Browsing by Author "Marttinen, Pekka"
Now showing 1 - 20 of 109
Results Per Page
Sort Options
Item Accounting for measurement errors in food diaries for continuous glucose prediction using Kalman filtering and Monte Carlo methods(2022-07-29) Väärä, Teijo; Särkkä, Simo; Perustieteiden korkeakoulu; Marttinen, PekkaContinuous blood glucose prediction for diabetic and non-diabetic patients has important medical applications. It can help in planning better personalized treatment for patients, and it may help in understanding how some severe complications develop in type 2 diabetic patients after bariatric surgery. Continuous blood glucose prediction is usually based on patient-kept food diaries and automated blood glucose level measurements. Often the patient-kept records of amounts of nutrients in meals, and the timing of meals, can have inaccuracies in them. Therefore, it is necessary to build prediction models which can take those measurement errors in the independent variables into account. Based on previous success of approaching this task with errors-in-variables models using Gaussian process regression, we propose using errors-in-variables models using Kalman filtering and Markov chain Monte Carlo methods for parameter estimation instead of Gaussian process regression. This approach should on theoretical grounds yield similar results to using Gaussian process regression, while having additional benefit of better scalability in computational complexity with respect to number of time steps in a time series. In this thesis, we provide the mathematical formulation of two different blood glucose response models. We test our approach to the problem with real-world dataset gathered from gastric bypass patients. While not outperforming the methods used in previous work, our preliminary results show promising prospects for further study and fine-tuning of the models.Item Assessing wearable data of chronic low back pain patients(2022-03-22) Gröhn, Tommi; Liikkanen, Sammeli; Perustieteiden korkeakoulu; Marttinen, PekkaThere is a high need for measuring fear of movement of chronic low back patients automatically. Firstly, Tampa Scale for Kinesiophobia and other scales to measure fear of movement require the patient to fill in questionnaires. These forms lead the patient repeatedly back to the disease. Secondly, automatic kinesiophobia measurement techniques would also be an easy, objective and passive way to personalize the treatment of pain. This thesis creates an algorithm to automatically detect fear of movement out of wearable data. The patients' and the control group's movement data is collected from a digital therapy in a game-like environment in virtual reality from a study called Painlab. Indeed, the Painlab data set enables finding differences between these two groups. On the other hand, the patients' changes in their movement data between different therapy sessions is investigated with a completely separate VIRPI data set. The methods in this thesis are constructed with an incremental design. Firstly, exploratory data analysis is conducted to detect which statistical features differentiate the patients and the control group well. Secondly, the players' time series data is automatically segmented with hidden Markov models, and the segmentation is carefully investigated. Thirdly, these segments are integrated together using the knowledge of exploratory data analysis. Lastly, the created methodology is evaluated with the Painlab test set and the VIRPI data set. Classifying the players into patients and healthy individuals was done reliably with the Painlab data set. Promising results in capturing the patients' development between therapy sessions were also obtained in the VIRPI study. Moreover, the segmentation with hidden Markov models functioned well, which may be helpful for further development of the algorithm. The findings of this research are a major step towards automatic classification of fear of movement. In the future, the created methodology could be used as a digital biomarker of kinesiophobia in the treatment of chronic pain.Item Automatic detection of adventitious respiratory sounds using deep learning(2022-01-24) Vilkki, Eetu; Kouros, Alexis; Perustieteiden korkeakoulu; Marttinen, PekkaAdventitious respiratory sounds are one of the most reliable and common indicators of pulmonary diseases. Automatic algorithms can increase the reliability and accessibility of their detection, and, consequently, research into such methods has been rapidly increasing. The state-of-the-art results today have been achieved using deep learning models. However, their performance is still lacking, there is no consensus on which methods perform the best, and it is not clear how different parts of the machine learning pipeline impact the results in this task. This thesis researched how the different parts of the pipeline impact the results and which methods are recommended to understand better how the automatic detection of adventitious respiratory sounds can be made more reliable with deep learning. These parts include data preprocessing, feature extraction, model architectures, data augmentation and balancing, and pretraining. The focus was on spectrograms as the input features and convolutional neural networks (CNNs) and Transformers as the model architectures. The first key finding was that deep learning could be applied to respiratory sound classification, the best model achieving an average score of 68.7\%. Each part of the pipeline had a significant impact, although some methods recommended by other studies, such as filtering and smart padding, had a minimal impact or were even detrimental. Mel-spectrograms outperformed the other spectrograms, and their effectiveness could be improved by combining three separate spectrograms into one three-channel input. CNNs were generally superior to Transformers, but both architectures showed potential. Data augmentation and pretraining were both highly impactful, improving the average score by 3.9\% and 6.1\%, respectively. Although the main limitation in performance for these models seemed to be the small amount of data, it was shown that advancements in other applications, such as image classification and speech recognition, also benefited the classification of respiratory sounds.Item Baseline gene expression in subcutaneous adipose tissue predicts diet-induced weight loss in individuals with obesity(PeerJ, 2023-03) Oghabian, Ali; van der Kolk, Birgitta W.; Marttinen, Pekka; Valsesia, Armand; Langin, Dominique; Saris, W. H.; Astrup, Arne; Blaak, Ellen E.; Pietiläinen, Kirsi H.; University of Helsinki; Computer Science Professors; Nestlé Institute of Health Sciences; Toulouse University Hospitals; Maastricht University; Novo Nordisk Fonden; Department of Computer ScienceBackground: Weight loss effectively reduces cardiometabolic health risks among people with overweight and obesity, but inter-individual variability in weight loss maintenance is large. Here we studied whether baseline gene expression in subcutaneous adipose tissue predicts diet-induced weight loss success. Methods: Within the 8-month multicenter dietary intervention study DiOGenes, we classified a low weight-losers (low-WL) group and a high-WL group based on median weight loss percentage (9.9%) from 281 individuals. Using RNA sequencing, we identified the significantly differentially expressed genes between high-WL and low-WL at baseline and their enriched pathways. We used this information together with support vector machines with linear kernel to build classifier models that predict the weight loss classes. Results: Prediction models based on a selection of genes that are associated with the discovered pathways ‘lipid metabolism’ (max AUC = 0.74, 95% CI [0.62–0.86]) and ‘response to virus’ (max AUC = 0.72,95% CI [0.61–0.83]) predicted the weight-loss classes high-WL/low-WL significantly better than models based on randomly selected genes (P < 0.01). The performance of the models based on ‘response to virus’ genes is highly dependent on those genes that are also associated with lipid metabolism. Incorporation of baseline clinical factors into these models did not noticeably enhance the model performance in most of the runs. This study demonstrates that baseline adipose tissue gene expression data, together with supervised machine learning, facilitates the characterization of the determinants of successful weight loss.Item Bayesian detection of genetic associations for metabolic syndrome(2009) Pennala, Eero; Marttinen, Pekka; Informaatio- ja luonnontieteiden tiedekunta; Teknillinen korkeakoulu; Helsinki University of Technology; Lampinen, JoukoThe causes for common Mendelian diseases involving only one gene have been thoroughly studied and now the interest is to find connections between multiple genes in risk stratification. In this thesis the focus is on the methods used for detecting genetic associations. The biological background and the prevalent methods in genome wide association studies are reviewed. The found effects on disease risk are small and therefore analysis is demanding. In this thesis logistic regression model is considered, based on the binomially distributed data. Additive, dominant and recessive models of inheritance are used and the effect of different priors experimented. In this work traditional frequentist methods are compared with Bayesian inference. Results from the single-locus analysis ignore association between loci. However, analyzing all possible combinations of gene-gene interactions is computationally intractable. Therefore Markov chain Monte Carlo and evolutionary algorithms are used to sample from the model space and model averaging is used to search for variants that occur in favourable samples. The methods are compared and evaluated with a small simulated data set. Additionally, real world data depicting metabolic syndrome X is analyzed and the results evaluated. From chromosomes 5 and 11 the algorithms are able find multiple loci, which have been previously found to be associated with metabolic syndrome X or related diseases. Confirming the validity of previously unknown associated loci is complicated in any event.Item A Bayesian model of acquisition and clearance of bacterial colonization incorporating within-host variation(Public Library of Science, 2019-04-01) Järvenpää, Marko; Sater, Mohamad R.Abdul; Lagoudas, Georgia K.; Blainey, Paul C.; Miller, Loren G.; McKinnell, James A.; Huang, Susan S.; Grad, Yonatan H.; Marttinen, Pekka; Department of Computer Science; Probabilistic Machine Learning; Helsinki Institute for Information Technology (HIIT); Professorship Kaski Samuel; Centre of Excellence in Computational Inference, COIN; Professorship Marttinen P.; Harvard University; Massachusetts Institute of Technology; University of California, Los Angeles; University of California, IrvineBacterial populations that colonize a host can play important roles in host health, including serving as a reservoir that transmits to other hosts and from which invasive strains emerge, thus emphasizing the importance of understanding rates of acquisition and clearance of colonizing populations. Studies of colonization dynamics have been based on assessment of whether serial samples represent a single population or distinct colonization events. With the use of whole genome sequencing to determine genetic distance between isolates, a common solution to estimate acquisition and clearance rates has been to assume a fixed genetic distance threshold below which isolates are considered to represent the same strain. However, this approach is often inadequate to account for the diversity of the underlying within-host evolving population, the time intervals between consecutive measurements, and the uncertainty in the estimated acquisition and clearance rates. Here, we present a fully Bayesian model that provides probabilities of whether two strains should be considered the same, allowing us to determine bacterial clearance and acquisition from genomes sampled over time. Our method explicitly models the within-host variation using population genetic simulation, and the inference is done using a combination of Approximate Bayesian Computation (ABC) and Markov Chain Monte Carlo (MCMC). We validate the method with multiple carefully conducted simulations and demonstrate its use in practice by analyzing a collection of methicillin resistant Staphylococcus aureus (MRSA) isolates from a large recently completed longitudinal clinical study. An R-code implementation of the method is freely available at: https://github.com/mjarvenpaa/bacterial-colonization-model.Item Bayesian modeling of the impact of antibiotic resistance on the efficiency of MRSA decolonization(Public Library of Science, 2023-10) Ojala, Fanni; Abdul Sater, Mohamad R.; Miller, Loren G.; McKinnell, James A.; Hayden, Mary K.; Huang, Susan S.; Grad, Yonatan H.; Marttinen, Pekka; Department of Computer Science; Professorship Marttinen P.; Computer Science Professors; Computer Science - Artificial Intelligence and Machine Learning (AIML); Harvard School of Public Health; Lundquist Institute; Rush University; University of California, IrvineMethicillin-resistant Staphylococcus aureus (MRSA) is a major cause of morbidity and mortality. Colonization by MRSA increases the risk of infection and transmission, underscoring the importance of decolonization efforts. However, success of these decolonization protocols varies, raising the possibility that some MRSA strains may be more persistent than others. Here, we studied how the persistence of MRSA colonization correlates with genomic presence of antibiotic resistance genes. Our analysis using a Bayesian mixed effects survival model found that genetic determinants of high-level resistance to mupirocin was strongly associated with failure of the decolonization protocol. However, we did not see a similar effect with genetic resistance to chlorhexidine or other antibiotics. Including strain-specific random effects improved the predictive performance, indicating that some strain characteristics other than resistance also contributed to persistence. Study subject-specific random effects did not improve the model. Our results highlight the need to consider the properties of the colonizing MRSA strain when deciding which treatments to include in the decolonization protocol.Item Bayesian Survival Analysis to Model Clearance of MRSA Colonization(2022-08-22) Ojala, Fanni; Marttinen, Pekka; Perustieteiden korkeakoulu; Marttinen, PekkaItem Bayesian Variable Selection in Searching for Additive and Dominant Effects in Genome-Wide Data(2012) Peltola, Tomi; Marttinen, Pekka; Jula, Antti; Salomaa, Veikko; Perola, Markus; Vehtari, Aki; Department of Computer ScienceAlthough complex diseases and traits are thought to have multifactorial genetic basis, the common methods in genome-wide association analyses test each variant for association independent of the others. This computational simplification may lead to reduced power to identify variants with small effect sizes and requires correcting for multiple hypothesis tests with complex relationships. However, advances in computational methods and increase in computational resources are enabling the computation of models that adhere more closely to the theory of multifactorial inheritance. Here, a Bayesian variable selection and model averaging approach is formulated for searching for additive and dominant genetic effects. The approach considers simultaneously all available variants for inclusion as predictors in a linear genotype-phenotype mapping and averages over the uncertainty in the variable selection. This leads to naturally interpretable summary quantities on the significances of the variants and their contribution to the genetic basis of the studied trait. We first characterize the behavior of the approach in simulations. The results indicate a gain in the causal variant identification performance when additive and dominant variation are simulated, with a negligible loss of power in purely additive case. An application to the analysis of high- and low-density lipoprotein cholesterol levels in a dataset of 3895 Finns is then presented, demonstrating the feasibility of the approach at the current scale of single-nucleotide polymorphism data. We describe a Markov chain Monte Carlo algorithm for the computation and give suggestions on the specification of prior parameters using commonly available prior information. An open-source software implementing the method is available at http://www.lce.hut.fi/research/mm/bmagwa/ and https://github.com/to-mi/.Item Beetainterferonilääkityksen vaikutus MS-taudissa – bayesilainen analyysi(2012-12-27) Kukkonen, Matleena; Marttinen, Pekka; Sähkötekniikan korkeakoulu; Turunen, MarkusItem biMM: efficient estimation of genetic variances and covariances for cohorts with high-dimensional phenotype measurements(2017) Pirinen, Matti; Benner, Christian; Marttinen, Pekka; Järvelin, Marjo-Riitta; Rivas, Manuel A; Ripatti, Samuli; Department of Computer Science; Probabilistic Machine Learning; Centre of Excellence in Computational Inference, COIN; University of Helsinki; University of Oulu; Stanford UniversityGenetic research utilizes a decomposition of trait variances and covariances into genetic and environmental parts. Our software package biMM is a computationally efficient implementation of a bivariate linear mixed model for settings where hundreds of traits have been measured on partially overlapping sets of individuals.Item Causal Modeling of Policy Interventions From Treatment-Outcome Sequences(PMLR, 2023-07) Hizli, Caglar; John, ST; Juuti, Anne; Saarinen, Tuure; Pietiläinen, Kirsi; Marttinen, Pekka; Department of Computer Science; Krause, Andread; Brunskill, Emma; Cho, Kyunghyun; Engelhardt, Barbara; Sabato, Sivan; Scarlett, Jonathan; Professorship Marttinen P.; Probabilistic Machine Learning; Professorship Kaski Samuel; Computer Science Professors; Computer Science - Artificial Intelligence and Machine Learning (AIML); University of HelsinkiA treatment policy defines when and what treatments are applied to affect some outcome of interest. Data-driven decision-making requires the ability to predict what happens if a policy is changed. Existing methods that predict how the outcome evolves under different scenarios assume that the tentative sequences of future treatments are fixed in advance, while in practice the treatments are determined stochastically by a policy and may depend, for example, on the efficiency of previous treatments. Therefore, the current methods are not applicable if the treatment policy is unknown or a counterfactual analysis is needed. To handle these limitations, we model the treatments and outcomes jointly in continuous time, by combining Gaussian processes and point processes. Our model enables the estimation of a treatment policy from observational sequences of treatments and outcomes, and it can predict the interventional and counterfactual progression of the outcome after an intervention on the treatment policy (in contrast with the causal effect of a single treatment). We show with real-world and semi-synthetic data on blood glucose progression that our method can answer causal queries more accurately than existing alternatives.Item Cluster analysis to estimate the risk of preeclampsia in the high-risk Prediction and Prevention of Preeclampsia and Intrauterine Growth Restriction (PREDO) study(2017-03-01) Villa, Pia M.; Marttinen, Pekka; Gillberg, Jussi; Inkeri Lokki, A.; Majander, Kerttu; Ordén, Maija Riitta; Taipale, Pekka; Pesonen, Anukatriina; Räikkönen, Katri; Hämäläinen, Esa; Kajantie, Eero; Laivuori, Hannele; Department of Computer Science; Centre of Excellence in Computational Inference, COIN; Helsinki Institute for Information Technology (HIIT); Professorship Kaski Samuel; Probabilistic Machine Learning; University of Helsinki; University of Tübingen; Kuopio University Hospital; Suomen Terveystalo Oy; Finnish Institute for Health and WelfareObjectives Preeclampsia is divided into early-onset (delivery before 34 weeks of gestation) and lateonset (delivery at or after 34 weeks) subtypes, which may rise from different etiopathogenic backgrounds. Early-onset disease is associated with placental dysfunction. Late-onset disease develops predominantly due to metabolic disturbances, obesity, diabetes, lipid dysfunction, and inflammation, which affect endothelial function. Our aim was to use cluster analysis to investigate clinical factors predicting the onset and severity of preeclampsia in a cohort of women with known clinical risk factors. Methods We recruited 903 pregnant women with risk factors for preeclampsia at gestational weeks 12+0-13+6. Each individual outcome diagnosis was independently verified from medical records. We applied a Bayesian clustering algorithm to classify the study participants to clusters based on their particular risk factor combination. For each cluster, we computed the risk ratio of each disease outcome, relative to the risk in the general population. Results The risk of preeclampsia increased exponentially with respect to the number of risk factors. Our analysis revealed 25 number of clusters. Preeclampsia in a previous pregnancy (n = 138) increased the risk of preeclampsia 8.1 fold (95% confidence interval (CI) 5.7-11.2) compared to a general population of pregnant women. Having a small for gestational age infant (n = 57) in a previous pregnancy increased the risk of early-onset preeclampsia 17.5 fold (95%CI 2.1-60.5). Cluster of those two risk factors together (n = 21) increased the risk of severe preeclampsia to 23.8-fold (95%CI 5.1-60.6), intermediate onset (delivery between 34+0-36+6 weeks of gestation) to 25.1-fold (95%CI 3.1-79.9) and preterm preeclampsia (delivery before 37+0 weeks of gestation) to 16.4-fold (95%CI 2.0-52.4). Body mass index over 30 kg/m2 (n = 228) as a sole risk factor increased the risk of preeclampsia to 2.1-fold (95%CI 1.1-3.6). Together with preeclampsia in an earlier pregnancy the risk increased to 11.4 (95%CI 4.5-20.9). Chronic hypertension (n = 60) increased the risk of preeclampsia 5.3-fold (95%CI 2.4-9.8), of severe preeclampsia 22.2-fold (95%CI 9.9-41.0), and risk of early-onset preeclampsia 16.7-fold (95%CI 2.0-57.6). If a woman had chronic hypertension combined with obesity, gestational diabetes and earlier preeclampsia, the risk of term preeclampsia increased 4.8-fold (95%CI 0.1-21.7). Women with type 1 diabetes mellitus had a high risk of all subgroups of preeclampsia. Conclusion The risk of preeclampsia increases exponentially with respect to the number of risk factors. Early-onset preeclampsia and severe preeclampsia have different risk profile from term preeclampsia.Item Community assessment to advance computational prediction of cancer drug combinations in a pharmacogenomic screen(NATURE PUBLISHING GROUP, 2019-06-17) Kaski, Samuel; Marttinen, Pekka; , AstraZeneca-Sanger Drug Combination DREAM Consortium; Department of Computer Science; Finnish Center for Artificial Intelligence, FCAI; Probabilistic Machine Learning; Helsinki Institute for Information Technology (HIIT); Professorship Kaski Samuel; Centre of Excellence in Computational Inference, COIN; Professorship Marttinen P.The effectiveness of most cancer targeted therapies is short-lived. Tumors often develop resistance that might be overcome with drug combinations. However, the number of possible combinations is vast, necessitating data-driven approaches to find optimal patient-specific treatments. Here we report AstraZeneca’s large drug combination dataset, consisting of 11,576 experiments from 910 combinations across 85 molecularly characterized cancer cell lines, and results of a DREAM Challenge to evaluate computational strategies for predicting synergistic drug pairs and biomarkers. 160 teams participated to provide a comprehensive methodological development and benchmarking. Winning methods incorporate prior knowledge of drug-target interactions. Synergy is predicted with an accuracy matching biological replicates for >60% of combinations. However, 20% of drug combinations are poorly predicted by all methods. Genomic rationale for synergy predictions are identified, including ADAM17 inhibitor antagonism when combined with PIK3CB/D inhibition contrasting to synergy when combined with other PI3K-pathway inhibitors in PIK3CA mutant cells.Item Comprehensive Identification of Single Nucleotide Polymorphisms Associated with Beta-lactam Resistance within Pneumococcal Mosaic Genes(2014) Chewapreecha, Claire; Marttinen, Pekka; Croucher, Nicholas J; Salter, Susannah J; Harris, Simon R; Mather, Alison E; Hanage, William P; Goldblatt, David; Nosten, Francois H; Turner, Claudia; Turner, Paul; Bentley, Stephen D; Parkhill, Julian; Helsinki Insititute for Information Technology HIIT; Tietojenkäsittelytieteen laitoTraditional genetic association studies are very difficult in bacteria, as the generally limited recombination leads to large linked haplotype blocks, confounding the identification of causative variants. Beta-lactam antibiotic resistance in Streptococcus pneumoniae arises readily as the bacteria can quickly incorporate DNA fragments encompassing variants that make the transformed strains resistant. However, the causative mutations themselves are embedded within larger recombined blocks, and previous studies have only analysed a limited number of isolates, leading to the description of “mosaic genes” as being responsible for resistance. By comparing a large number of genomes of beta-lactam susceptible and non-susceptible strains, the high frequency of recombination should break up these haplotype blocks and allow the use of genetic association approaches to identify individual causative variants. Here, we performed a genome-wide association study to identify single nucleotide polymorphisms (SNPs) and indels that could confer beta-lactam non-susceptibility using 3,085 Thai and 616 USA pneumococcal isolates as independent datasets for the variant discovery. The large sample sizes allowed us to narrow the source of beta-lactam non-susceptibility from long recombinant fragments down to much smaller loci comprised of discrete or linked SNPs. While some loci appear to be universal resistance determinants, contributing equally to non-susceptibility for at least two classes of beta-lactam antibiotics, some play a larger role in resistance to particular antibiotics. All of the identified loci have a highly non-uniform distribution in the populations. They are enriched not only in vaccine-targeted, but also non-vaccine-targeted lineages, which may raise clinical concerns. Identification of single nucleotide polymorphisms underlying resistance will be essential for future use of genome sequencing to predict antibiotic sensitivity in clinical microbiology.Item Comprehensive self-tracking of blood glucose and lifestyle with a mobile application in the management of gestational diabetes: a study protocol for a randomised controlled trial (eMOM GDM study)(BMJ Publishing Group, 2022-11-07) Kytö, Mikko; Markussen, Lisa Torsdatter; Marttinen, Pekka; Jacucci, Giulio; Niinistö, Sari; Virtanen, Suvi M.; Korhonen, Tuuli E.; Sievänen, Harri; Vähä-Ypyä, Henri; Korhonen, Ilkka; Heinonen, Seppo; Koivusalo, Saila B.; Department of Computer Science; Computer Science Professors; Computer Science - Artificial Intelligence and Machine Learning (AIML); Professorship Marttinen P.; University of Helsinki; Finnish Institute for Health and Welfare; UKK Institute – Centre for Health Promotion Research; Tampere UniversityIntroduction Gestational diabetes (GDM) causes various adverse short-term and long-term consequences for the mother and child, and its incidence is increasing globally. So far, the most promising digital health interventions for GDM management have involved healthcare professionals to provide guidance and feedback. The principal aim of this study is to evaluate the effects of comprehensive and real-time self-tracking with eMOM GDM mobile application (app) on glucose levels in women with GDM, and more broadly, on different other maternal and neonatal outcomes. Methods and analysis This randomised controlled trial is carried out in Helsinki metropolitan area. We randomise 200 pregnant women with GDM into the intervention and the control group at gestational week (GW) 24-28 (baseline, BL). The intervention group receives standard antenatal care and the eMOM GDM app, while the control group will receive only standard care. Participants in the intervention group use the eMOM GDM app with continuous glucose metre (CGM) and activity bracelet for 1 week every month until delivery and an electronic 3-day food record every month until delivery. The follow-up visit after intervention takes place 3 months post partum for both groups. Data are collected by laboratory blood tests, clinical measurements, capillary glucose measures, wearable sensors, air displacement plethysmography and digital questionnaires. The primary outcome is fasting plasma glucose change from BL to GW 35-37. Secondary outcomes include, for example, self-tracked capillary fasting and postprandial glucose measures, change in gestational weight gain, change in nutrition quality, change in physical activity, medication use due to GDM, birth weight and fat percentage of the child. Ethics and dissemination The study has been approved by Ethics Committee of the Helsinki and Uusimaa Hospital District. The results will be presented in peer-reviewed journals and at conferences. Trial registration number NCT04714762.Item Computational modelling of self-reported dietary carbohydrate intake on glucose concentrations in patients undergoing Roux-en-Y gastric bypass versus one-anastomosis gastric bypass(Informa Healthcare, 2021) Ashrafi, Reza A.; Ahola, Aila J.; Rosengård-Bärlund, Milla; Saarinen, Tuure; Heinonen, Sini; Juuti, Anne; Marttinen, Pekka; Pietiläinen, Kirsi H.; Department of Computer Science; Professorship Marttinen P.; Computer Science Professors; Computer Science - Artificial Intelligence and Machine Learning (AIML); University of HelsinkiObjectives: Our aim was to investigate in a real-life setting the use of machine learning for modelling the postprandial glucose concentrations in morbidly obese patients undergoing Roux-en-Y gastric bypass (RYGB) or one-anastomosis gastric bypass (OAGB). Methods: As part of the prospective randomized open-label trial (RYSA), data from obese (BMI ≥35 kg/m2) non-diabetic adult participants were included. Glucose concentrations, measured with FreeStyle Libre, were recorded over 14 preoperative and 14 postoperative days. During these periods, 3-day food intake was self-reported. A machine learning model was applied to estimate glycaemic responses to the reported carbohydrate intakes before and after the bariatric surgeries. Results: Altogether, 10 participants underwent RYGB and 7 participants OAGB surgeries. The glucose concentrations and carbohydrate intakes were reduced postoperatively in both groups. The relative time spent in hypoglycaemia increased regardless of the operation (RYGB, from 9.2 to 28.2%; OAGB, from 1.8 to 37.7%). Postoperatively, we observed an increase in the height of the fitted response curve and a reduction in its width, suggesting that the same amount of carbohydrates caused a larger increase in the postprandial glucose response and that the clearance of the meal-derived blood glucose was faster, with no clinically meaningful differences between the surgeries. Conclusions: A detailed analysis of the glycaemic responses using food diaries has previously been difficult because of the noisy meal data. The utilized machine learning model resolved this by modelling the uncertainty in meal times. Such an approach is likely also applicable in other applications involving dietary data. A marked reduction in overall glycaemia, increase in postprandial glucose response, and rapid glucose clearance from the circulation immediately after surgery are evident after both RYGB and OAGB. Whether nondiabetic individuals would benefit from monitoring the post-surgery hypoglycaemias and the potential to prevent them by dietary means should be investigated.KEY MESSAGES The use of a novel machine learning model was applicable for combining patient-reported data and time-series data in this clinical study. Marked increase in postprandial glucose concentrations and rapid glucose clearance were observed after both Roux-en-Y gastric bypass and one-anastomosis gastric bypass surgeries. Whether nondiabetic individuals would benefit from monitoring the post-surgery hypoglycaemias and the potential to prevent them by dietary means should be investigated.Item Contextualized Graph Embeddings for Adverse Drug Event Detection(2023-03-17) Gao, Ya; Ji, Shaoxiong; Zhang, Tongxuan; Tiwari, Prayag; Marttinen, Pekka; Department of Computer Science; Amini, Massih-Reza; Canu, Stéphane; Fischer, Asja; Guns, Tias; Kralj Novak, Petra; Tsoumakas, Grigorios; Professorship Marttinen P.; Computer Science Professors; Computer Science - Artificial Intelligence and Machine Learning (AIML); Department of Computer Science; Tianjin Normal UniversityAn adverse drug event (ADE) is defined as an adverse reaction resulting from improper drug use, reported in various documents such as biomedical literature, drug reviews, and user posts on social media. The recent advances in natural language processing techniques have facilitated automated ADE detection from documents. However, the contextualized information and relations among text pieces are less explored. This paper investigates contextualized language models and heterogeneous graph representations. It builds a contextualized graph embedding model for adverse drug event detection. We employ different convolutional graph neural networks and pre-trained contextualized embeddings as the building blocks. Experimental results show that our methods can improve the performance by comparing recent ADE detection models, suggesting that a text graph can capture causal relationships and dependency between different entities in a document.Item Continual Reinforcement Learning in a Resource Allocation Simulator(2021-08-23) Kumpumäki, Antti; Lee, Denny; Perustieteiden korkeakoulu; Marttinen, PekkaThe ability to operate in a changing environment is seen as an essential aspect of any reinforcement learning algorithm designed for real-world usage, as models trained only with historical data will lead to a decrease in performance. Traditional deep reinforcement learning algorithms have suffered from slow learning of new concepts and catastrophic forgetting, where previously learned information is lost when new information is presented. The aim of this thesis is to explore how deep reinforcement learning algorithms can be modified to make them more resilient to changes in the environment, and also to evaluate the usability of such algorithms in a resource allocation problem. These topics are approached by developing a simulator that mimics the internet usage of a population in an imaginary city, where the movement and behaviour of the population change suddenly. The network demand of different parts of this city are predicted using a Soft Actor-Critic algorithm, that is implemented with an experience replay buffer that can favour experiences from different time scales. The results show that the developed algorithm can learn new information faster, as well as to keep hold of older memories when compared to a baseline solution. Furthermore, the algorithm is found to be one potential approach to the resource allocation problem. However, the results also show that the time it takes to learn new information leaves room for improvement.Item COVIDNet: An Automatic Architecture for COVID-19 Detection with Deep Learning from Chest X-ray Images(IEEE, 2022-07-01) He, Lang; Tiwari, Prayag; Su, Rui; Shi, Xiuying; Marttinen, Pekka; Kumar, Neeraj; Department of Computer Science; Professorship Marttinen P.; Computer Science Professors; Computer Science - Artificial Intelligence and Machine Learning (AIML); Xi'an University of Posts and Telecommunications; Northwest University; Yan’an University; Asia University TaiwanUp to now, the COVID-19 has been sweeping across all over the world, which has affected individual’s lives in an overwhelming way. To fight efficiently against the COVID-19, radiography and radiology images are used by clinicians in hospitals. This paper presents an integrated framework, named COVIDNet, for classifying COVID-19 patients and healthy controls. Specifically, ResNet (i.e., ResNet-18 and ResNet-50) is adopted as a backbone network to extract the discriminative features first. Second, the spatial pyramid pooling (SPP) layer is adopted to capture the middle-level features from the features of ResNet. To learn the high-level features, the NetVLAD layer is used to aggregate the features representation from middle-level features. Context gating (CG) mechanism is adopted to further learn the high-level features for predicting the COVID-19 patients or not. Finally, extensive experiments are conducted on the collected database, showing the excellent performance of the proposed integrated architecture, with the sensitivity up to 97%, and specificity of 99.5% of the ResNet-18, and with the sensitivity up to 99%, and specificity of 99.4% of the ResNet-50.
