar; innehålla information om morrhårens postion relativt till varandra över tid; utvidga spårningen till 3D; och utvärdera spårning med hjälp av biomekaniska modeller. Dessa förbättringar möjliggör observation i en mer realistisk miljö och under en längre tid. Slutligen kommer koppling av morrhårspositionens data till elektrofysiologiska inspelningar förhoppningsvis att resultera i ny insikt i förhållandet mellan sensorimotoriskt beteende och global hjärndynamik.

This master thesis investigates the possibility of using machine learning methods to predict patient length of stay at the time of admission to a clinical ward from the emergency department. The main aim of this thesis is to provide a comparative analysis of different algorithms and to suggest a suitable model that can be used in a hospital prediction software. The results show that it is possible to achieve a balanced accuracy of 0.72 at the time of admission and of 0.75 at a later stage in the process. The suggested algorithm was Random Forest which combines good accuracy with effective training time, making it suitable for on-line use in a hospital. The study shows that there is a clear potential for the use of machine learning methods for predicting length of stay, but that further improvements have to be made before adaption into the healthcare. / Detta masterexamensarbete utforskar möjligheten att använda maskin-inlärning för att förutspå vårdtiden för en patient då denne skrivs in på en vårdavdelning från akutvårds-avdelningen vid ett sjukhus. Huvudmålet för arbetet är att tillhandahålla en jämförelse av olika maskininlärnings-algoritmer  och föreslå en algoritm som är lämplig att integrera i en mjukvara på sjukhuset. Resultaten visar att det är möjligt att nå en balanced accuracy på 0.72 vid inskrivningstillfället samt 0.75 vid en senare tidpunkt i vårdprocessen. Den föreslagna algoritmen var Random Forest som kombinerade bra prestanda med effektiv träningstid, något som gör den lämplig för att köras på sjukhuset. Projektet visar att det finns en tydlig potential för att använda maskininlärning för att prediktera vårdtid men att förbättringar krävs innan det kan nå hela vägen in i sjukhuset.

Existing method for the antibiotic susceptibility test (AST) requires 2-3 days to produce the result. For patients with urinary tract infection (UTI), this lengthy process is critical because they must bear the pain during this period. A newly proposed idea to shorten the time is by using phenotypic antibiotic susceptibility test (AST) method on single bacteria.The first step to achieve the goal is to design and fabricate a platform to capture single bacteria and remove unwanted particles. An integrated micro and nanomembrane structure fulfill the requirement. The micro membrane structure is used to capture intended particles and block larger particles, while nanomembrane is used to remove unwanted nanoparticles. Micro-membrane structure is designed as an array of the microwell, and each microwell is expected to capture the bacteria. The thickness of microwell must be comparable to the bacteria dimension to make the observation of single bacteria be possible. Otherwise, there will be a stack of bacteria and only the bacteria at the top can be observed.The nanoporous membrane is created using aluminum foil through the anodization process and microwell structure is created using OSTE polymer through photolithography. Both materials are inexpensive and the fabrication process is fairly simple. In the final result, the nanoporous membrane is successfully integrated with microwell array. Each microwell has diameter 200 µm and thickness 100 µm. The nanoporous membrane has pore diameters 70nm, thickness 200nm, and approximate porosity 24.6%.This project successfully produces a complete device by integrating microwell array, nanoporous membrane, and a device substrate. However, the desired thickness for the microwell array is not achieved. The thickness of the microwell is 100 µm while the size of a bacteria is only a few microns. Thin microwell array with 5 µm thickness is easily detached from the substrate during development. To solve this problem, a different mixture of developer solution is needed that will develop the OSTE but will not destroy the structure. Nevertheless, this early design of the platform is a promising start in the early development of rapid antibiotic susceptibility test (AST).

Fall-related incidents are the most common cause of injury among elderly, and may result in hip fractures. Svein Kleiven and Hans von Holst, professors at the Royal Institute of Technology, have developed a technology for a compliant flooring system with the intention of reducing the peak force acting on the proximal femur during a fall. A project is underway to make the floor commercially available, where this thesis was a part of the first phase of the project. The goal with this thesis was to modify a computational human body model (HBM) to predict hip fractures when falling, using different material and geometry regarding the flooring system. It was also to compile a set of requirements that the final product would need to fulfill. The human body model was validated and modified using a study where cadavers had been tested. With the Finite Element Method (FEM), impacts were performed with the human body model and a flooring system. Requirements regarding the flooring system were compiled using literature studies, a study visit in a geriatric care facility and dialogues with well-informed people. Modifications involving contacts, material and the proximal femur were made on the model. A total of 18 simulations were performed using different flooring systems. When compared to rigid floor condition, all configurations showed a reduction in peak force on the proximal femur. The maximal attenuation was calculated to 33.04%, provided by pins with a diameter of 3 mm and with a distance of 5 mm between their midpoints.

Neuronavigation is the technology used during brain surgery to verify the location in which the surgeon is working in terms of depth and width. Despite being considered to be helpful, the technology lacks when it comes to the consideration of brain shift, the dynamic changes of the brain during surgery. The purpose of this report was to verify the suspected need of an additional tool and investigate the market size and dynamics. The thesis work mainly followed the method of Biodesign – The process of innovating medical technologies, which covers all steps in the innovation process. To fit the specific purpose, the method was partly amended.The results of the report are divided into two categories – need and market. A need statement was developed and the market was mapped in terms of size and other players in the market space. The conclusions drawn are that there is a need for technology development in order to safely account for brain shift during tumour resection surgery, but that it is likely to be hard to enter the market due to other researchers who have made more progress in the area.

Image quality in medical x-ray imaging is dependent on the characteristics of the x-ray photons produced  by  the  x-ray  tube,  the  detection  of  the  x-ray  photons  in  the  detector  and  the  image processing. X-ray image guidance during endovascular interventions is usually performed with C-arm fluoroscopy. It is of major importance that the image quality is sufficient for the procedure. The understanding of what affects the image quality could be made easier if the performance of the different parts of the system were to be judged separately, rather than considering the system as a whole.   The term Image QualityTube Output that is used throughout this thesis describes the image quality that only depends on the characteristics of the x-ray beam, such as spectral distribution and fluence, without the consideration of detector characteristics or image processing. The aim of this thesis was to  propose  a  method  to  evaluate  the  Image  QualityTube  Output  and  then  compare  two  different fluoroscopy systems used for endovascular interventions with respect to Image QualityTube  Output, patient dose and personnel exposure, all as a function of patient thickness.   To limit to the x-ray tube output a third x-ray system was used. In that way detector and image processing would be the same and the differences in image quality would only be due to the tube output. The imaging parameters set by the Automatic Dose Rate Control, ADRC, for various patient thicknesses, was found by using Polymethyl-methacrylate, PMMA, as patient. A test object was introduced and exposed together with the PMMA of varying thickness, using the parameters set by the ADRC. The images were evaluated with respect to low contrast sensitivity and contrast to noise ratio, CNR. The effective dose rate to the patient was calculated using Monte Carlo simulation. Finally, the personnel exposure was measured using active dosimeters. The  results  showed  differences  in  the  systems  performance  and  Image  QualityTube  Output.  Also interesting differences in the effective dose rate to the patient was found. Altogether the results indicate that the proposed method is a feasible way to compare the Image QualityTube Output of two fluoroscopic systems.

Instruction manuals are supposed to be used to ensure safe and proper use of medical devices. The instruction manuals could be paper or digital docu- ments, animations or videos. Conveying information and instructions through these media might not be the most efficient approach, especially in healthcare where physicians and nurses experience high cognitive load and prefer hands-on experience with the devices [1]. This thesis investigates how a possibly more efficient and intuitive instruction manual could be developed using mobile marker-based Augmented Reality; a technology that combines a virtual and physical world. By analyzing and comparing different marker-based Augmented Reality sys- tems and Software Development Kits (SDKs), the most suitable system, in terms of availability and functionality, was chosen for the implementation. Two mobile AR instruction manual prototypes were developed, as a proof-of-concept, using Unity3D with Qualcomm’s Vuforia extension. The resulting prototypes demonstrate the opportunity to integrate informa- tion and instructions in our physical world and the difficulties associated with the interaction and user experience with the relatively novel Augmented Reality interface.

No description available.

No description available.