The deployment of human-augmented machine learning (ML) systems has become a recommended organizational best practice. ML systems use algorithms that rely on training data labeled by human annotators. However, human involvement in reviewing and labeling consumers' voice data to train speech recognition systems for Amazon Alexa, Microsoft Cortana, and the like has raised privacy concerns among consumers and privacy advocates. We use the enhanced APCO model as the theoretical lens to investigate how the disclosure of human involvement during the supervised machine learning process affects consumers' privacy decision making. In a scenario-based experiment with 499 participants, we present various company privacy policies to participants to examine their trust and privacy considerations, then ask them to share reasons why they would or would not opt in to share their voice data to train a companies' voice recognition software. We find that the perception of human involvement in the ML training process significantly influences participants' privacy-related concerns, which thereby mediate their decisions to share their voice data. Furthermore, we manipulate four factors of a privacy policy to operationalize various cognitive biases actively present in the minds of consumers and find that default trust and salience biases significantly affect participants' privacy decision making. Our results provide a deeper contextualized understanding of privacy-related concerns that may arise in human-augmented ML system configurations and highlight the managerial importance of considering the role of human involvement in supervised machine learning settings. Importantly, we introduce perceived human involvement as a new construct to the information privacy discourse. Although ubiquitous data collection and increased privacy breaches have elevated the reported concerns of consumers, consumers' behaviors do not always match their stated privacy concerns. Researchers refer to this as the privacy paradox, and decades of information privacy research have identified a myriad of explanations why this paradox occurs. Yet the underlying crux of the explanations presumes privacy concern to be the appropriate proxy to measure privacy attitude and compare with actual privacy behavior. Often, privacy concerns are situational and can be elicited through the setup of boundary conditions and the framing of different privacy scenarios. Drawing on the cognitive model of empowerment and interest, we propose a multidimensional privacy interest construct that captures consumers' situational and dispositional attitudes toward privacy, which can serve as a more robust measure in conditions leading to the privacy paradox. We define privacy interest as a consumer's general feeling toward reengaging particular behaviors that increase their information privacy. This construct comprises four dimensions—impact, awareness, meaningfulness, and competence—and is conceptualized as a consumer's assessment of contextual factors affecting their privacy perceptions and their global predisposition to respond to those factors. Importantly, interest was originally included in the privacy calculus but is largely absent in privacy studies and theoretical conceptualizations. Following MacKenzie et al. (2011), we developed and empirically validated a privacy interest scale. This study contributes to privacy research and practice by reconceptualizing a construct in the original privacy calculus theory and offering a renewed theoretical lens through which to view consumers' privacy attitudes and behaviors. / Doctor of Philosophy / The deployment of human-augmented machine learning (ML) systems has become a recommended organizational best practice. ML systems use algorithms that rely on training data labeled by human annotators. However, human involvement in reviewing and labeling consumers' voice data to train speech recognition systems for Amazon Alexa, Microsoft Cortana, and the like has raised privacy concerns among consumers and privacy advocates. We investigate how the disclosure of human involvement during the supervised machine learning process affects consumers' privacy decision making and find that the perception of human involvement in the ML training process significantly influences participants' privacy-related concerns. This thereby influences their decisions to share their voice data. Our results highlight the importance of understanding consumers' willingness to contribute their data to generate complete and diverse data sets to help companies reduce algorithmic biases and systematic unfairness in the decisions and outputs rendered by ML systems. Although ubiquitous data collection and increased privacy breaches have elevated the reported concerns of consumers, consumers' behaviors do not always match their stated privacy concerns. This is referred to as the privacy paradox, and decades of information privacy research have identified a myriad of explanations why this paradox occurs. Yet the underlying crux of the explanations presumes privacy concern to be the appropriate proxy to measure privacy attitude and compare with actual privacy behavior. We propose privacy interest as an alternative to privacy concern and assert that it can serve as a more robust measure in conditions leading to the privacy paradox. We define privacy interest as a consumer's general feeling toward reengaging particular behaviors that increase their information privacy. We found that privacy interest was more effective than privacy concern in predicting consumers' mobilization behaviors, such as publicly complaining about privacy issues to companies and third-party organizations, requesting to remove their information from company databases, and reducing their self-disclosure behaviors. By contrast, privacy concern was more effective than privacy interest in predicting consumers' behaviors to misrepresent their identity. By developing and empirically validating the privacy interest scale, we offer interest in privacy as a renewed theoretical lens through which to view consumers' privacy attitudes and behaviors.

Argumentation framework is a common technique in Artificial Intelligence and related fields. It is a good way of formalizing, resolving conflicts and helping with defeasible reasoning. This thesis discusses the exploration of the quantitative bipolar argumentation framework applied in multi-agent systems. Different agents in a multi-agent systems have various capabilities, and they contribute in different ways to the system goal. The purpose of this study is to explore approaches of explaining the overall behavior and output from a multi-agent system and enable explainability in the multi-agent systems. By exploring the properties of the quantitative bipolar argumentation framework using some techniques from explainable Artificial Intelligence (AI), the system will generate output with explanations given by the argumentation framework. This thesis gives a general overview of argumentation frameworks and common techniques from explainable AI. The study mainly focuses on the exploration of properties and interactive algorithms of quantitative bipolar argumentation framework. It introduces explanation techniques to the quantitative bipolar argumentation framework. A Graphical User Interface (GUI) application is included in order to present the results of the explanation. / Argumentationsramar är en vanlig teknik inom artificiell intelligens och relaterade områden. Det är ett bra sätt att formalisera, lösa konflikter och hjälpa till med defekta resonemang. I den här avhandlingen diskuteras utforskningen av den kvantitativa bipolära argumentationsramen som tillämpas i fleragentsystem. Olika agenter i ett system med flera agenter har olika kapacitet och bidrar på olika sätt till systemets mål. Syftet med den här studien är att utforska metoder för att förklara det övergripande beteendet och resultatet från ett system med flera agenter och möjliggöra förklarbarhet i systemen med flera agenter. Genom att utforska egenskaperna hos den kvantitativa bipolära argumentationsramen med hjälp av vissa tekniker från förklaringsbara AI kommer systemet att generera utdata med förklaringar som ges av argumentationsramen. Denna avhandling ger en allmän översikt över argumentationsramar och vanliga tekniker från förklaringsbara AI. Studien fokuserar främst på utforskandet av egenskaper och interaktiva algoritmer för det kvantitativa bipolära argumentationsramverket och introducerar tillämpningen av förklaringstekniker på det kvantitativa bipolära argumentationsramverket. En GUI-applikation ingår för att presentera resultaten av förklaringen.

The overprescription of antibiotics has resulted in bacteria resistance, which is considered a global threat to global health. Deciding if antibiotics should be prescribed or not from individual visits of patients’ medical records in Swedish can be considered a text classification task, one of the applications of Natural Language Processing (NLP). However, medical experts and patients can not trust a model if explanations for its decision are not provided. In this work, multilingual and monolingual Transformer models are evaluated for the medical classification task. Furthermore, local explanations are obtained with SHapley Additive exPlanations and Integrated Gradients to compare the models’ predictions and evaluate the explainability methods. Finally, the local explanations are also aggregated to obtain global explanations and understand the features that contributed the most to the prediction of each class. / Felaktig utskrivning av antibiotika har resulterat i ökad antibiotikaresistens, vilket anses vara ett globalt hot mot global hälsa. Att avgöra om antibiotika ska ordineras eller inte från patientjournaler på svenska kan betraktas som ett textklassificeringproblem, en av tillämpningarna av Natural Language Processing (NLP). Men medicinska experter och patienter kan inte lita på en modell om förklaringar till modellens beslut inte ges. I detta arbete utvärderades flerspråkiga och enspråkiga Transformersmodeller för medisinska textklassificeringproblemet. Dessutom erhölls lokala förklaringar med SHapley Additive exPlanations och Integrated gradients för att jämföra modellernas förutsägelser och utvärdera metodernas förklarbarhet. Slutligen aggregerades de lokala förklaringarna för att få globala förklaringar och förstå de ord som bidrog mest till modellens förutsägelse för varje klass.

Syftet med denna studie är att undersökavilket användargränssnitt som potentiella användare föredrar för att utbyta trafikrelateradinformation. Forskningsfrågan som ska besvaras är följande. Vilket användargränssnittföredras för kommunikation i ett självkörande fordon? Genom att läsa denna studie fårläsaren en fördjupad insikt för hur föredragna användargränssnitt kan öka acceptansen hospotentiella användare. En kvantitativ metod användes för att genomföra enstickprovsundersökning med hjälp av en webbaserad enkät som distribuerades på olika sättsom Facebook, Linkedin, m.m, för att besvara studiens syfte. Den empiriskadatainsamlingen resulterade i 201 insamlade svar. Resultatet visade att 41,3 % avrespondenterna föredrog skärmgränssnitt och 35,3% föredrog ett multimodalt gränssnitt föratt integrera med ett självkörande fordon. Sammanlagt 84,1% av respondenterna besvaradeatt användningen av det önskade gränssnittet skulle öka effektiviteten ochkommunikationen vid utbyte av information med fordonet. Slutsatsen är att valet avanvändargränssnitt kan påverkas av olika faktorer, såsom erfarenheter och teknologiskaförväntningar. Framtida utveckling av gränssnitt och teknologier bör sträva efter attinkludera en mångfald av alternativ för att tillgodose användarnas behov och preferensernär det gäller att kommunicera med fordon. / Syftet med denna studie är att undersöka vilket användargränssnitt som potentiella användare föredrar för att utbyta trafikrelaterad information. Forskningsfrågan som ska besvaras är följande. Vilket användargränssnitt föredras för kommunikation i ett självkörande fordon? Genom att läsa denna studie får läsaren en fördjupad insikt för hur föredragna användargränssnitt kan öka acceptansen hos potentiella användare. En kvantitativ metod användes för att genomföra en stickprovsundersökning med hjälp av en webbaserad enkät som distribuerades på olika sätt som Facebook, Linkedin, m.m, för att besvara studiens syfte. Den empiriska datainsamlingen resulterade i 201 insamlade svar. Resultatet visade att 41,3 % av respondenterna föredrog skärmgränssnitt och 35,3% föredrog ett multimodalt gränssnitt för att integrera med ett självkörande fordon. Sammanlagt 84,1% av respondenterna besvarade att användningen av det önskade gränssnittet skulle öka effektiviteten och kommunikationen vid utbyte av information med fordonet. Slutsatsen är att valet av användargränssnitt kan påverkas av olika faktorer, såsom erfarenheter och teknologiska förväntningar. Framtida utveckling av gränssnitt och teknologier bör sträva efter att inkludera en mångfald av alternativ för att tillgodose användarnas behov och preferenser när det gäller att kommunicera med fordon.

Early detection reduces mortality in lung cancer, but it is also considered as a challenge for oncologists and for healthcare systems. In addition, screening modalities like CT-scans come with undesired effects, many suspected patients are wrongly diagnosed with lung cancer. This thesis contributes to solve the challenge of early lung cancer detection by utilizing unique data consisting of self-reported symptoms. The proposed method is a predictive machine learning algorithm based on natural language processing, which handles the data as an unstructured data set. A replication of a previous study where a prediction model based on a conventional multivariate machine learning using the same data is done and presented, for comparison. After evaluation, validation and interpretation, a set of variables were highlighted as early predictors of lung cancer. The performance of the proposed approach managed to match the performance of the conventional approach. This promising result opens for further development where such an approach can be used in clinical decision support systems. Future work could then involve other modalities, in a multimodal machine learning approach. / Tidig lungcancerdiagnostisering kan öka chanserna för överlevnad hos lungcancerpatienter, men att upptäcka lungcancer i ett tidigt stadie är en av de större utmaningarna för onkologer och sjukvården. Idag undersöks patienter med riskfaktorer baserat på rökning och ålder, dessa undersökningar sker med hjälp av bland annat medicinskt avbildningssystem, då oftast CT-bilder, vilket medför felaktiga och kostsamma diagnoser. Detta arbete föreslår en maskininlärninig algoritm baserad på Natural language processing, som genom analys och bearbetning av ostrukturerade data, av patienternas egna anamneser, kan prediktera lungcancer. Arbetet har genomfört en jämförelse med en konventionell maskininlärning algoritm baserat på en replikering av ett annat studie där samma data behandlades som strukturerad. Den föreslagna metoden har visat ett likartat resultat samt prestanda, och har identifierat riskfaktorer samt symptom för lungcancer. Detta arbete öppnar upp för en utveckling mot ett kliniskt användande i form av beslutsstödsystem, som även kan hantera elektriska hälsojournaler. Andra arbeten kan vidareutveckla metoden för att hantera andra varianter av data, så som medicinska bilder och biomarkörer, och genom det förbättra prestandan.

A Neurodegenerative Disease (ND) is progressive damage to brain neurons, which the human body cannot repair or replace. The well-known examples of such conditions are Dementia and Alzheimer’s Disease (AD), which affect millions of lives each year. Although conducting numerous researches, there are no effective treatments for the mentioned diseases today. However, early diagnosis is crucial in disease management. Diagnosing NDs is challenging for neurologists and requires years of training and experience. So, there has been a trend to harness the power of deep learning, including state-of-the-art Convolutional Neural Network (CNN), to assist doctors in diagnosing such conditions using brain scans. The CNN models lead to promising results comparable to experienced neurologists in their diagnosis. But, the advent of transformers in the Natural Language Processing (NLP) domain and their outstanding performance persuaded Computer Vision (CV) researchers to adapt them to solve various CV tasks in multiple areas, including the medical field. This research aims to develop Vision Transformer (ViT) models using Alzheimer’s Disease Neuroimaging Initiative (ADNI) dataset to classify NDs. More specifically, the models can classify three categories (Cognitively Normal (CN), Mild Cognitive Impairment (MCI), Alzheimer’s Disease (AD)) using brain Fluorodeoxyglucose (18F-FDG) Positron Emission Tomography (PET) scans. Also, we take advantage of Automated Anatomical Labeling (AAL) brain atlas and attention maps to develop explainable models. We propose three ViTs, the best of which obtains an accuracy of 82% on the test dataset with the help of transfer learning. Also, we encode the AAL brain atlas information into the best performing ViT, so the model outputs the predicted label, the most critical region in its prediction, and overlaid attention map on the input scan with the crucial areas highlighted. Furthermore, we develop two CNN models with 2D and 3D convolutional kernels as baselines to classify NDs, which achieve accuracy of 77% and 73%, respectively, on the test dataset. We also conduct a study to find out the importance of brain regions and their combinations in classifying NDs using ViTs and the AAL brain atlas. / <p>This thesis was awarded a prize of 50,000 SEK by Getinge Sterilization for projects within Health Innovation.</p>

One of the biggest challenges in healthcare is Emergency Department (ED)crowding which creates high constraints on the whole healthcare system aswell as the resources within and can be the cause of many adverse events.Is is a well known problem were a lot of research has been done and a lotof solutions has been proposed, yet the problem still stands unsolved. Byanalysing Real-World Data (RWD), complex problems like ED crowding couldbe better understood. Currently very few applications of survival analysis hasbeen adopted for the use of production data in order to analyze the complexityof logistical problems. The aims for this thesis was to apply survival analysisthrough advanced Machine Learning (ML) models to RWD collected at aSwedish hospital too see how the Length Of Stay (LOS) until admission ordischarge were affected by different factors. This was done by formulating thecrowding in the ED for survival analysis through the use of the LOS as thetime and the decision regarding admission or discharge as the event in order tounfold the clinical complexity of the system and help impact clinical practiceand decision making.By formulating the research as time-to-event in combination with ML, thecomplexity and non linearity of the logistics in the ED is viewed from a timeperspective with the LOS acting as a Key Performance Indicator (KPI). Thisenables the researcher to look at the problem from a system perspective andshows how different features affect the time that the patient are processedin the ED, highlighting eventual problems and can therefore be useful forimproving clinical decision making. Five models: Cox Proportional Hazards(CPH), Random Survival Forests (RSF), Gradient Boosting (GB), ExtremeGradient Boosting (XGB) and DeepSurv were used and evaluated using theConcordance index (C-index) were GB were the best performing model witha C-index of 0.7825 showing that the ML models can perform better than thecommonly used CPH model. The models were then explained using SHapleyAdaptive exPlanations (SHAP) values were the importance of the featureswere shown together with how the different features impacted the LOS. TheSHAP also showed how the GB handled the non linearity of the features betterthan the CPH model. The five most important features impacting the LOS wereif the patient received a scan at the ED, if the visited and emergency room,age, triage level and the label indicating what type of medical team seemsmost fit for the patient. This is clinical information that could be implementedto reduce the crowding through correct decision making. These results show that ML based survival analysis models can be used for further investigationregarding the logistic challenges that healthcare faces and could be furtherused for data analysis with production data in similar cases. The ML survivalanalysis pipeline can also be used for further analysis and can act as a first stepin order to pinpoint important information in the data that could be interestingfor deeper data analysis, making the process more efficient. / En av de största utmaningarna inom vården är trängsel på akuten som skaparstora ansträngninar inom vårdsystemet samt på dess resurser och kan varaorsaken till många negativa händelser. Det är ett välkänt problem där mycketforskning har gjorts och många lösningar har föreslagits men problemetär fortfarande olöst. Genom att analysera verklig data så kan komplexaproblem som trängsel på akuten bli bättre förklarade. För närvarande harfå tillämpningar av överlevnadsanalys applicerats på produktionsdata för attanalysera komplexiteten av logistiska problem. Syftet med denna avhandlingvar att tillämpa överlevnadsanalys genom avancerade maskininlärningsmetoderpå verklig data insamlat på ett svenskt sjukhust för att se hur vistelsens längdför patienten fram till inläggning påverkades av olika faktorer. Detta gjordesgenom att applicera överlevnadsnanalys på trängsel på akuten genom attanvända vistelsens längd som tid och beslutet om intagning eller utskrivningsom händelsen. Detta för att kunna analysera systemets kliniska komplexitetoch bidra till att påverka klinisk praxis och beslutsfattande.Genom att formulera forskningsfrågan som en överlevnadsanalys i kombinationmed maskininlärning kan den komplexitet och icke-linjäritet som logistikenpå akuten innebär studeras genom ett tidsperspektiv där vistelsens längdfungerar som ett nyckeltal. Detta gör det möjligt för forskaren att ävenstudera problemet från ett systemperspektiv och visar hur olika egenskaperoch situationer påverkar den tid som patienten bearbetas på akuten. Detta uppmärksammar eventuella problem och kan därför vara användbart för attförbättra det kliniska beslutsfattandet. Fem olika modeller: CPH, RSF, GB,XGB och DeepSurv användes och utvärderades med hjälp av C-index där GBvar den bäst presterande modellen med ett C-index på 0.7825 vilket visar attmaskininlärningsmetoderna kan prestera bättre än den klassiska och vanligtförekommande CPH modellen. Modellerna förklarades sedan med hjälp utavSHAP värden där vikten utav de olika variablerna visades tillsammmans med deras påverkan. SHAP visade även att GB modellen hanterade icke-linjäriteten bättre än CPH modellen. De fem viktigaste variablerna som påverkade vistelsens längd till intagning var om patienten blev scannad påakutmottagningen, om de blev mottagna i ett akutrum, ålder, triagenivå ochvilket medicinskt team som ansågs bäst lämpat för patienten. Detta är kliniskinformation som skulle kunna implementeras genom beslutsfattande för attminska trängseln på akuten. Dessa resultat visar att maskininlärningsmetoderför överlevnadsanalys kan användas för vidare undersökning angående de logistiska utmaningar som sjukvården står inför och kan även användas ytterligareför datanalys med produktionsdata i liknande fall. Processen med överlevnadsanalys och ML kan även användas för vidare analys och kan agera som ett förstasteg för att framhäva viktig information i datan som skulle vara intressant fördjupare data analys. Detta skulle kunna göra processen mer effektiv.

The growing presence of algorithm-generated recommendations in AI-powered services highlights the importance of responsible systems that explain outputs in a human-understandable form, especially in an automotive context. Implementing explainability in recommendations of AI-powered eco-driving is important in ensuring that drivers understand the underlying reasoning behind the recommendations. Previous literature on explainable AI (XAI) has been primarily technological-centered, and only a few studies involve the end-user perspective. There is a lack of knowledge of drivers' needs and requirements for explainability in an AI-powered eco-driving context. This study addresses the attributes that make a “satisfactory” explanation, i,e., a satisfactory interface between humans and AI. This study uses scenario-based interviews to understand the explainability attributes that influence truck drivers' intention to use eco-driving recommendations. The study used thematic analysis to categorize seven attributes into context-dependent (Format, Completeness, Accuracy, Timeliness, Communication) and generic (Reliability, Feedback loop) categories. The study contributes context-dependent attributes along three design dimensions: Presentational, Content-related, and Temporal aspects of explainability. The findings of this study present an empirical foundation into end-users' explainability needs and provide valuable insights for UX and system designers in eliciting end-user requirements.

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.

The riveting interplay of industrialization, informalization, and exponential technological growth of recent years has shifted the attention from classical machine learning techniques to more sophisticated deep learning approaches; yet its intrinsic black-box nature has been impeding its widespread adoption in transparency-critical operations. In this rapidly evolving landscape, where the symbiotic relationship between research and practical applications has never been more interwoven, the contribution of this paper is twofold: advancing gradient faithfulness of CAM methods and exploring new frontiers beyond it. In the first part, we theorize three novel gradient-based CAM formulations, aimed at replacing and superseding traditional Grad-CAM-based methods by tackling and addressing the intricately and persistent vanishing and saturating gradient problems. As a consequence, our work introduces novel enhancements to Grad-CAM that reshape the conventional gradient computation by incorporating a customized and adapted technique inspired by the well-established and provably Expected Gradients’ difference-from-reference approach. Our proposed techniques– Expected Grad-CAM, Expected Grad-CAM++and Guided Expected Grad-CAM– as they operate directly on the gradient computation, rather than the recombination of the weighing factors, are designed as a direct and seamless replacement for Grad-CAM and any posterior work built upon it. In the second part, we build on our prior proposition and devise a novel CAM method that produces both high-resolution and class-discriminative explanation without fusing other methods, while addressing the issues of both gradient and CAM methods altogether. Our last and most advanced proposition, Hyper Expected Grad-CAM, challenges the current state and formulation of visual explanation and faithfulness and produces a new type of hybrid saliencies that satisfy the notion of natural encoding and perceived resolution. By rethinking faithfulness and resolution is possible to generate saliencies which are more detailed, localized, and less noisy, but most importantly that are composed of only concepts that are encoded by the layerwise models’ understanding. Both contributions have been quantitatively and qualitatively compared and assessed in a 5 to 10 times larger evaluation study on the ILSVRC2012 dataset against nine of the most recent and performing CAM techniques across six metrics. Expected Grad-CAM outperformed not only the original formulation but also more advanced methods, resulting in the second-best explainer with an Ins-Del score of 0.56. Hyper Expected Grad-CAM provided remarkable results across each quantitative metric, yielding a 0.15 increase in insertion when compared to the highest-scoring explainer PolyCAM, totaling to an Ins-Del score of 0.72.