The data were acquired during the day over a 6-min period, using two vertical and two horizontal Ag-AgCl electrodes placed around the left eye. The sampling rate was 100 Hz. The vertical EOG signal , used for eye blink assessment, was obtained from a bipolar montage using the electrodes placed above and below the eye. The horizontal EOG signal, used to exclude artefacts produced by saccades, was obtained from a bipolar montage using the electrodes placed lateral to the external canthi. Participants were comfortably seated and instructed to fixate the wall in front of them while they believed the system was being calibrated and the experimenter was outside of the room. They were not instructed in any way about blinking and were asked not to move their head or activate their facial muscles to minimize EOG artefacts. The vEOG signals were rectified and band-pass filtered between 0.5 and 20 Hz. Eye blinks were detected using an automated procedure based on a voltage change threshold of 100 lV in a time interval of 400 ms. The vEOG signal was then visually inspected by two of the authors to assess detection accuracy and remove potential artefacts resulting from muscle activity and saccades as detected in the hEOG signal. As the inter-rater reliability was very high , the resulting sEBR measures from the two raters were averaged for subsequent analyses.All PET scans were acquired at the department of Nuclear Medicine of the Radboud University Medical Center using a Siemens mCT PET/CT camera . Participants were positioned as comfortably as possible,rolling grow trays in a supine position, with the head slightly fixated in a headrest to avoid movement. First, a low-dose CT was made for attenuation correction of the PET images, followed by an 89-min dynamic PET scan. The scan started at the same time as the bolus injection of the [18F]DOPA into an antecubital vein.
Images were reconstructed with an ordered subset expectation maximization algorithm with weighted attenuation and time-of-flight recovery, scatter corrected and smoothed with a 4-mm FWHM kernel.We realigned the [18F]DOPA images to the middle frame to correct for movement during scanning using SPM8 . The mean [18F]DOPA image and the realigned frames were coregistered to the structural MRI scan using SPM8. Higher [18F]DOPA uptake, indexed by higher Ki values, is established to reflect higher dopamine synthesis capacity . To create Ki images representing the amount of tracer accumulated in the brain relative to a cerebellar region of reference, we used an in-house graphical analysis program implementing Patlak plotting . Ki images were generated from PET frames corresponding to 24–89 min. These images are comparable to Ki images obtained using a blood input function but are scaled to the volume of tracer distribution in the reference region. Finally, structural MRI scans were normalized to a standard MNI template , and the transformation matrix was applied to coregistered Ki images. As expected from previous studies , the voxels with the highest Ki values were located in the striatum and the brainstem . For the main analyses, Ki values were extracted from an independently and functionally defined striatal region of interest, based on a whole-brain [18F]DOPA PET template normalized to MNI space Specifically, we retained the template voxels in which Ki values were at least three standard deviations above the template mean, thus approximating the anatomical boundaries of the striatum . For the exploratory voxel-wise analysis , we performed a linear regression using SPM8. The analysis was restricted to an anatomical mask covering the entire striatum , within which the results were corrected for multiple comparisons using a voxel-wise family-wise error -corrected threshold of P < 0.05.
With the bold promise to revolutionize clinical practice in psychiatry, the emerging field of Predictive Analytics in Mental Health has recently generated tremendous interest, paralleling similar developments in personalized and precision medicine. Here, we provide an overview of the key-questions and challenges in the field, aiming to 1) propose general guidelines for Predictive Analytics projects in psychiatry, 2) provide a conceptual introduction to core-aspects of predictive modelling technology, and 3) foster a broad and informed discussion involving all stakeholders including researchers, clinicians, patients, funding bodies, and policymakers.The immense economic loss mirrors the enormous suffering of patients and their friends and relatives.In addition, health care costs as well as the number of individuals diagnosed with psychiatric disorders are projected to disproportionately rise within the next twenty years.13 With an ever-growing number of patients, the future quality of health-care in psychiatry will crucially depend on the timely translation of research findings into more effective and efficient patient care. Despite the certainly impressive contributions of psychiatric research to our understanding of the aetiology and pathogenesis of mental disorders, the ways in which we diagnose and treat psychiatric patients have largely remained unchanged for decades.Recognizing this translational roadblock, we currently witness an explosion of interest in the emerging field of Predictive Analytics in Mental Health, paralleling similar developments in personalized or precision medicine.In contrast to the vast majority of investigations employing group-level statistics, Predictive Analytics aims to build models which allow for individual predictions, thereby moving from the description of patients and the investigation of statistical group differences or associations toward models capable of predicting current or future characteristics for individual patients , thus allowing for a direct assessment of a model’s clinical utility . In summary, valid models in this area would be instrumental, both, for minimizing patient suffering and for maximizing the efficient allocation of resources for research.
For example, children and adults are diagnosed with Attention-Deficit-Hyperactivity-Disorder every day and prescribed medications with little or no scientific evidence as to which patient will be likely to benefit from one or the other of the two major classes of medications or unlikely to benefit from either medication. In the same vein, the STAR*D study – a large evaluation of depression treatment including 4,041 outpatients – showed that approximately 50% of patients respond. In both cases, patients would greatly benefit from Predictive Analytics models predicting which treatment would be most effective . Against this background, Predictive Analytics in general and its potential applications in health have simultaneously been met with exuberant enthusiasm as well as with substantial skepticism: On the one hand, some see “previously unimaginable opportunities to apply machine learning to the care of individual patients”, prompting others to even propose “a shift from a search for elusive mechanisms to implementing studies that focus on predictions to help patients now”.On the other hand, critics have pointed out problems of an all too care-free view of Predictive Analytics in general and Big Data in particular.Considering the tremendous investment into Big Data infrastructure and Predictive Analytics capabilities in all areas of science and in the private sector16, most will agree, however, that this technology – to quote a recent New York Times article – “is here to stay”, but that we ought to see it as “an important resource for anyone analyzing data, not a silver bullet”. From this, the question arises: How can we best steer the development and implementation of Predictive Analytics technology to effect the clinical innovations demanded by researchers and practitioners alike? Now that evidence from initial proof-of-concept studies is accumulating in all areas – from genetics to neuroimaging, from blood-based markers to ambulatory assessments – and the approach is gaining momentum ,horticulture trays this question is particularly pressing. As the field of Predictive Analytics in Mental Health is faced with strategic choices which will have formative influence on research and clinical practice for the decades to come, we seek to move beyond the numerous descriptions and reviews of this beginning transformation of psychiatry by 1) proposing general guidelines for Predictive Analytics projects in psychiatry, 2) providing a conceptual introduction to the core-aspects of predictive modelling technology which distinguish Predictive Analytics in Mental Health from other areas of medicine or predictive analytics applications, and3) fostering a broad and informed discussion involving all stakeholders including researchers, clinicians, patients, funding bodies, and policymakers. To this end, we will, first, provide an overview of the steps of a Predictive Analytics project. Secondly, we will consider the challenges that arise from the unique, multivariate and multi-modal nature of mental disorders and argue that the combination of expert domain-knowledge and data integration technology is the key for overcoming, both, the conceptual and practical obstacles ahead. Finally, we will briefly discuss perspectives for the field.Every Predictive Analytics project can be described as a series of steps aimed at ensuring the utility of the resulting model. While this process is similar for all Predictive Analytics projects, numerous questions, problems and opportunities are unique to the area of mental health. The guiding-questions in Box 1 are intended primarily as a means to support explicit reflection of the essential steps of a project – from defining objectives to deploying the model. Thereby, we hope to foster a broad discussion leading to common standards and procedures in the field. Predictive Analytics efforts in psychiatry parallel developments in other fields of medicine. Generally, we have witnessed a trend towards ever more precise specification of the genetic, molecular and cellular aspects of disease. This so-called Precision Medicine approach , in many cases, led to the realization that disease entities which appear to be a single disorder actually have distinct genetic precursors and pathophysiology. For example, cancer diagnosis is – for many forms of cancer – defined by analysis of genetic variants based on which the optimal treatment can be predicted.
While communalities are particularly obvious with regard to technology, researchers in psychiatry are also faced with rather unique challenges. Apart from the massively multivariate and multi-modal nature of mental disorders which we will discuss in detail below, a traditionally much discussed issue arises from the often-times fuzzy and relatively unreliable labels of disease entities in psychiatry. As predictive models learn from examples, training a model aiming to support the differentiation between patients suffering from Major Depressive Disorder and individuals with Bipolar Disorder prior to conversion manic symptoms have become apparent), for instance, might proof difficult simply because it may be very hard to reliably categorize patients with certainty. In practice, most studies either mitigate this problem by employing resource-intensive, state-of-the-art diagnostic procedures in combination with multiple clinical expert ratings or circumvent it by acquiring data first and then waiting for the quantity of interest to become more easily accessible . Complementing these efforts to render labels more accurate, fuzzy and unreliable labels can also be handled directly using machine learning algorithms specifically designed for this purpose . While currently, it seems as if researchers in psychiatry almost exclusively rely on the optimization of data acquisition rather than trying to inherently model label uncertainty, combining the two approaches might be highly beneficial. In addition, current disease entities as defined by DSM-5 or ICD-10 are very heterogeneous regarding, both, physiology as well as clinical endophenotypes.On the one hand, this will make the classification of disease entities difficult as each entity is in fact a conglomerate of different physiological and behavioral deviations. On the other hand, the underlying causes or correlates of therapeutic response or disease trajectory may qualitatively as well as quantitatively vary for different, more homogeneous sub-samples of the data. While this makes training predictive models more difficult , machine learning algorithms are generally well equipped to handle such cases. In fact, learning multiple rules mapping features to labels are quite common . That said, homogeneous disease entities would not only make discovering rules easier , but definitely lead to more interpretable models which – though not technically the goal of predictive analytics – is still desirable from a scientific point of view. Most importantly, however, discovering homogeneous disease entities would enable us to move beyond merely reproducing the presently established diagnostic classification using considerably more expensive and complicated procedures. While this has thus far been a seemingly unattainable goal not only for DSM-5, the recent success of so-called unsupervised machine learning approaches might reinvigorate this line of research .Overwhelming evidence shows that no single measurement – be it a gene, a psychometric test or a protein – explains substantial variance with regard to any practically relevant aspect of a psychiatric disorder .