Search

Background: The clinical effects of smartphone-based interventions for bipolar disorder (BD) have yet to be established. Objectives: To examine the efficacy of smartphone-based interventions in BD and how the included studies reported user-engagement indicators. Methods: We conducted a systematic search on January 24, 2022, in PubMed, Scopus, Embase, APA PsycINFO, and Web of Science. We used random-effects meta-analysis to calculate the standardized difference (Hedges' g) in pre-post change scores between smartphone intervention and control conditions. The study was pre-registered with PROSPERO (CRD42021226668). Results: The literature search identified 6034 studies. Thirteen articles fulfilled the selection criteria. We included seven RCTs and performed meta-analyses comparing the pre-post change in depressive and (hypo)manic symptom severity, functioning, quality of life, and perceived stress between smartphone interventions and control conditions. There was significant heterogeneity among studies and no meta-analysis reached statistical significance. Results were also inconclusive regarding affective relapses and psychiatric readmissions. All studies reported positive user-engagement indicators. Conclusion: We did not find evidence to support that smartphone interventions may reduce the severity of depressive or manic symptoms in BD. The high heterogeneity of studies supports the need for expert consensus to establish ideally how studies should be designed and the use of more sensitive outcomes, such as affective relapses and psychiatric hospitalizations, as well as the quantification of mood instability. The ISBD Big Data Task Force provides preliminary recommendations to reduce the heterogeneity and achieve more valid evidence in the field.

Drug design is a critical step in the drug discovery process, where promising drug molecules are engineered to be later evaluated preclinically and perhaps clinically. Phenotypic drug design has again gained traction. Cancer cell lines, a frequently adopted in vitro model for phenotype drug design, can be used to evaluate the drug resistance level (lack of inhibitory activity, for example) of a large number of molecules, and discard those that are the least likely to become drug candidates. By reusing these datasets, supervised learning models have been built to predict drug resistance on cancer cell lines. Usually, these methods have assigned reliability to the whole model rather than reliability to individual predictions (molecules). In problems such as drug design, accurately achieving the latter would revolutionize decision making. Conformal prediction is a model-agnostic method to assign reliability to each model prediction. In this study, we investigated the impact of conformal prediction on the prediction of inhibitory activity of molecules on a given cancer cell line. This analysis was carried out in each of the 60 cell lines from the NCI-60 panel to understand the variability of the results across cancer types. We also discussed the implications of predicting the molecules considered most potent. In addition, we investigated how the further subdivision of the training set to build conformal prediction models may affect the results obtained. Overall, we observed that those molecules deemed most reliable by conformal prediction are substantially better predicted than those that are not. This suggest that such computational tools are promising to guide phenotypic drug design., The directories of this repository contain the following information: - CP_NCI60_data: Data needed, in .feather format, to recreate the main results of this paper. - CPModels: Saved models in .pkl format. - python_code: Scripts to generate the main results of this paper. See the README.MD for more information.

Non

Introduction: Through large neuroimaging consortia, researchers have identified a series of neuroanatomical alterations in mood and psychotics disorders, such as major depressive disorder (MDD), bipolar disorder (BD), and schizophrenia (SCZ). However, the mechanism behind these alterations is not well understood. One of the existing hypotheses suggests that the observed brain changes are related to a process of accelerated brain ageing. Studies investigating this hypothesis use a measure called the brain age gap (i.e., the difference between machine learning model predictions of brain age and chronological age). Thus far, there is limited understanding on how mood and psychotic disorders affect model predictions, how can predictions be clinically useful, and what is the biological meaning behind the brain age gap. In this thesis, we investigated accelerated brain ageing in mood and psychotic disorders. We sought to estimate the effect of the brain age gap and propose new ways of modeling brain age. We also explored the clinical utility and meaning of the brain age gap. Results: We confirmed the presence of a brain age gap in MDD, BD, and SCZ through a systematic review and meta-analysis. SCZ presented the highest levels of brain age gap, followed by BD and MDD. We analyzed the clinical utility of brain age for antidepressant treatment response and concluded that the brain age gap is not a predictor of antidepressant treatment response in weeks 8 and 16. We proposed a new method for brain age prediction that is more interpretable than previous approaches while preserving good predictive performance. We have also used model explanation strategies and identified that the brain age gap is largely associated with total gray matter volume reduction and ventricle enlargement in SCZ. Conclusion: The results of this thesis suggest that the brain age gap is present across mood and psychotic disorders. The results have also helped to clarify the meaning behind the brain age gap, a largely used but still poorly understood measure in neuroimaging research. So far, there is no indication that the brain age gap can be a useful tool for treatment response prediction in MDD. Thesis Doctor of Philosophy (PhD)

Resumo não disponível.

Objectives The International Society for Bipolar Disorders Big Data Task Force assembled leading researchers in the field of bipolar disorder (BD), machine learning, and big data with extensive experience to evaluate the rationale of machine learning and big data analytics strategies for BD. Method A task force was convened to examine and integrate findings from the scientific literature related to machine learning and big data based studies to clarify terminology and to describe challenges and potential applications in the field of BD. We also systematically searched PubMed, Embase, and Web of Science for articles published up to January 2019 that used machine learning in BD. Results The results suggested that big data analytics has the potential to provide risk calculators to aid in treatment decisions and predict clinical prognosis, including suicidality, for individual patients. This approach can advance diagnosis by enabling discovery of more relevant data-driven phenotypes, as well as by predicting transition to the disorder in high-risk unaffected subjects. We also discuss the most frequent challenges that big data analytics applications can face, such as heterogeneity, lack of external validation and replication of some studies, cost and non-stationary distribution of the data, and lack of appropriate funding. Conclusion Machine learning-based studies, including atheoretical data-driven big data approaches, provide an opportunity to more accurately detect those who are at risk, parse-relevant phenotypes as well as inform treatment selection and prognosis. However, several methodological challenges need to be addressed in order to translate research findings to clinical settings.

While deep learning models become more widespread, their ability to handle unseen data and generalize for any scenario is yet to be challenged. In medical imaging, there is a high heterogeneity of distributions among images based on the equipment that generates them and their parametrization. This heterogeneity triggers a common issue in machine learning called domain shift, which represents the difference between the training data distribution and the distribution of where a model is employed. A high domain shift tends to implicate in a poor generalization performance from the models. In this work, we evaluate the extent of domain shift on four of the largest datasets of chest radiographs. We show how training and testing with different datasets (e.g., training in ChestX-ray14 and testing in CheXpert) drastically affects model performance, posing a big question over the reliability of deep learning models trained on public datasets. We also show that models trained on CheXpert and MIMIC-CXR generalize better to other datasets., 10 pages, 3 figures

Machine learning applications make several assumptions regarding the scenario where they are employed. One common assumption is that data distribution in the test environment follows the same distribution of the training set. This assumption is systematically broken in most real-world scenarios; the difference between these distributions is commonly known as domain shift. Unsupervised domain adaptation aims at suppressing this problem by leveraging knowledge with unlabeled data from the test environment. One of the most sensitive fields for domain shift is medical imaging. Due to the heterogeneity in data distributions from scanners, models tend to vary in predictive performance when dealing with images from scanners with no examples in the training set. We propose two contributions in this work. First, we introduce the use of self-ensembling domain adaptation in the field of medical imaging segmentation in a spinal cord grey matter segmentation task. Next, based on the success of self-ensembling, we adapt two other recent work from the semi-supervised learning literature to the same task, namely, unsupervised data augmentation and MixMatch. We conduct ablation studies and other experiments in order to understand the behavior of each method and compare their best results. The results show improvements over training models in a supervised learning fashion and demonstrate that recent semi-supervised learning methods are promising for domain adaptation in medical imaging segmentation. Aplica??es com aprendizado de m?quina possuem diversas suposi??es sobre o cen?rio em que s?o colocadas. Uma suposi??o comum ? a de que o ambiente de teste segue a mesma distribui??o dos dados de treino. Essa suposi??o ? sistematicamente quebrada em c?narios do mundo real; a diferen?a entre essas distribui??es ? conhecida como domain shift. Adapta??o de dom?nio n?o-supervisionada visa mitigar esse problema impulsionando o conhecimento dos modelos com dados do ambiente de teste. Uma das ?reas mais sens?veis a domain shift ? a de imagens m?dicas. Devido a heterogeneidade das distribui??es de dados das m?quinas de aquisi??o de imagens, os modelos tendem a variar sua performance preditiva quando lidam com imagens vindas de m?quinas sem nenhum exemplo no conjunto de treino. Este trabalho prop?e duas contribui??es. Primeiramente, o uso de self ensembling em adapta??o de dom?nio para segmenta??o de imagens m?dicas para segmenta??o de subst?ncia cinzenta na medula espinhal ? introduzido. Em seguida, baseado no sucesso do self-ensembling, outros trabalhos recentes da literatura de aprendizado semi-supervisionado s?o adaptados para o contexto apresentado, nominalmente, unsupervised data augmentation e MixMatch. Foram conduzidos estudos de abla??o e experimentos para compreens?o do comportamento dos m?todos e compara??o dos seus melhores resultados. Os resultados indicam uma melhoria em rela??o a treinamento puramente supervisionado, al?m de demonstrar que os m?todos recentes de aprendizado semi-supervisionado s?o promissores para o campo de adapta??o de dom?nio em segmenta??o de imagens m?dicas.