Medical Machine Learning Lab
Mental disorders are among the most debilitating diseases in industrialized nations today. 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 – particularly for affective disorders such as depression and bipolar disorder. 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 etiology and pathogenesis of mental disorders, the ways in which we diagnose and treat psychiatric patients have largely remained unchanged since the discovery of antidepressants and antipsychotics decades ago.
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 so-called personalized or precision medicine in general. In contrast to the vast majority of investigations employing group-level statistics, Predictive Analytics aims to build models which allow for individual (i.e. single-subject) predictions. More specifically, the former analyses seek to explain variance in the data (thereby enabling researchers to test hypotheses and gain further theoretical insights) while for the latter approach, a model based on one group of individuals (the so-called training samples) is constructed. This model is then used to predict characteristics of new, previously unseen individuals. In short, Predictive Analytics in Mental Health is moving from the description of patients (“hindsight”) and the investigation of statistical group differences or associations (“insight”) toward models capable of predicting current or future characteristics for individual patients (“foresight”), thereby allowing for a direct assessment of a model’s clinical utility (Figure 1). Within this framework, this group aims to build models in three broad areas of clinical application:
- Supporting differential diagnoses is crucial whenever the clinical picture alone is ambiguous (which especially in prodromal or first episode cases is the rule rather than the exception in mental disorders). Providing additional model-based information to clinicians thus enables a timely administration of disease-specific interventions. This increases adherence, minimizes undesired side-effects, and shortens the time until effective treatment (which may act neuroprotective and prevent a chronic disease course) is administered. The differentiation of patients suffering Major Depressive Disorder (MDD) from patients with Bipolar Disorder before the first manic episode is but one prominent example illustrating clinical utility in this area.
- The prediction of therapeutic response can support the selection of optimized interventions through comparative effectiveness research, thereby improving the trial-and-error-based approach common in psychiatry. This maximizes adherence and minimizes undesired side-effects. Importantly, it also allows clinicians to focus resources on patients who will most likely not benefit from the first-line treatment and allocate additional resources to those who will require second line or other treatments. Finally, identifying treatment-resistant individuals with high accuracy would also simplify the development and evaluation of novel drugs and interventions as research efforts could be custom-tailor kto homogeneous sub-populations.
- Models predicting individual risk are important in two respects: On the one hand, short-term prediction of risk can greatly improve outpatient management – for example with regard to conversion risk evaluation in prodromal states. On the other hand, long-term risk prediction would allow for a targeted application of preventive measures in early stages of a disorder or even before disease onset, which gains even more importance when current staging concepts in depression and bipolar disorder are taken into account. Equally important, individual risk prediction could greatly increase the efficiency of the development and evaluation of preventive interventions as research efforts could be focused specifically on at-risk individuals.
In summary, valid predictive models would be instrumental, both, for minimizing patient suffering and for maximizing the efficient allocation of resources. Realizing this potential is the core goal of this research group. To this end, we employ state-of-the-art tools from machine learning, artificial intelligence and statistical learning such as Deep Neural Networks, Random Forests etc.. In particular, we are developing a comprehensive Machine Learning Toolbox aiming to accelerate and simplify complex machine learning analyses using e.g. neuroimaging, genetic and psychometric data. Find the PHOTON here.
