Sarah Itani

At the Crossroads Between Psychiatry and Machine Learning: Insights Into Paradigms and Challenges for Clinical Applicability

While the prevalence of Autism Spectrum Disorder (ASD) is increasing, research continues in an effort to identify common etiological and pathophysiological bases. In this regard, modern machine learning and network science pave the way for a better understanding of the neuropathology and the development of diagnosis aid systems. The present work addresses the classification of neurotypical and ASD subjects by combining knowledge about both the structure and the functional activity of the brain. In particular, we model the brain structure as a graph, and the resting-state functional MRI (rs-fMRI) signals as values that live on the nodes of that graph. We then borrow tools from the emerging field of Graph Signal Processing (GSP) to build features related to the frequency content of these signals. In order to make these features highly discriminative, we apply an extension of the Fukunaga-Koontz transform. Finally, we use these new markers to train a decision tree, an interpretable classification scheme, which results in a final diagnosis aid model. Interestingly, the resulting decision tree outperforms state-of-the-art methods on the publicly available Autism Brain Imaging Data Exchange (ABIDE) collection. Moreover, the analysis of the predictive markers reveals the influence of the frontal and temporal lobes in the diagnosis of the disorder, which is in line with previous findings in the literature of neuroscience. Our results indicate that exploiting jointly structural and functional information of the brain can reveal important information about the complexity of the neuropathology.

One-class Classification (OCC) is an area of machine learning which addresses prediction based on unbalanced datasets. Basically, OCC algorithms achieve training by means of a single class sample, with potentially some additional counter-examples. The current OCC models give satisfaction in terms of performance, but there is an increasing need for the development of interpretable models. In the present work, we propose a one-class model which addresses concerns of both performance and interpretability. Our hybrid OCC method relies on density estimation as part of a tree-based learning algorithm, called One-Class decision Tree (OC-Tree). Within a greedy and recursive approach, our proposal rests on kernel density estimation to split a data subset on the basis of one or several intervals of interest. Thus, the OC-Tree encloses data within hyper-rectangles of interest which can be described by a set of rules. Against state-of-the-art methods such as Cluster Support Vector Data Description (ClusterSVDD), One-Class Support Vector Machine (OCSVM) and isolation Forest (iForest), the OC-Tree performs favorably on a range of benchmark datasets. Furthermore, we propose a real medical application for which the OC-Tree has demonstrated its effectiveness, through the ability to tackle interpretable diagnosis aid based on unbalanced datasets.

Attention Deficit/Hyperactivity Disorder (ADHD) is a neurodevelopmental disorder that has heavy consequences on a child’s wellbeing, especially in the academic, psychological and relational planes. The current evaluation of the disorder is supported by clinical assessment and written tests. A definitive diagnosis is usually made based on the DSM-V criteria. There is a lot of ongoing research on ADHD, in order to determine the neurophysiological basis of the disorder and to reach a more objective diagnosis. The advent of Machine Learning (ML) opens up promising prospects for the development of systems able to predict a diagnosis from phenotypic and neuroimaging data. This was the reason why the ADHD-200 contest was launched a few years ago. Based on the publicly available ADHD-200 collection, participants were challenged to predict ADHD with the best possible predictive accuracy. In the present work, we propose instead a ML methodology which primarily places importance on the explanatory power of a model. Such an approach is intended to achieve a fair trade-off between the needs of performance and interpretability expected from medical diagnosis aid systems. We applied our methodology on a data sample extracted from the ADHD-200 collection, through the development of decision trees which are valued for their readability. Our analysis indicates the relevance of the limbic system for the diagnosis of the disorder. Moreover, while providing explanations that make sense, the resulting decision tree performs favorably given the recent results reported in the literature.

Over the last years, the evolution of the Diagnostic and Statistical Manual of mental disorders (DSM) has been subject to a lively debate between people for and against the resulting changes in clinical practices. Autism Spectrum Disorder (ASD) figures among the syndromes that have seen the most significant changes between the fourth and fifth versions of DSM. The ASD category includes three distinct DSM-IV conditions: autistic disorder, Asperger’s and Pervasive Developmental Disorder Not Otherwise Specified (PDD-NOS). This revision is highly controversial, especially since it awakens fears of reduced access to healthcare. Indeed, the clinical criteria related to ASD are perceived as less stringent and may thus exclude a part of children previously diagnosed with Asperger’s or PDD-NOS. In this work, we shed light on the interest of mathematical approaches to investigate whether the changes involved by DSM-V as regards ASD are genuinely required. Moreover, we show how modern Machine Learning (ML) can be used to capture the complexity of the neuropathology through the detection of explanatory markers and the development of systems able to recommend a diagnosis. For such a purpose, we considered a data sample extracted from the publicly available ABIDE (Autism Brain Imaging Data Exchange) dataset, including neurotypical and ASD children aged between 6 and 12 years old (n = 177). The ASD children were also diagnosed as autistic, Asperger’s or PDD-NOS, based on DSM-IV criteria. Moreover, for each subject, the sample contains Blood Level Oxygen Dependent (BOLD) signals, at resting-state, given a brain parcellation in 90 brain regions. The results tend to show that the autism group is distinct from the Asperger’s and PDD-NOS conditions. It also appears that the assessment of the brain activity is relevant to make a diagnosis with interesting predictive performances. These results are promising and gives grounds of hope of completing the current formal assessment based on descriptive clinical criteria.